Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Housework in Canada : uneven convergence of the gender gap in domestic tasks, 1986-2010 Kolpashnikova, Kamila 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2017_february_kolpashnikova_kamila.pdf [ 3.63MB ]
Metadata
JSON: 24-1.0340502.json
JSON-LD: 24-1.0340502-ld.json
RDF/XML (Pretty): 24-1.0340502-rdf.xml
RDF/JSON: 24-1.0340502-rdf.json
Turtle: 24-1.0340502-turtle.txt
N-Triples: 24-1.0340502-rdf-ntriples.txt
Original Record: 24-1.0340502-source.json
Full Text
24-1.0340502-fulltext.txt
Citation
24-1.0340502.ris

Full Text

HOUSEWORK IN CANADA: UNEVEN CONVERGENCE OF THE GENDER GAP IN DOMESTIC TASKS, 1986-2010 by  Kamila Kolpashnikova  MA, The University of Tokyo, 2009  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Sociology)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   December 2016  © Kamila Kolpashnikova, 2016 ii  Abstract  Housework is one of the last bastions of gender inequality. The persistence of the cultural association of housework with “women's work” and its significance in reflecting societal power differentials between women and men makes research on the division of household labour important. My work explores the division of domestic work in Canada, paying special attention to changes over time and to the economic and cultural explanations of women’s and men’s differential participation in routine and non-routine domestic tasks. First, I decompose the gender gap in time allotted to housework tasks using five time use cycles of the Canadian General Social Survey. Then, using OLS regression with the Heckman correction, I investigate whether and to what extent economic or cultural factors play a role in the division of individual domestic tasks. The gender gap analysis shows that tasks, like shopping, which is culturally understood as a more gender-neutral activity, are best explained by the time availability framework, whereas the economic factors, in general, can explain a sizeable share of the participation in tasks traditionally associated with women such as household cleaning. For instance, the latter account for around 39% of the gender gap in time spent on cleaning among all married and cohabiting Canadians. However, the economic and gender-centred factors are least likely to explain the gender gap in tasks where there is a clear cultural change in attitudes and participation. For example, they can explain only 31% of the gender gap in cooking. Additionally, the findings suggest that pressures for breadwinning Canadian women to compensate for gender deviance in paid work are more severe than those faced by men. Thus breadwinning women continue to reproduce traditional gender patterns in cooking, cleaning, and shopping tasks. On the other hand, Canadian men perform a new behavioural pattern in cooking tasks: breadwinning men iii  break traditional gender patterns and spend more time on cooking than can be predicted by economic exchange theory. In total these patterns reveal the processes through which cultural changes around a domestic task propel the changes toward gender equality in the division of housework. iv  Preface  This dissertation is an original intellectual product of the author, Kamila Kolpashnikova. The analyses of the Canadian General Social Survey data were conducted with the granted access at the University of British Columbia Research Data Centre and all the results released with the permission of Statistics Canada. This research was supported by funds to the Canadian Research Data Centre Network (CRDCN) from the Social Sciences and Humanities Research Council (SSHRC), the Canadian Institute for Health Research (CIHR), the Canadian Foundation for Innovation (CFI), and Statistics Canada. Although the research and analysis are based on data from Statistics Canada, the opinions expressed do not represent the views of Statistics Canada. Research Supervisors Dr. Neil Guppy, Dr. James White, and Dr. David Tindall contributed to the research design, development of the theoretical framework, and manuscript edits. Data from the American Time Use Survey Public-Use Microdata files were used for comparison.     v  Table of Contents  Abstract ................................................................................................................................................ ii Preface ................................................................................................................................................. iv Table of Contents ................................................................................................................................ v List of Tables ...................................................................................................................................... ix List of Figures .................................................................................................................................... xii List of Abbreviations ........................................................................................................................ xv Glossary ............................................................................................................................................. xvi Acknowledgements .........................................................................................................................xvii Dedication ...................................................................................................................................... xviii Chapter 1: Introduction..................................................................................................................... 1 1.1 Overview of the Research Project ...................................................................................... 5 1.2 Research Agenda ................................................................................................................. 7 1.3 Research Questions ........................................................................................................... 13 Chapter 2: Building a Theory of Housework ............................................................................... 14 2.1 Perspectives within the Economic Approach .................................................................. 17 2.1.1 Economic Approach to the Gendered Division of Housework.................................. 17 2.1.2 Time Availability Perspective ...................................................................................... 21 2.1.3 Autonomy or Absolute Resources Perspective ........................................................... 23 2.1.4 Relative Resources Perspective ................................................................................... 25 2.2 Gender-centred Perspectives............................................................................................. 28 2.2.1 Gender Perspective ....................................................................................................... 29 vi  2.2.2 Focusing on Resistance and Undoing Gender in Housework .................................... 35 2.3 Unifying Framework ......................................................................................................... 37 2.3.1 Brines and Greenstein’s Testable Specifications and Their Extensions ................... 38 2.3.2 Limitations of the Specifications ................................................................................. 40 2.3.3 Gender Neutralization – The Ignored Dimensions ..................................................... 41 2.3.4 Housework Task Hypotheses ....................................................................................... 44 Chapter 3: Data and Methods ........................................................................................................ 48 3.1 Datasets and Samples ........................................................................................................ 48 3.1.1 General Social Survey .................................................................................................. 49 3.1.2 American Time Use Survey ......................................................................................... 50 3.1.3 Sample Selection ........................................................................................................... 51 3.1.4 Specifics of Time Use Diaries ..................................................................................... 52 3.1.5 Collection and Limitations of Time Use Diaries ........................................................ 53 3.1.6 Heckman Two-Step Correction for Selection Bias..................................................... 53 3.1.7 Comparing Canadian and American Samples ............................................................ 55 3.2 Measures ............................................................................................................................ 62 3.2.1 Dependent Variables ..................................................................................................... 62 3.2.2 Independent Variables .................................................................................................. 64 3.2.3 Control and Selection Variables .................................................................................. 67 3.3 Models ................................................................................................................................ 69 3.3.1 Comparing Gaps: Decomposition Method .................................................................. 71 3.3.2 Ordinary Least Squares Models ................................................................................... 76 Chapter 4: Decomposing the Gender Gap.................................................................................... 77 vii  4.1 Immigration to Canada: Historic Perspective .................................................................. 77 4.2 Results: Time Spent on Housework Tasks Trends .......................................................... 79 4.3 Gender Norms and Same-sex couples.............................................................................. 85 4.4 Results: Analyzing the Gender Gap in Housework......................................................... 87 4.4.1.1 Indoor Routine Tasks: Cooking ............................................................................ 92 4.4.1.2 Indoor Routine Tasks: Cleaning ......................................................................... 103 4.4.1.3 Outdoor Routine Tasks: Shopping ..................................................................... 113 4.4.1.4 Non-Routine Tasks: Maintenance ...................................................................... 122 4.4.1.5 Gender Gap Decomposition Discussion ............................................................ 130 Chapter 5: Gendered Division of Labour: Relative Resources or Gender-centred Approach? ........................................................................................................................................ 134 5.1 Routine Indoor Housework and Relative Resources..................................................... 134 5.1.1 Results Testing Specifications on Aggregate Housework ....................................... 135 5.1.2 Results for Individual Housework Tasks .................................................................. 140 5.1.2.1 Cooking ................................................................................................................ 140 5.1.2.2 Cleaning ............................................................................................................... 147 5.1.2.3 Shopping and Maintenance ................................................................................. 153 5.2 Discussions and Chapter Conclusions ........................................................................... 159 Chapter 6: Conclusions .................................................................................................................. 170 6.1 Addressing the Hypotheses ............................................................................................. 172 6.2 Strengths and Limitations, Implications, and Future Research .................................... 179 Bibliography .................................................................................................................................... 185 Appendices ....................................................................................................................................... 205 viii  Appendix A Variables Coding Supplement ............................................................................... 205 Appendix B Summary of the Decomposition of the Gender Gap in the US ........................... 210 Appendix C Supplement to the Theoretical Framework of Brines (1994) on the Domain of Relative Economic Contribution .................................................................................... 213 Appendix D OLS Regression Results for the US Sample......................................................... 215 Appendix E Trends for Cohabiting Canadians and Bivariate Correlation Tables for Main Continuous Dependent and Independent Variables ...................................................... 223  ix  List of Tables  Table 1 Summary of the Hypotheses by Research Question ........................................................... 46 Table 2 Descriptive Statistics for the Main Variables for the Canadian GSS ................................ 56 Table 3 Descriptive Statistics for the ATUS, 2003-2015................................................................. 56 Table 4 Independent and Control Variables, Canadian GSS ........................................................... 60 Table 5 Independent and Control Variables, ATUS......................................................................... 61 Table 6 Descriptive Statistics of Dependent Variables by Gender Between Same-Sex couples, 1998-2010, and All Couples, 1998-2010 .......................................................................................... 86 Table 7 Explained and Unexplained Gender Gap in Domestic Tasks by Pooled Decomposition, in % ...................................................................................................................................................... 89 Table 8 Percent Explained in Gender Gap in Time Spent on Cooking among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ........................................................................ 97 Table 9 Percent Explained in Gender Gap in Time Spent on Cooking among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ........................................................................ 98 Table 10 Percent Explained in Gender Gap in Time Spent on Cooking among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition..................................................... 99 Table 11 Percent Explained in Gender Gap in Cooking among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 100 Table 12 Percent Explained in Gender Gap in Time Spent on Cooking among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 101 Table 13 Explained and Unexplained Gender Gap among Those Who Report Any Housework Time by Pooled Decomposition, in %............................................................................................. 104 x  Table 14 Percent Explained in Gender Gap in Time Spent on Cleaning among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ...................................................................... 107 Table 15 Percent Explained in Gender Gap in Time Spent on Cleaning among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ...................................................................... 108 Table 16 Percent Explained in Gender Gap in Time Spent on Cleaning among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 109 Table 17 Percent Explained in Gender Gap in Cleaning among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 110 Table 18 Percent Explained in Gender Gap in Time Spent on Cleaning among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 111 Table 19 Percent Explained in Gender Gap in Time Spent on Shopping among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ...................................................................... 117 Table 20 Percent Explained in Gender Gap in Time Spent on Shopping among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 118 Table 21 Percent Explained in Gender Gap in Time Spent on Shopping among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 119 Table 22 Percent Explained in Gender Gap in Shopping among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 120 Table 23 Percent Explained in Gender Gap in Time Spent on Shopping among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 121 Table 24 Percent Explained in Gender Gap in Time Spent on Maintenance among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition................................................... 125 xi  Table 25 Percent Explained in Gender Gap in Maintenance among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 126 Table 26 Percent Explained in Gender Gap in Maintenance among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 127 Table 27 Percent Explained in Gender Gap in Maintenance among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ...................................................................... 128 Table 28 Percent Explained in Gender Gap in Maintenance among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition ................................................................................ 129 Table 29 Models on Logged Time Spent on Housework for Canadian Women and Men, 1986-2010 ................................................................................................................................................... 137 Table 30 Year Fixed Effects Models on Log of Time Spent on Housework for American Women and Men, 2003-2015 ......................................................................................................................... 146 Table 31 Year and Province Fixed-Effects Models on Log of Time Spent Cooking for Women and Men with Heckman Adjustment, 1986-2010 ........................................................................... 162 Table 32 Year and Province Fixed-Effects Models on Log of Time Spent on Cleaning for Women and Men with Heckman Adjustment, 1986-2010............................................................. 164 Table 33 Year and Province Fixed-Effects Models on Log of Time Spent on Shopping for Women and Men with Heckman Adjustment, 1986-2010............................................................. 166 Table 34 Year and Province Fixed-Effects Models on Log of Time Spent on Maintenance for Women and Men with Heckman Adjustment, 1986-2010............................................................. 168 Table 35 Summary of the Results Associated with Hypotheses ................................................... 172  xii  List of Figures Figure 1 Gender-centred Model ......................................................................................................... 33 Figure 2 Bargaining Linear and Cumulative Disadvantage Models ............................................... 39 Figure 3 Constant Association between Resources and Housework............................................... 41 Figure 4 Gender Neutralization Specification: Linear and Cumulative Variants........................... 44 Figure 5 Time Spent on Cooking, Canada and US........................................................................... 57 Figure 6 Time Spent on Cleaning, Canada and US .......................................................................... 58 Figure 7 Time Spent on Shopping, Canada and US ......................................................................... 58 Figure 8 Time Spent on Maintenance, Canada and US ................................................................... 59 Figure 9 Time Spent on Domestic Tasks, Canada and US .............................................................. 59 Figure 10 Decomposition Method ..................................................................................................... 74 Figure 11 Time Spent on Housework by Canadians, 1986-2010 .................................................... 80 Figure 12 Time Spent on All Housework by Americans, 2003-2015 ............................................. 80 Figure 13 Time Spent on Cooking Trends Among English and French Canadians, 1986-2010 .. 83 Figure 14 Time Spent on Domestic Tasks Among Chinese (Left), South Asian (Centre), and Filipino (Right) Canadians, 1986-2010 ............................................................................................. 84 Figure 15 Percent Explained of the Gender Gap in Cooking by Different Frameworks ............... 95 Figure 16 Percent Explained of the Gender Gap in Cleaning by Different Frameworks ............ 105 Figure 17 Percent Explained of the Gender Gap in Shopping by Different Frameworks ........... 115 Figure 18 Percent Explained of the Gender Gap in Maintenance by Different Frameworks ...... 124 Figure 19 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cooking (Y), Canadian Women....................................................................................... 142 xiii  Figure 20 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cooking (Y), Canadian Men ............................................................................................ 142 Figure 21 Predicted Association between Relative Economic Contribution and Logged Time Spent on Cooking, American Women (Upper) and American Men (Bottom) ............................. 144 Figure 22 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), Canadian Women ...................................................................................... 148 Figure 23 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), Canadian Men ........................................................................................... 149 Figure 24 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), American Women ..................................................................................... 151 Figure 25 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), American Men ........................................................................................... 152 Figure 26 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), Canadian Women ..................................................................................... 156 Figure 27 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), Canadian Men........................................................................................... 156 Figure 28 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Maintenance (Y), Canadian Men ..................................................................................... 158 Figure 29 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), American Women .................................................................................... 158 Figure 30 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), American Men .......................................................................................... 159 xiv  Figure 31 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Maintenance (Y), American Men .................................................................................... 159  xv  List of Abbreviations ATUS – American Time Use Survey ATUS-X - American Time Use Survey Data Extract Builder CPS - Current Population Survey (US) GSS – Canadian General Social Survey (Canada) MTUS – Multinational Time Use Study PUMF - Public Use Microdata File RDC – Research Data Centre  xvi  Glossary Heckman correction – a method that allows corrections for the selection bias in sampling. This correction is applied to regression coefficients. Pooled Blinder-Oaxaca Decomposition – one of the variations of decomposition modelling, based on the regressions for two groups and comparison of their slopes. The method allows the analysis of the extent to which the differentials in the outcome variable are associated with differentials in prediction factors. The pooled variation of the Blinder-Oaxaca decomposition technique involves pooled methods of assigning weights to resulting coefficients in order to identify the theoretical non-discriminatory coefficient vector. The decomposition differentiates between the explained part of the gap and the unexplained part of the gap. The explained part represents the portion of the gap that can be accounted for by the factors of the tested model. The unexplained portion is usually ascribed to discrimination or other unexplained differences. Time Use Diaries – self-reported estimates of how time is spent on diary days. Time Use Surveys are usually collected in the form of time use diaries. In the case of Canada and the US, respondents report on their activities over a 24-hour period starting from 4 am on the diary day to 4 am of the next day.  xvii  Acknowledgements  I incurred numerous debts in the preparation of this dissertation. Let me begin by thanking friends and colleagues at the University of British Columbia and especially those who are members of St. John’s College: Drea Glen, James Gauthier, Markus Schultz-Weiling, Dr. Blake Allen, Zoe Lam, Nathaniel Lim, Evan Koike, Edgar Liao, Cary Wu, Dr. Tatsuya Suzuki, Dr. Atsushi Kanazawa, Dr. Kiwa Nakano for creating a supporting and stimulating atmosphere for me to be able to finish my thesis. Special thanks are owed to my mother and my little sister, who have morally supported me in my long years of education. I offer my enduring gratitude to the faculty, staff and my fellow sociology students at UBC, who have inspired me to continue my work in this field. I owe particular thanks to Dr. Neil Guppy, whose guidance and support made this dissertation possible. I am also grateful to Dr. James White, whose insights taught me to question my understanding more deeply. I thank the government of Japan for supporting me financially in my endeavors and for preparing me for the educational challenges abroad. I am thankful to Tokyo University and its faculty and students for the moral support that I have received from them throughout my career. I am especially grateful to my professors and colleagues at the Centre for Time Use Research at Oxford University: Dr. Jonathan Gershuny, Dr. Oriel Sullivan, Dr. Man-Yee Kan, Dr. Kimberly Fisher, Dr. Jooyeoun Suh, Dr. Ewa Jarosz-Gugushvili, and Giacomo Vagni. I also wish to express thanks to the Maryland Population Center for supporting my trip to the 2016 ATUS-X Time Use Data Access System Workshop. Finally, I owe a continuing debt of gratitude to Patricia Houle of Statistics Canada for her feedback and clarifications on time use survey methods and sampling procedures. xviii  Dedication       To Tolyana and Stas, friends that influenced my life. 1  Chapter 1: Introduction  Do we live in an egalitarian society where women and men share housework equally? There are two main approaches to answering the question in the pertinent literature. The argument in recent decades has been dominated by the view that the gender revolution has stalled (Hochschild and Machung 1989) and that we have not reached gender equality. However, even though full equality in terms of sharing housework has not yet been attained, to some extent the glass is half-full, as the other, more optimistic, side argues (Gimenez-Nadal and Sevilla 2012; Kan, Sullivan, and Gershuny 2011; Marshall 2011; Robinson and Godbey 1997).  Women today are no longer held to the standards of the traditional highly-segregated gendered division of household labour. For instance, new generations are usually socialized in ways that promote more egalitarian gender roles compared to older generations (Alwin and McCammon 2004). Thus, people of the baby busters generation hold more egalitarian beliefs in terms of gender norms than do baby boomers (Bianchi, Robinson, and Milkie 2006). Moreover, today men are increasingly more involved in the domestic chores. Thus, Marshall (2011) finds that Generation Y men spend, on average, 17 minutes more on the unpaid work around the home than men from the late baby boomers generation. Similarly, an array of recent studies demonstrate that women devote less time to housework and that men gradually take on more responsibility both in Canada (Marshall 1986, 2011) and in other countries, including the European Union (Gimenez-Nadal and Sevilla 2012; Kan et al. 2011) and the United States (Coltrane 2000). The stalled revolution view supporters cite a number of hurdles met on the way to reaching gender equality in unpaid work, both structural and individual in character. Thus, Hochschild and 2  Machung (1989) confirm that even in the scenario where women are the breadwinners, men still do less housework as an act of culturally-driven resistance and to compensate for the loss of their self-esteem due to their perception of being in an inadequate employment situation. Yet because the stalled revolution view has circumscribed the scholarship to the reasons as to why equality has not been reached, it has been relatively blind to the causes of the gender convergence in the division of domestic labour, if it were in fact occurring which the gender convergence side argues. In this regard, Sullivan (2006) notes that there was a bias in the earlier studies toward seeing the glass as half-empty: “[t]he majority of the literature in the area has focused…on why gender inequalities in the home are so persistent.” Consequently, more recent studies in the field, including the scholarship of Sullivan herself, try to fill in the lacunae in explaining the causes behind the converging housework time of men and women, in the field of unpaid labour research (Gershuny 2000; Heisig 2011; Hook 2006, 2010, Marshall 2009, 2011).  There are two main ways to approach the explanation of the gender convergence of household labour: using the tenets of the rational choice theory (more economic) or the gender-centred perspective. Within the economic camp, there are three major arguments that are worth mentioning. First, the relative resources argument is based on the premise that because households allocate their time so as to maximize the utility of that time for the household (Becker 1981), one of the partners would inevitably specialize more on the labour market activities and thus accumulate more economic resources than the other. Consequently, the partner that has more ‘tangible’ and ‘intangible’ resources (Scanzoni 1980) will have more power to bargain her- or himself out of housework, exchanging resources for housework (England and Farkas 1986; Scanzoni 1978). Gupta (2007) criticizes the stance that relative resources and power drive the gendered division of housework specifically through bargaining 3  mechanisms and insists that absolute, not relative, resources affect the disproportionate allocation of time to domestic chores, emphasizing the autonomous decision-making within households 1  or ‘silent agreements’ (Scanzoni 1980). Thus the second approach within the economic perspectives is dubbed the autonomy perspective. The autonomy perspective also implicitly focuses on the relevance of class differences to the division of unpaid labour. Third, another argument is that in a household, people bargain with time rather than with other types of resources. A partner who has more time is expected to do more unpaid work (Blood and Wolfe 1960; Coverman 1985; Hook 2004). Because this argument is based on the availability of time, it is called the time availability or time constraint perspective. There are numerous extensions of the resource-based economic perspective, including the theory of resources in cultural context attributing the marital power to cultural expectations about its distribution (Rodman 1967) or the exchange theory positing that the partner with more alternatives outside of the marriage will have the greater power (Heer 1963). I focus on those three resource-based explanations which continue to be most prominent in the current empirical research literature and which I can adequately test with the best available evidence at hand. On the other hand, the gender-centred perspective supports the idea that individuals learn their roles through gendered socialization and re-enact what they perceive a woman and a man should normatively do in a heavily gendered arena like the household. Gender socialization and its performance affect the gendered division of household labour. The studies of this approach, stressing a more cultural framing, claim that a part of the gender gap narrowing in time devoted to housework is due to new gender socialization patterns of younger generations (Gershuny                                                    1 Husbands’ and wives’ domination may also be reflected in each having autonomous domains of decision-making, or implicit bargaining (Scanzoni and Polonko 1980). 4  2000) resulting in new generational attitudes and behaviours, which “are thought to be created in the formative years and often stabilize in adulthood” (Marshall 2011). This perspective evolved with the emphasis on the performative side of gender into the ‘gender display’ and the ‘doing gender’ argument, which views a household as a “site for doing gender” (West and Zimmerman 1987) or the ‘gender factory’ (Berk 1985). The ‘doing gender’ perspective is the predominant approach in the quantitative analysis of the gendered division of housework. In this cultural, or gender-centred, framing of social action, ‘doing gender’ emphasizes performances, agency, and expectations that reproduce traditional patterns of action, whereas ‘undoing gender’ highlights new, unconventional ways of acting outside of traditional gender scripts. Additionally, there are a few smaller scale perspectives sometimes used in studying the gendered division of labour. For instance, there is an argument that new labour saving technology has eased women’s time commitment2 and increased opportunities for participation of women in the labour market (Heisig 2011). The perspective, however, adds very little as to the understanding of the decision-making process because it is related to the idea of absolute resources and gives limited additional insight into negotiating mechanisms between women and men regarding the division of household labour, and thus, did not receive much traction within the field. Other theoretical approaches called for the use of the life-course perspective on the changes in the ideology and its association with the allocation of time in households (Coltrane and Ishii-Kuntz 1992), of the functionalist approach to specialization in housework tasks (Parsons 1942), and of the class conflict perspective (Ehrenreich 2001; Nakhaie 2002). While these are useful theoretical perspectives, I limit the scope to the theoretical frameworks germane                                                    2 It might be worth noting that there are studies in the US that find that the technological revolution actually leads to more domestic work, which some attribute to raised standards of cleanliness and housework in general (Vanek 1978). But because these are studies conducted quite a while ago, it could be meaningful to retest the conclusions they present in the future research. 5  to the empirical research of the division of housework. As to the social class perspective, although useful for the analysis of variance within gender groups, it is not applicable to the analysis of the differences contrasting one gender group against the other, i.e. for the analysis of the gender gap. In households, partners are usually assumed to be of the same social class when class is understood as a certain lifestyle, tastes, and developed cultural preferences (Bourdieu 1980). The economic part of social class, as it applies to comparing women to men, can be captured by personal income as Gupta (2007) argues and by employment status, whereas gendered leisure differences (Berk 1985; Coverman and Sheley 1986; Hochschild and Machung 1989; Sayer 2005) are reflected in time spent on leisure activities. 1.1 Overview of the Research Project An important shortcoming of the literature on the division of household labour is its inadequate explanation of patterns of participation in housework that contradict the expectations of the economic theory or predominant gender roles. Thus the current theoretical development and its testable specifications do not account for patterns that are gender non-normative such as behaviour consistent with ideas of ‘undoing’ gender (Butler 2004; Deutsch 2007). To address this limitation, Chapter 2 explicates and extends the current theoretical standing on the analysis of the gendered division of housework. In the Chapter, the specifications testing individual theoretical ideas are taken one step further to accommodate the possibility of non-normative gender behaviour. This Chapter frames the theoretical arguments that are tested in subsequent chapters. The data on which the analyses are based are described in Chapter 3. The datasets come from the five cycles of the Canadian General Social Survey (GSS) available at the Research Data Centre of the University of British Columbia, with access granted by Statistics Canada. The 6  target population for this survey were Canadian residents aged 15 or older with the exception of the residents of the Yukon, Northwest Territories and Nunavut and full-time residents of institutions. The GSS uses a stratified sampling design with provinces as the first strata, census metropolitan areas (CMA) as the second strata and separate strata for big metropolitan areas such as Toronto and Montreal. The non-CMA areas were grouped together in each province to create 10 additional strata. For the comparison with the Canadian sample, I also utilize the American Time Use Survey data for the period between 2003 and 2015 to investigate whether the trends evident from the Canadian sample are also present in the American subset. Additionally, Chapter 3 describes methods and analytical strategies utilized for the present project in detail. The sample selection and choices made to address the sampling bias are also discussed here. The literature of the gendered division of labour mostly focuses on the analysis of the effects that resources have on participation in housework among women or among men. To this day there is no comprehensive study that systematically addresses the analysis of the gender gap in individual tasks. To fill in this lacunae, I start my analysis with the Blinder Oaxaca decomposition of the gender gap in housework tasks in Chapter 4. The decomposition method is widely used in the analysis of the gender gap in wages. The Chapter analyzes the trends in the housework tasks and the factors that contribute to the explanation of gender differences. The analyses in the Chapter are extended to several major cultural groups in Canada, namely Anglo-, French, Chinese, South Asian, and Filipino Canadians. The investigation of the gender gap in housework by ethnic or cultural origin is novel in Canada and can only be performed on all datasets using the access to the Research Data Centres Network. Following the traditional analyses in the field of the division of unpaid labour, Chapter 5 tests to what extent the differences in resources can explain the participation among women and men  7  in domestic chores. The analyses in this Chapter use the extended theoretical specifications which allow conclusions regarding the change in housework that happens with the gender group which faces the least pressure in term of conforming to societal gender roles – men are ‘undoing’ gender in routine housework, whereas women are still pressured to ‘do’ gender. The project concludes with recapturing the findings concerning the main research hypotheses and integrating them into the bigger picture of the research agenda in the gendered division of household labour.  1.2 Research Agenda I investigate whether the two approaches, those lying within the economic theory and those adhering to a more gender-centred cultural approach, are able to explain the discrepancies in the division of domestic labour in Canada. I do not, however, expect to arrive at a monistic answer or to find the approaches to be mutually exclusive. Instead, I aim at providing a pluralistic explanation and describe a combination of the factors enhancing or stalling gender convergence in housework.  First, I want to advance scholarship on the topic of the gendered division of housework by investigating whether gender convergence happened evenly for all types of domestic tasks. Housework labour represents a set of domestic activities “required for the maintenance of household” and is usually unwaged (Shelton 1992).3 Some previous studies in the field claim that some household tasks are ‘gender-typed’ (Coltrane 2000; Coltrane, Parke, and Adams 2004; Craig and Baxter 2016; Lee and Waite 2005), reflecting the idea that behaviour in a household is driven by what is perceived as gender normative and that housework by itself is a ‘symbolic                                                    3 The focus of the present paper is on housework and not on child care which is not considered a type of housework but more commonly analyzed as a distinct care activity (Lee and Waite 2005; Sayer 2010; Shelton 1992). This is important to distinguish housework from child care because to analyze care activities, we would need additional theoretical frameworks aimed at explaining child and adult care. 8  enactment of gender relations’ (Evertsson and Nermo 2004), while households are ‘sites for doing gender’ (West and Zimmerman 1987) and ‘gender factories’ (Berk 1985). Scholars use, though, different notions to describe the phenomenon (Coltrane 2000) and refer to routine housework as either ‘female’ (Presser 1994), ‘female-dominated’ (Blair and Lichter 1991; Coltrane and Ishii-Kuntz 1992), 4  ‘female-stereotypic’ (Cunningham 2005, 2007; Dotti Sani 2016; Goldberg 2013), ‘female gender-typed’ (Starrels 1994), ‘traditionally feminine’ (Sutphin 2010; Szabo 2014), or ‘feminine’ (Antill et al. 1996; Kan et al. 2011).5 In general, what this classification suggests is that routine housework such as cooking, cleaning, and shopping has traditionally been associated with women and thus considered more ‘feminine’ behaviour.6  On the other hand, Coltrane (2000) also notes that non-routine tasks such as “household repairs, mowing the lawn, taking care of cars,” or snow shoveling7 are usually called ‘male,’ ‘male-dominated,’ ‘male-typed,’ or ‘masculine’ (Blair and Lichter 1991; Shelton 1992), emphasizing that non-routine housework has been traditionally associated with men rather than women. Moreover, there are a few tasks that are regarded as ‘gender-neutral’ which are associated neither predominantly with women nor with men. Coltrane (2000) mentions bill paying and driving as gender-neutral but the housework activity of this category that takes the most time is shopping (Baxter 2002; Craig, Powell, and Brown 2015). This gendered categorization reminds us once again that housework is a heavily gendered process, “accompanied by significant task segregation” (Shelton 1992), and that domestic tasks vary in the degree of their association with a certain gender group (Coltrane 2000). This approach to                                                    4 The terms ‘female-dominated’ or ‘male-dominated’ are appropriations from the literature on gendered occupations (Chang 2000; Reskin and Hartmann 1986; Smith 2000). 5 The summary is adopted from Coltrane (2000). 6 The ‘feminine’ character of activities (performances) is socially constructed through accounts of supposed femininity. 7 In the context of Canada. 9  housework tasks as heavily-gendered is consistent both with ‘socialization gender-roles attitudes’ (Coltrane 2000; Hook 2006, 2010; Kan et al. 2011) and situational gender performative explanations (Deutsch 2007; West and Zimmerman 1987, 2009). Of course, the gendered character of doing housework goes beyond the simple binary categorization of typically ‘feminine’ and ‘masculine’ (Goldberg 2013) but these are but crude ways the housework tasks could be categorized on aggregate level.8 Another way to distinguish housework tasks, which was also developed in the 1990s and is more predominant now, is by their routine or non-routine nature (Kan 2008; Kan et al. 2011). Thus, in tasks such as cooking and cleaning respondents have little control over their schedule because the task is routinized. This type of task is referred to as a routine indoor task. On the other hand, maintenance can be dubbed a non-routine housework task since for the most part it does not have to be performed on a regular basis. Grocery shopping falls into routine but outdoor activity or otherwise, routine outdoor housework. Researchers increasingly view shopping as gender-neutral (Baxter 2002; Craig et al. 2015) while still routine. Coincidentally, routine housework tasks usually burden women more than men and overall a greater share of housework is still considered stereotypically feminine.9 Thus, using data from the Multinational Time Use Survey, Kan, Sullivan, and Gershuny (2011) show that women still do most of the routine housework whereas men have increased their involvement mainly in non-routine domestic work. Similarly, Guppy and Luongo (2015) show that domestic work remains gender segregated, where women do most of the routine housework and men – non-routine.                                                    8 The fact that national surveys up until this day often force individuals to choose from the only two possible choices for their gender reflects the persisting hegemony of cisnormativity in the two North American societies. The aggregate data on non-cisgender groups, however, would be laden with threats to respondents’ confidentiality. 9 In other words, routine housework is still socially constructed as ‘feminine’. 10  Therefore, whatever the explanation of gendered processes behind the clustering of housework types is, whether it is their routine character, gender socialization, or gender performance, we can expect to find differences in the types of tasks women do compared to the types of tasks men routinely do in their households. These ideas stem from the recent scholarship (Hook 2006, 2010) arguing that tasks should be analyzed separately such as cooking and other housework. Thus, Sullivan and Gershuny (2011) propose four types of housework to be analyzed separately: care work (e.g., children, elderly), food preparation and  housekeeping (e.g., laundry, cleaning), domestic travel (e.g., grocery shopping), and maintenance work (e.g., gardening, repairs, painting). Using more meticulous categories like those mentioned above, as opposed to an aggregate measure of housework, ought to help highlight housework dimensions where change is and is not occurring for groups in focus. This is especially relevant in the view of persisting gender-essentialist stereotypes and dispositions, which  are at times blamed for the tenacity of gender inequality (Charles and Bradley 2009; Hakim 1998). It is also important to distinguish housework from child care. Although there are many instances in the early literature considering child care and parenting as housework (Ferree 1990; Hochschild and Machung 1989; Thompson and Walker 1989), overall, researchers of unpaid labour acknowledge that parenting is a more complex phenomenon calling for a conceptualization distinct from housework and thus requiring an other, albeit interrelated, set of theoretical modelling than housework (Davis and Greenstein 2013; Ishii-Kuntz and Coltrane 1992). Many studies since then analyze housework (Lee and Waite 2005; Sayer 2010; Shelton 1992) and child care (Hofferth and Goldscheider 2010; Hofferth and Lee 2016) as distinct concepts. In this project, I also follow the convention and view housework as a distinct, but interrelated, concept from parenting and child care. 11  The present project focuses both on routine and non-routine housework, dividing the former into indoor and outdoor routine tasks. Furthermore, there are two big categories of the routine indoor housework: cooking and cleaning. These tasks, though, are not of the same character because as much as domestic work researchers like to suggest that all types of housework are equally unpleasant, it is not actually so empirically (Greenstein 2000). Thus, in 1998, the Canadian General Social Survey (GSS) team interviewed the respondents using two questions tapping their subjective rapport of enjoyment of housework. On average, using a five point Likert scale with five standing for high enjoyment, Canadians residents responded that they enjoyed cooking (mean = 3.29, SD=1.31 10 ) more than cleaning (mean=2.46, SD= 1.26). Therefore, the distribution of the participation in the tasks more proportionately to those with more power in the partnership might also reflect the undesirability of the task. Another type of housework is routine outdoor tasks, which refers mainly to shopping activities. This task is reported to be as unpleasant as cleaning. It has, though, a slightly higher ranking of enjoyability (mean=2.67, SD=1.23, according to the results of the Cycle 12 of the GSS). Maintenance is one of the most common non-routine tasks and it also is ranked as more enjoyable (mean=2.87, SD=1.33, according to the results of the Cycle 12 of the GSS) than shopping and cleaning but lower than cooking. These examples from the General Social Survey illustrate the inherent error in the previous literature that relied on the assumption that all housework was equally unpleasant (Greenstein 2000). I find that individual tasks differ on how pleasurable they are perceived to be and thus, the participation in the tasks and the gender gap can depend on the type of housework. Moreover, the previous studies aggregate time commitments in a way that does not take into consideration the differences that may arise due to ancestral and ethnic group belonging, or                                                    10 3 – neutral; above 3 – more positive attitudes; below 3 – more negative attitudes. 12  sexuality. I argue that, as with any type of social change, it is unlikely that the gendered division of labour has converged evenly for everyone. A dimension for comparison, mostly ignored by the previous research especially within the context of the Canadian society, is ethnic and ancestral group belonging. Due to the effects of the cultural inheritance of gender socialization traditions other than those common in North America from countries where the cultural groups claim origin from, I also anticipate that there are differences in time allocation to household tasks between women and men of various cultural groups and ancestral background. This contention also relies on and is connected to the findings on the effects of welfare states and the systemic variability on the division of household labour (Hook 2010; Treas and Tai 2016; Williams 1995).  Furthermore, another dimension worthy of comparison that I add to my study which was not yet explored in the previous literature is comparing the convergence of time devoted to housework by family types such as between heterosexual couples and couples of the same sex. The interest that drives this analysis is engendered by the question whether women and men in same-sex partnerships would not meet the same societal pressures to conform to gender norms imposed on other partnerships. Canada is a great example of a progressive society which allows diversity of family types in terms of their sexuality, making it an opportune case study for exploring the differences in same sex partners’ time allocation. Do all Canadians follow the prevalent gender norms, or do we expect same-sex couples to be more egalitarian11 (Kurdek 1993, 2006; Patterson 2000)?                                                    11 Following the convention in the literature on the gendered division of labour, I use ‘egalitarian’ and ‘equalitarian’ as synonyms with regards to the time spent on housework. By using both, however, I mean a more equal division of housework, recognizing the existence of the paradox of perceived fairness in the division of household tasks while it remains unequal in favour of men(Perales et al. 2015; Polachek and Wallace 2015; Young et al. 2015).  13  1.3 Research Questions The purpose of this project, therefore, is to examine the gender differences in the allocation of time to household labour. The study will address three principal research questions: 1. To what extent has the time allotted to an array of household tasks converged over time in the domestic work of women and men in Canada? In other words, does the time spent on all individual housework tasks converge in Canada? 2. What drives a more equitable division of labour? What are the factors behind the gender gap in time spent on housework tasks? Can the gender gap in time spent on different housework tasks be accounted for by the same theoretical framework(s)?  3.   Do factors that explain the gender gap in time allocated to domestic labour work in the ways predicted by the economic perspectives? Are there differences in results by individual housework types?     This study explores the above questions and tries to illuminate the mechanisms that propel the gender convergence in time spent on unpaid work in Canada.  14  Chapter 2: Building a Theory of Housework In order to build a theory of housework several underlying issues need delineation. First, the distinction between explaining the gender gap in time spent on housework and the variation in women’s and men’s time devoted to household tasks needs explication. Second, the bundle of tasks that entail housework needs to be clearly differentiated. After specifying these underlying issues, I turn to theoretical accounts of why time devoted to household tasks is different for women and men. Differences in factors driving the gender gap are distinct from the factors that explain the variation among women and among men (Sayer 2010), yet no study has endeavoured to analyze those factors in tandem. Explaining the gender gap involves the analysis of factors that account for the difference in women’s time spent on housework compared to men’s (analogous to explaining the gap in pay between the sexes). Explaining the variance in time devoted to housework means the analysis of how much factors can account for the variance in time spent on housework within a gender group (analogous to how different factors affect the pay of men versus women). The two are related because they analyze the contribution of explanatory factors to participation (measured in time) of women and men in housework but the former compares women to men and the latter women to other women and men to other men. Moreover, processes explaining participation in individual housework tasks vary, yet no study has provided a comprehensive analysis of the social mechanisms influencing the performance of different household tasks. In the 1990s, housework was often categorized into ‘female-typed’ and ‘male-typed’ tasks. This was done because some tasks such as cooking and cleaning were often considered stereotypically feminine (Twiggs, McQuillan, and Ferree 1999) but maintenance tasks were viewed as more stereotypically masculine (Coltrane 2000). Another way 15  to distinguish housework tasks, which is more predominant in the discussion of housework nowadays (Guppy and Luongo 2015), is by their routine or non-routine character (Barnett and Shen 1997; Starrels 1994). Thus, in tasks such as cooking and cleaning respondents have little control over their schedule because these tasks are routinized, and can be referred to as routine indoor tasks. On the other hand, maintenance can be termed as non-routine housework due to its more sporadic occurrence, whereas grocery shopping falls into routine but outdoor activity. Overall routine housework is usually associated more with feminine than masculine activities and women continue to do a greater share of housework (measured in time) (Gimenez-Nadal and Sevilla 2012; Hook 2006; Kan et al. 2011). In contrast to routine indoor and non-routine tasks, researchers increasingly view shopping as ‘gender-neutral’ (Craig et al. 2015).  Most research on how women and men share housework focuses on two types of factors influencing time allocation to housework: economic and gender-centred factors (Brines 1994; Greenstein 2000; Hook 2010). While economic factors tend to be more concentrated on the individual level and are more economic and human capital oriented by nature, the gender-centred approach tends to draw more attention to cultural and institutional factors, highlighting the constraints faced as a result of institutional inequalities or the consequences of diverse socialization. Often this categorization is simplified into the dichotomy of the bargaining and gender display perspectives (Bianchi et al. 2000; Bittman et al. 2003; Brines 1994; Coltrane 2000; Greenstein 2000; Hook 2010; Shelton and John 1996). The bargaining mechanisms are more economic in nature based on the processes of economic exchange,12 while the gender display explanation deals with the performative side of the gendered division of housework. For                                                    12 Even though the mechanism can be explained more by the economic exchange processes, the objects of bargaining can also be non-economic such as ‘sexual bargaining’ (Scanzoni 1972), divorce threats, sense of entitlement, among others (Baxter and Kane 1995; Breen and Cooke 2005).  16  example, men display gender by performing acts that are typically understood as masculine (e.g., fixing things). The gender display framework can explain gender differences in participation in domestic tasks of a routine character which are traditionally associated more with women. Displaying gender for men, for instance, leads them to avoid routine indoor tasks (Bittman et al. 2003). When a task does not have a clear gendered character at a certain historic time and when controlled for the generational effects, the bargaining perspective could be able to explain the division of household labour. Thus, partners in a household may bargain to buy themselves out of a neutral task such as shopping. This Chapter outlines the main ideas behind the economic and cultural approaches to explaining gender inequality in household labour. Most of these theories deal on the individual and interactional levels, yet they represent the reflection of structural opportunities and constraints met by individuals within the larger societal context. Previous research has shown that the variation among individuals in housework is higher than the variation among countries (van der Lippe 2010), making the analysis of factors driving gender disparities in housework within countries more pressing. Therefore, the focus of the project is on a single country, Canada, with some illustrative comparisons to the US. It is necessary to remember the limitations of the present research agenda and that the results obtained herewith may only be extended to the ‘liberal’ regimes (Esping-Andersen 1990) and only cautiously extended to countries in conservative, social-democratic, or other regimes outside of the Esping-Andersen Eurocentric regime categorization.  17  2.1  Perspectives within the Economic Approach In the following sections, I describe the main economic approaches that will be tested with regard to gender differences in participation in housework: specialization, time availability, autonomy, and relative resources explanations of the gender gap and variation in time spent on domestic tasks. These frameworks are predominantly used in the literature to analyze the variation in women’s and men’s time spent on housework (Artis and Pavalko 2003; Baxter and Hewitt 2013; Brines 1994; Coverman 1985; Greenstein 2000; Gupta 2007), while they aim to explain the gender gap in time. The interrelatedness of the issues, however, allows me to apply the frameworks both to the explanation of the gender gap in time and the variation in women’s and men’s time spent on housework. 2.1.1  Economic Approach to the Gendered Division of Housework There are three major arguments regarding the division of domestic labour within the economic perspective: time availability, autonomy, and relative resources. These arguments originate from neoclassical economic theory and the new home economics (Becker 1981). Both are based on the premise that households allocate their time in a way that maximizes the utility for the household. Historically, the approach advocated by Becker (1981) is often referred to as the specialization or human capital approach, which later evolved into the three main contemporary arguments tested in the present project. The specialization approach posits that even small differences in biological or institutional factors result in the specialization of partners where one of them does more paid work and the other more housework (Becker 1981). Thus, small advantages in factors help partners learn to specialize and develop their human capital accordingly. Therefore, a spouse with a higher wage would continuously allocate more time to 18  paid work relative to their partner, while a spouse with a lower wage would spend more time on housework.13  “The various divisions of labor among family members are determined partly by biological differences and partly by different experiences and different investments in human capital. Specialization in the allocation of time and in the accumulation of human capital would be extensive in an efficient family even if all members were biologically identical” (Becker 1981).  According to the specialization argument, the gender convergence in housework would occur only if women’s and men’s human capital were similarly developed and specialized and by the utility maximization principle, the market powers would encourage gender-equal allocation of time devoted to housework. Thus, in the case of persisting gender inequality in housework, within the specialization perspective, it is assumed that men's human capital brings higher returns from paid labour, while women's human capital can be maximized when the paid job is combined with or completely abandoned for the sake of domestic unpaid work. Likewise, gender convergence does not happen at this historical moment because of prolonged differences in human capital investments between women and men, encouraging the unequal allocation of time to housework and to a smaller extent, because of the imperfect information about skills of different actors. This approach allows for the fact that the gender convergence might also occur differently for different households. For instance, because returns from labour for the skilled professional households may well exceed the utility of performing the household tasks by themselves, people with higher returns to their human capital may buy themselves out of housework more easily than people with lower levels of human capital because their time has                                                    13 This generalization certainly does not apply to all women and men. There might be considerable differences by social class. The time of upper class women might be worth much more on the labour market and these women may choose to specialize on market activities while outsourcing housework.    19  higher value on the market and the alternative allocation of time has lower utility for the household. If the housework is performed by a hired labourer, which is more common for people of certain social status, the gender convergence in the division of unpaid housework for couples is more likely to take place faster for people with higher levels of human capital than for others, for whom market returns from their human capital does not allow the hiring of help. This type of rhetoric appears, for instance, from the qualitative interviews of women of different positions within a big corporation (BBC) to explain the consequences of the differences in earnings: “If you earn big bucks, you can afford a nanny – you have a choice – but for most of us, that is not an option – simple as that. (No. 14 Basic: Female 30s) A lot of the senior women here absolutely depend on their nannies if they have kids – and that makes a big difference in how far they can go with their career – and obviously most women are not in that position (No 15: Basic: Female 40s).” (Browne 2004).  According to the economic approaches, not only do more resources result in less housework but also the gender convergence is driven by the competitive market forces assuming that gender inequality is inconsistent with the demands of market profitability. Yet it is not intuitive why gender inequality has not resolved itself considering the abundance of information on human capital.  The empirical evidence repeatedly shows that discrimination against women in paid and unpaid work persists. The faith in an egalitarian ‘invisible hand’ of the market is perpetuated by “a pervasive myth of rising female employment” (Hakim 1995).14 The UK data, analyzed by Hakim (1995) demonstrates that workforce participation of women, on the contrary, did not                                                    14 Here, Hakim (1995) emphasizes the differences between full-time and part-time employment. 20  change much in the period from 1851 to the late 1950, and the full time employment rates remained virtually the same from 1841 to 1993 because the majority of women are still occupying low status part-time jobs. On the other hand, in Canada, Guppy and Luongo (2015) show that despite a significant rise in overall participation of women in labour force in the 20 th century with a precipitous acceleration in 1950s and steady increase until 1980s, it was stalled since the 1980s and the most increase happened for part-time jobs rather than full-time. A similar situation is reported in the US with an observed peak of women’s employment in 1999 and slow decrease in the first half of 2000s (Bianchi and Milkie 2010). Analogously to job satisfaction with lower wages than those of men, there is a plethora of recent studies on the paradox of the perceived fair division of housework where women do most of the housework but still perceive this arrangement as fair (Perales, Baxter, and Tai 2015; Polachek and Wallace 2015; Young, Wallace, and Polachek 2015), usually reported as a result of the resolution of the time-based conflict (Young et al. 2015) or life transitions such as parenthood (Perales et al. 2015). One of the limitations of the economic approach is the assumption that as long as women are encouraged to allocate their time in developing their human capital to make it sellable in the labour market, their parity with men with respect to economic returns will be eventually reached. However, even with the seemingly similar hours of work of the modern women and men, there is a sizeable difference in how time is compensated or whether opportunities are available for each gender group (Kay and Hagan 1998). To address this limitation, the approaches within the economic camp not only tackle the differences in time availability but also look into market compensation of their human capital in terms of their income and other accumulated resources. Thus the approaches discussed further start with the time availability argument which directly 21  flows from the specialization perspective and then addresses two resource-driven arguments, the autonomy and relative resources arguments. 2.1.2 Time Availability Perspective The time availability, or time constraint, argument posits that a partner who has more available time performs more of the unpaid work in a household (Artis and Pavalko 2003; Coverman 1985; Presser 1994; Silver and Goldscheider 2013). This perspective is a direct extension of the specialization perspective and Becker’s ideas, sharing similar strengths and shortcomings as the original. Within this approach, a partner who has more time chooses to do more housework, sometimes even in a case where they could allocate resources into more profitable activities because of the assumed necessity of specialization – someone has to do the housework in a partnership. The time availability perspective is also closely intertwined with time use allocation and treats time as a scare resource which individuals have limited supply of (Silver and Goldscheider 2013), thus being constrained by the available time. That is why this perspective is also often referred to as the time constraint approach – people have only 1440 minutes each day to allocate to various activities. Gender convergence in unpaid work would result from the equitable availability of spare time for both partners in a household and with changes toward more egalitarian allocation of time among people in general. Because the partner who works more hours for pay has less time to do housework, with time the increased specialization would result in increased productivity of the partner’s time spent on paid work and decreased productivity of the time spent on housework, leaving housework to the partner who had more time to allocate to unpaid work to start with and thus having developed more human capital conducive to housework. Moreover, studies show 22  that there are generational differences in how women and men allocate their time and how much they, especially women, choose to spend on housework instead of paid work. Thus Artis and Pavalko (2003) report a considerable change in the socialization patterns among women and men regarding the division of domestic labour since 1970s. Thus the gender convergence might reflect cultural changes in attitudes toward how to allocate time. This approach shares the same limitations as were discussed above with the specialization approach, namely, more allocation of time does not always result in higher returns from invested time. In fact, we live now in the age that the poor work longer hours and gain lower wages (Chetty et al. 2014; Jäntti et al. 2006). Most of the wage structure depends on the factors outside of the individual – on the structure of market opportunities. Another flaw of this perspective is the assumption that the available time outside paid work would be spent on housework (Presser 1994). There are other options as to how to use one’s spare time and a likely option is spending time on leisurely activities. Thus men are more likely than women to spend their free time on leisure rather than on housework (Mincer and Polachek 1974). There is a tendency in the literature analyzing housework to simplify daily activities into the dichotomy of paid work and housework, disregarding another big part of the human life – leisure. Additionally, Coverman (1985; Coverman and Sheley 1986) emphasizes the presence of children, both of pre-school and of school-age, finding that these two are significant predictors of husband’s engagement in housework together with the time spent on paid work. In fact, she finds that the time availability framework is the most important in predicting time spent on housework for men compared to the relative resources and the ‘sex role ideology’ perspectives. Similarly, Artis and Pavalko (2003) 23  show that the changes in time availability has significantly contributed to the changes in the division of household labour in the period between 1974 and 1988, using longitudinal analysis.  2.1.3 Autonomy or Absolute Resources Perspective One of the limitations of the time availability framework is the fact that the time of different individuals is of different worth in the labour market. This limitation is addressed by the absolute resources or autonomy approach to explaining the gender gap in housework. Under this perspective, resources other than time, particularly compensation for time such as wages or personal income, are considered to be strong predictors of participation in housework. Availability of resources provide women and men with the ability to allocate their time and resources to other activities rather than housework.  This argument is conceptually broader than the relative resources argument (discussed in the following section) because it allows the explanation of the gender gap to rely on other resources and processes in decision-making about the division of the housework rather than just bargaining and negotiation with the partner. Having resources, in general, is assumed to provide individuals with more freedom to decide about their time and resources and their allocation. This argument is also empirically driven. Thus Baxter and Hewitt (2013) summarize many US studies that find that women’s absolute earnings rather than their earnings relative to spouses are more important in determining how much time they would spend on housework. They explain that in the opposite contexts, such as in Australia, where relative earnings are considered a stronger predictor of housework participation, the social forces favour the traditional male-breadwinner institutional framework compared to countries where we find that absolute earnings matter more. Similarly, Gupta (2007) argues that it is the autonomy of having one’s own resources and not 24  relative resources that matter more in the division of housework. That is, according to him, it is a woman’s absolute earnings not her earnings relative to her partner/spouse that contribute to the gender gap in domestic labour. The convergence of the gender gap in the division of household labour occurs when women reach a similar level of autonomy with men and have more of their own resources. These individual resources are also considered to be a leverage in the event of the dissolution of the partnership, therefore, having something to rely on if the marriage does not work out. Thus, researchers in the field argue that it is the availability of alternatives outside the partnership and related portable resources that a spouse will still have in the event of a divorce that results in the unequal division of housework (England and Kilbourne 1990; Hobson 1990; Lundberg and Pollak 1996).  The inherent focus of the autonomy argument on the individual rather than the interaction between partners limits the scope of the explanation of the gender gap in housework to the resources belonging to a partner rather than to the relative resources. The argument, therefore, lacks the interactive side and disregards power relations unlike the relative resources argument, discussed in the following section. However, contrary to the relative resources argument, the autonomy argument (1) allows for the differentiation between couples that make joint decisions on household-related issues and those that divide the decision-making areas between spouses, or two types of “equalitarian” marriages – “syncratic” and “autonomic” (Blood and Wolfe 1960); (2) captures the idea of class, at least partially because it does not narrow down the focus on the value of the differences in resources between partners but puts more attention on the absolute value of a partner’s resources. Thus, a woman might make relatively more than her partner in the 25  labour market but still not be able to afford extra help around house, which a woman, who earns less than her partner but more in absolute terms, can hire with her own earnings. 2.1.4 Relative Resources Perspective The relative resources perspective results from the interactive approach to the division of labour. It centres attention on the power relations in a partnership, coming in response to the feminist critique of the economic theorists’ take on the domestic division of labour. This perspective incorporates ideas of domination and gender conflict within a household. Blood and Wolfe (1960) present evidence that men’s power in households increases with higher levels of education, income, and occupational status compared to their partners. The relative resources perspective considers the interrelatedness of the partners in a household as the main drive for decision-making, thus simplifying the process of partners’ joint decision making to bargaining between partners, where a partner with more relative resources has more power to bargain, and in terms of unpaid work, buy themselves out of housework. Thus implicitly the relative resources argument assumes that individuals with conflicting interests both strive to do as little housework as possible and bargain over who is to perform certain domestic tasks (Blood and Wolfe 1960; Heer 1963; Lloyd and South 1996; Manser and Brown 1980; McElroy and Horney 1981).  Due to its reliance on the explanation through the joint decision-making and bargaining process, the relative resources perspective is often referred to as the bargaining perspective. The approach draws focus to the power balance and bargaining with available resources (Blood and Wolfe 1960; Rodman 1972). And because at its core there is the process of negotiation between partners, it is often myopic to other types of decision-making, based on processes other than bargaining. This limitation of the perspective was discussed by the proponents of the more 26  autonomous decision-making in a household such as Gupta (2007) and Baxter and Hewitt (2013). Gupta (2007), however, focuses more to the resources themselves rather than the processes that underlie the division of labour. It benefits the research field to emphasize that negotiation through bargaining is not the only way through which decisions can be made in a household. Already in the 1960s, Blood and Wolfe (1960) distinguish between “syncratic” and “autonomous” decisions as discussed in the section above. What this practically means is that decisions in the egalitarian households, where partners have some equality in voicing their opinions when it comes to household decisions, can not only occur via bargaining and joint decision-making but also through an initial division of ‘spheres of influence’, or domains (Heath 1976),that allows partners to have more autonomy in making decisions within their own ‘areas of expertise.’ Inevitably, these types of decisions might depend more on ‘own’ autonomous resources rather than relative resources compared to their partner’s. The convergence of the gender gap under the relative resources explanation occurs simultaneously with the increase in equality between women and men in resources that they bargain with. Thus, in their work Husbands and Wives, Blood and Wolfe (1960) argue that the shifts in power to make decisions within the family are the result primarily of the changes in the share of resources, or the relative resources of partners. Even though they show that the division of housework did not change much in their Detroit study, and most of the housework still stayed gendered, they did, however, emphasize the importance of the relative resources to the balance of power within a household. Whether the lagged progression toward gender convergence in housework that we experience nowadays is the result of the change in the distribution of 27  resources can be tested by analyzing how sensitive the allocation of time to housework is, depending on the level of wives’ reliance on their husbands.  More precisely, gender convergence, according to the bargaining perspective, can be achieved when the gap between relative resources of spouses is minimized (Bianchi et al. 2000; Evertsson and Nermo 2004). The argument first led many to conclude that  the increase in bargaining power of women will happen with their specialization and more time spent on their human capital on the labour market (Becker 1965) yet the empirical results again and again show that this is not always true (DiPrete and Buchmann 2013; Hakim 1995). The dependence of women on men for economic resources persists even with the increase of hours women spend on paid work because for the most part, the subordinate position of women at home reflects the equally subordinate and often menial character of their paid work, leaving women with “no exit” (Meissner et al. 1975). The analysis of the relative resources, therefore, improves on the specialization and time availability perspectives because it reflects the necessity to account for power inequalities between partners. However, its greatest contribution is also its biggest flaw because this perspective underplays the importance of processes other than bargaining in a households’ decision-making. Julie Brines (1994) devised one of the most compelling theoretical developments and testable specifications of the economic exchange in housework, specifically of the bargaining perspective. She proposed specifications for the link between household labour participation and relative economic contribution, measuring the relative income contributions of partners by a variable called income transfer.15 This effectively captures the relative share of the income                                                    15 (Incomepartner1 – incomepartner2)/ (Incomepartner1 + incomepartner2) 28  contribution of each partner to the household income. I discuss her specifications in more detail in the following sections to emphasize the importance it plays in testing a unified theory of the gendered division of housework. Her framework draws from the relative resources perspective and thus shares most of its limitations such as an overreliance on the power-related bargaining process in household decision-making. 2.2 Gender-centred Perspectives The economic approaches have been contested by the gender-centred perspectives, showing that even in the woman-breadwinner families, ‘gender trumps money’ (Bittman et al. 2003; Brines 1994). The economic perspectives often claim that they rely on processes that are ‘gender-neutral’ and that they would work for people of any gender. As long as there are differences in the resources, there are differences in how partners share housework. Often, however, the reality of the division of domestic labour does not meet such expectations and represents by itself a historically highly gendered process. The gender-centred approach attempts to tackle the problem of gender inequality via the prism of self-awareness of one’s own gender and the rights and responsibilities associated with it (Gerson and Peiss 1985). The omnipresence of gender is also apparent in the paradox of why women are satisfied with the arrangements of the unequal division of domestic labour (Acock and Demo 1994; Demo and Acock 1993; Perales et al. 2015; Polachek and Wallace 2015; Treas 2010; Young et al. 2015). The following section describes in more detail this opposing framework borne out of the feminist criticism of the economic explanations of the gendered division of housework.  29  2.2.1 Gender Perspective The gender-centred approaches to the quantitative analysis of the division of housework evolved from arguments putting an emphasis on gender roles in 1980s and 1990s (Blair and Lichter 1991; Coverman 1985; Presser 1994) to a more performative approach highlighting gender display (Baxter and Hewitt 2013; Bianchi et al. 2000; Gupta 2007), and then to a more comprehensive perspective combining socialization, roles, expectations and performance – ‘doing gender’ (Artis and Pavalko 2003; Bianchi et al. 2000; Brines 1994; Butler 1990; Evertsson and Nermo 2004; Greenstein 2000; Hook 2010; Ting, Perales, and Baxter 2015).  The ‘doing gender’ perspective within the literature on the gendered division of household labour is itself based on the theoretical work of West and Zimmerman (1987). This interactive approach emphasizes the performative side of gender, stating that every situation and with it, all households, are in fact “site[s] for doing gender” where men and women repeatedly and constantly perform and reaffirm their gender identities (Berk 1985; Goffman 1976; West and Zimmerman 1987).  In reality, however, the ‘gender display’ (a name often used for the gender-centred perspectives in general) is a bit of a misnomer for the ‘doing gender’ perspective as developed by West and Zimmerman (1987), and with the ‘doing gender’ as developed by Butler (1990). The former framework employs not only the performative side of gender as it is traditionally understood (Goffman 1976), but it also relies on a more ethnomethodological approach to the concept of gender. Thus, for the ‘doing gender’ approach, gender is omnipresent in everyday interactions where women and men achieve16 it in all situations. This gender achievement is                                                    16 Achievement of gender is distinct from gender performance (West and Zimmerman 1987). 30  distinct from the gender display because in a ‘display’ each scene is divided into actions and performing gender becomes an action by itself. Being an action, therefore, gender is not omnipresent in all actions and interactions but only relevant during the performances of gender. West and Zimmerman (1987) criticize the view of gender in the gender display framework where it is located mostly on the fringes of other actions sometimes deemed as ‘more important’ or more central. Instead, they assert that gender is a feature of all actions and interactions and cannot be, and is not, secondary.  On the other hand, as a critique to these theatrical approaches to gender display, the development of idea of the ‘doing gender’ by Butler (1990) provides another view of ‘gender’ as a social construction17 that does not exist except in these performances. Gender is not something innate, belonging to an individual, but a construct outside of the individual and existing in all performances not only in interactions. She views gender as a result of those performances and the direct outcome of the subtle and blatant social sanctions to comply with gender roles. Both West and Zimmerman (1987) and Butler (1990) emphasize the social construction of gender and its omnipresence. For example, driving a car, traditionally men’s prerogative, nowadays is a more common activity among both women and men. While the abstract idea of the activity itself might not be gendered, the way an individual does the activity may reflect a gendered performance. What type of a car one is driving? How one is holding the steering wheel? Whether one criticizes other drivers while driving? All of these are gendered performances. Thus gender is these performances.                                                    17 Unlike West and Zimmerman (1987), Butler also views ‘sex’ as a social construction. 31  This misnomer ‘gender display’18 is widely used in the literature analyzing gendered housework (Bittman et al. 2003; Brines 1994; Evertsson 2014; Evertsson and Nermo 2004; Gupta 2007), where researchers draw focus on the behavioural representations of gendered identities rather than specifying the process by which gender19 becomes relevant in the domestic realm. It is, however, sufficient for the quantitative studies which aim predominantly at exploring whether the division of household labour aligns with the expectations of the economic and resource-driven explanations. And if it is not, the division of housework is deemed to occur due to other gendered processes rather than those born out of the labour market differences. Specifically, gendered performances become more evident in situations when an individual deviates from established gender roles through their actions in other aspects of life such as on the labour market and then compensates through increased performance in other domains such as domestic labour.  This process is called gender neutralization technique (Greenstein 2000; Sykes and Matza 1957). A neutralization of gender deviance in the labour market is the result of social control and especially of the internalized control, i.e. self-control because of the internalization of gendered attitudes and values and the traumatic experiences of gender socialization (Edlund and Oun 2016). The complexity of the interrelation between the gender display and ethnomethodological explanations led me to refer to the present approach as the gender-centred framework. Even more broadly, the gender-centred framework falls within the cultural narrative (Blau, Brinton, and Grusky 2006) because gender norms are a part of broader cultural and social norms. Gender                                                    18 For understanding gender by Butler (1990), the phrase ‘gender display’ becomes tautological because gender is the display. 19 Or more precisely, the belief in gender identity, created as a result of retrospection on prior gender performances (Butler 1990). 32  narratives around housework fall within the broader (heteronormative) social context (Goldberg 2013), within which the gendered ‘character’ of certain tasks is constructed. For example, in North American societies, ‘cleaning the house’ is socially constructed and mutually agreed as a ‘feminine’ task, whereas ‘fixing things around the house’ is socially constructed as a more ‘masculine’ performance.    The gender-centred framework posits that women and men follow social norms in their behaviour to achieve gender. Achieving gender is simultaneously institutional and interactional, working on both levels, because accountability is the innate feature of social relationships (West and Zimmerman 2009). For instance, traditionally, women spend more of their time on housework in the dependent positions when their share of contribution to the household income is small. Such traditional behaviour goes hand in hand with the predictions of the economic approaches. Furthermore, women are also expected to do more housework when they are the main breadwinners where their economic contribution exceeds their partners’, thus achieving their traditionally defined femininity in housework in order to neutralize the non-normative gender behaviour in the labour market (Greenstein 2000). The husbands’ pattern predicted by the gender-centred perspective, on the other hand, assumes less housework for higher earning husbands and similarly, less housework for dependent husbands who earn less than their partners because men in the dependent positions are expected to display their traditional gender in the household by compensating for their non-normative gender behaviour in paid work (Brines 1994).    Brines (1994) shows that the gender-centred model has a higher predictive power for men in the dependent positions, where their belief in gendered identity results in performances 33  consistent with traditionally ‘masculine’ acts such as eschewing core housework, even though these men do not provide for their families as much as their partners. Greenstein (2000) also shows that gendered performances of breadwinning women are also consistent with a more traditional construct of ‘femininity’, where they do more core housework, even though they earn more than their partners. Similar results were obtained by Bittman et al. (2003) for Australian women and by Evertsson and Nermo (2004) for American women but not for Swedish women. The gender-centred framework predicts patterns that are distinct from the expectations laid out by the economic exchange perspective. The predicted patterns between the relative economic contribution and housework participation under the gender display explanation would resemble those illustrated in Figure 1 (adapted from Brines (1994)).  Figure 1 Gender-centred Model  The traditional wives’ pattern follows the upper curve in Figure 1. Women do more housework in the economic dependence position when they contribute fewer economic resources 34  to the household (upper left quadrant of Figure 1). They also do the least amount of housework compared to other women when their contribution to the share of household income is roughly the same as their partners’. Additionally, according to the gender display perspective, breadwinning women are also expected to do more housework compared to women in households where both partners contribute equally. This situation of imbalanced housework participation for breadwinning women occurs because breadwinning women tend to do more housework in order to neutralize the non-normative gender behaviour in the labour market and to achieve their gender. The traditional husbands’ pattern predicted by the gender display perspective, on the other hand, follows the lower curve in Figure 1 – less housework for breadwinning husbands, and similarly less housework for dependent husbands who earn less than their partners because men in dependent positions compensate for their non-normative gender behaviour in the labour market (lower left quadrant). Thus the gender adherence becomes more pronounced for women in breadwinning positions and for men in dependent positions. In other family permutations, the economic predictions coincide with the gender-centred explanations.  The ability of the gender-centred framework to incorporate both the economic and gender-driven explanations for the division of domestic labour makes it a potentially more potent explanation than the economic approaches. Yet one of the main shortcomings of the framework is its process-oriented approach which circumscribes it within less specifiable concepts and its association with enormous difficulty of measurement. Brines (1994) in her work, provides us with a framework that can test at least the situations where the gendered processes overrun 35  economic predictions and establish the rule of gender achievement in the production of housework. 2.2.2 Focusing on Resistance and Undoing Gender in Housework In 2007, Deutsch published an article titled Undoing Gender20, as an extension to the original West and Zimmerman (1987), which was supported by other feminist researchers (Risman 2009). She argues that the concept of doing gender inevitably limits the focus to the normative gender behaviour, yet individuals are capable and in fact very often show resistance against conventional gender relations. She pushes the emphasis of the approach of gendered relations toward gender deconstruction and introduces a new term – ‘undoing gender’ – by which she means social interactions that reduce gender differences, or specific acts and behaviours that accomplish this agenda. On the other hand, Butler (2004) does not view ‘undoing gender’ as the matter of individual agency, she asserts it outside of the individual in the intricate web of social interrelations. The belief in the ‘undone’ gender, however, also comes from the changes in the larger social fabric and reflects the shifts in the association of certain acts with gendered performance.  All in all, the ‘undoing gender’ approach draws our attention to identifying conditions under which the traditional association of belief in gender (Butler 2004) no longer matters or matters less, not just different. When there is still difference in gendered performances around housework, is the difference reduced over time? What are the conditions that can potentially reduce the differences in the domestic division of labour and are reducing the gendered association of certain housework activities with gendered performance across time? Deutsch (2007) also emphasizes that it is not the difference that is problematic but the power differentials                                                    20 The original article also included an apology to Judith Butler, who published a book in 2004 with the same title. 36  between women and men. She argues that we need to not simply change the structural arrangements but also the dynamics of power that underlie those arrangements because without changing the biases, the higher female presence in an occupation would only lead to feminization of it rather than lessen the social construction of power differentials between women and men. Thus we need not just look at the differences where gender can be undone by more participation in non-normative behaviour but actively seek what structural factors drive the ‘undoing gender’ and the arrangements that render gendered performances in housework less pronounced or even irrelevant.  The doing gender approach helps us understand why certain performances remain gendered when women contribute the same amount of resources to the household (Berk 1985) and why gender norms might be coerced more strictly when women earn more than men in a heteronormative household (Greenstein 2000). Yet the degree of the variability is not much of the focus of the ‘doing gender’ perspective for which Deutsch (2007) criticizes West and Zimmerman (1987) because such focus renders the performances associated with ‘undoing’ gender invisible. The conditions of undoing gender, furthermore, are also important because they reflect where and how the societal shifts regarding the association of certain acts with a gender occur and these conditions can be different for different cultural groups and social classes. The achievement of general differences between incumbents of different gender, ethnic, and class identities are also referred to as “doing difference” (West and Fenstermaker 1995).  In the present project, I try to avoid sematic differences, draw focus to the conditions that drive the acts of non-normative gendered performances in housework, thus emphasizing the conditions underlying ‘undoing’ gender given the process that gender is continuously ‘done’, ‘redone’(West and Zimmerman 2009), and ‘undone’ (Butler 2004). I focus herein not in the 37  characteristics of gender differences but rather on characteristics of gendered housework and the preconditions that change associations of housework with gendered distributions of unpaid domestic work. Thus I focus on conditions that ‘undo’ the gendered character of the relationships of beliefs in gendered identities (women and men) to the types of housework as understood by Butler (2004), and distinguish these relationships from gender, the achieved feature of individuals, which cannot be completely ‘undone’ as conceptualized by West and Zimmerman (1995) and Deutsch (2007). Thus, I use concepts of ‘undoing gender’ in in housework in a way that housework is seen as the reflection of the social (de)construction of gendered associations of housework acts with gender performance but not the situations where individuals are viewed as actors of such undoing.  2.3 Unifying Framework In this section, I will bring together the economic and gender-centred perspectives into testable specifications extending the work of Brines (1994). The economic expectations are discussed first, then I introduce how Brines (1994) proposed to test the relative resources and gender-centred approaches. I further extend her ideas to the time availability, the autonomy, the variations in the ‘undoing’ gender approach specification, and add more on the gender neutralization specifications. I conclude this Chapter with hypotheses predicting the outcomes with regard to individual housework types, using the frameworks, the proposed and extended specifications.   38  2.3.1 Brines and Greenstein’s Testable Specifications and Their Extensions Two main specifications lying within the economic theory expectations correspond to the predictions of the bargaining perspective. The linear and cumulative disadvantage models of the bargaining perspective are represented graphically in Figure 2 (adapted from Brines (1994)). Figure 2a represents the linear relationship between the relative economic contribution of a partner to their participation in housework. In this model, participation in housework is inversely proportional to the economic contribution to the household. That is, the higher the relative contribution (X-axis), the lower the participation in housework (Y-axis). Figure 2b outlines a similar economic exchange scenario in which the contribution of fewer economic resources results in an even greater proportion of housework assumed. This scenario represents the cumulative disadvantage model, or the cubic specification. Conversely, a higher relative contribution to household income is associated with even lower participation in housework. Thus, the cumulative advantage model predicts the same cubic specification where more economic resources would let a partner do even less housework. The cumulative ‘advantage’ model is not explicitly broached in Brines (1994). She points out, instead, that a flatter line is a more likely outcome for higher-earning partners, meaning that the amount of housework for breadwinners would be similar for all levels of relative economic contribution, while the dependent partners will assume increasingly more housework with more economic dependence on their spouses. 39   Figure 2 Bargaining Linear and Cumulative Disadvantage Models  Similar specifications can be extended to other economic approaches accounting for the division of household labour: the time availability and the autonomy perspective. Thus the linear model would apply to situations where the predictions of the time spent on housework would depend on whether more or less time is spent on paid work or leisure activities. The more women and men would spend time on paid work and leisurely activities, the less time they will be expected to spend on housework. Likewise, women and men who have higher income in absolute terms are expected to have more autonomy with regard to having the option to shirk housework. Therefore, the relationship between personal income (or other personal resources) is also inversely associated with time spent on housework, resulting in the same patterns depicted by Figure 2. Brines (1994) also developed specifications for the gender display model where women who provide a higher share of the household income do more housework than women who earn as much as their partners because breadwinning women try to neutralize their gender non-normative behaviour on the labour market. Conversely, she specifies the gender display model for men who are in dependent positions. These men are expected to do less housework than men who earn as 40  much as their partners to neutralize non-normative gender behaviour in paid work. These specifications are graphically represented in Figure 1 and discussed together with the gender-centred perspective. Moreover, Brines (1994) finds that the display model has more explanatory power than the bargaining model for the housework participation of husbands. Greenstein (2000) extended Brines’ findings establishing the similar gender effect for breadwinning women  in the United States. American women who earned more than their partners in the 1980s in the United States, did more housework to compensate for their non-normative gender performance in paid work.  2.3.2 Limitations of the Specifications Brines’ (1994) theoretical specifications for the association between relative resources and housework are based on the assumption that all housework is stereotypically feminine and that bargaining oneself out of housework is the most likely rational decision for any individual. In the present dissertation project, I argue that it is important to separate domestic tasks by their type because not all tasks are considered equally unpleasant (Blair and Lichter 1991). Thus, my first line of criticism comes from the assumption that all housework tasks are similar and would result in the same patterns of association with the explanatory factors. The second line of criticism follows the path of extending the testable specifications to other theoretical frameworks and not only the relative resources and the bargaining mechanisms in the domestic division of labour, especially following the criticisms of Gupta (2007) and the autonomous decision-making process within household partnerships. Moreover, the over-reliance on the relative resources may conceal the differences in social class and the association of housework tasks with a gender (Ehrenreich 2001; Gupta 2007). 41  Furthermore, it is important to separate the factors aimed at explanation of the gender gap from the factors explaining variation in time allocation to household tasks among women and men, even though the processes and therefore the factors themselves are interconnected, the gender gap and within-gender variation are different concepts, not to be confused with each other. In the following subsections, I propose a couple of further specifications and put forward hypotheses with regard to the explanation of the gap and of the variance among women and among men by individual tasks. 2.3.3 Gender Neutralization – The Ignored Dimensions There are other specifications that can be envisioned with regard to the association between resources of an individual in a household and housework that this partner provides. First, participation in housework may not be associated with the amount of resources provided. Thus the relationship between the two may be invariant for all or, at least, some levels of resources, relative or absolute. This case is represented in Figure 3 and was not explicitly presented by Brines (1994).   Figure 3 Constant Association between Resources and Housework 42  The situation of no association between resources and participation in housework may indicate a more egalitarian division of housework between women and men or an unequal division of housework which is unrelated to the amount of resources partners bring to households on average. Moreover, it is worth emphasizing that the pattern presented in Figure 3 does not have to be present for all levels of resources but can also work only for some levels of relative resources. For instance, there can be almost no association between relative resources and participation in housework for dependent and equal-earning partners while there is some association for breadwinners. Empirically, a situation like this would pronounce itself in a more equal variation for lower levels of resource contribution to a household with a more certain association for higher levels of resource contribution. More specifically, in such a hypothetical case, we could observe that the left and the middle parts of the graph as in Figure 2 standing for dependent and equal-earners would be level as in Figure 3. This specification represents no association for dependent and equal-earning partners and the right side of the graph (where the breadwinners are) would show a linear association suggesting some distinct pattern for breadwinners.  These ideas can be equally extended to other frameworks such as the time availability and autonomy approaches, where instead of the association of the participation in housework with relative resources, the analysis would focus on the association with absolute resources and the amount of time spent on activities other than housework, more commonly in market activities. It is more relevant to situations where we have fewer cases for some of the groups for instance for the dependent men, we might have not enough observations and the spread of the variation for the group would be greater to have any confidence in establishing a statistically significant 43  association between resources and participation in housework. Ideally, though, the situation with no association would also be characteristic of two extreme cases: (1) when women’s and men’s contribution to housework is independent of their relative or absolute resources, while the gender gap is closed; (2) when women’s and men’s contribution to housework remains on the same level regardless of their resources, while the gender gap remains sizeable. The former situation would suggest reaching gender equality in the division of labour regarding the housework task, while the latter would suggest inelasticity of the housework participation to the contribution of resources and the persistence of gender inequality in households regardless of one’s contribution in the form of economic resources. Second, the situation where the gender neutralization process would be present for any level of working women is also possible. It can be that working women in general would tend to do more housework with the increase in their absolute or relative income, not only women who bring more income than their partners. Such a situation is particularly likely if any paid work done by women is considered a violation of gender norms and requires neutralization techniques in a given society or a group. Thus, for these situations, we could observe an increasing participation in a type of housework with increasing amount of relative or absolute resources as in Figure 4, specifically among women. 44   Figure 4 Gender Neutralization Specification: Linear and Cumulative Variants Figure 4 represents two main possibilities of such case. First, the linear association of housework with the relative share of household income (Figure 4a). With each unit of increased share of contribution to household income, a woman would also increase her participation in housework so as to neutralize the violation of gender norms in the labour market. Second, a cubic model is represented in Figure 4b, which I refer to as the cumulative neutralization model, where breadwinning women, especially the ones that contribute close to 100% of the household income, have to work twice as hard at home to neutralize gender norm violation in the labour market. In general, any upward trend on the whole graph or part of it suggests an association clashing against the predictions of the economic frameworks. 2.3.4 Housework Task Hypotheses  The hypotheses with the relation to the research questions are summarized in Table 1. In my analysis I want to explore (1) whether the gender convergence of time spent on housework is achieved in individual housework tasks, including in the tasks that are considered stereotypically feminine such as the routine indoor tasks. While I expect to observe the pattern of convergence in time spent on housework tasks in Canada as in many other countries (Gimenez-Nadal and Sevilla 2012; Guppy and Luongo 2015; Hook 2006; Kan et al. 2011; Robinson and Godbey 45  1997; Sayer 2005), because the performance of gender-neutral tasks should have weaker association with gender, I expect the following results with regard to the gender gap in individual housework tasks:  Hypothesis 1.1. The gender gap is expected to be narrower in gender-neutral tasks such as shopping than in housework tasks that are traditionally gendered such as cooking and cleaning, which are seen more frequently as women’s as opposed to men’s work. The second question that I aim to address is: (2) what theoretical perspective (one of the economic perspectives or the gender-centred perspective) explains the gender gap in women’s and men’s time spent on individual housework tasks. This question is driven by the identified research lacuna in the analysis of the actual gender gap along with the lack of a dominant consensus around the predictive power of theoretical frameworks (Bianchi and Milkie 2010). Because economic mechanisms are assumed to work regardless of gendered expectations (Baxter and Hewitt 2013; Brines 1994; Greenstein 2000; Gupta 2007) resulting in the patterns laid out by the rational choice and economic exchange theories, I expect the following to be observed with regard to the gender gap in time: Hypothesis 2.1. The gender gap in time is more likely to be explained by the availability of resources, relative or absolute, including time, for the tasks that are more gender-neutral such as shopping compared to tasks that are still considered more typically feminine – such as routine indoor housework tasks, cooking and cleaning.   46   Table 1 Summary of the Hypotheses by Research Question Question 1. Whether the gender convergence of time spent on housework is achieved in individual housework tasks, including in the tasks that are considered stereotypically feminine such as routine indoor tasks? Question 2. What theoretical perspective (one of the economic perspectives or the gender-centred perspective) explains the gender gap in women’s and men’s time spent on individual housework tasks. Question 3. How factors that drive the changes in the gender gap in domestic chore time work to predict the variation within gender groups? Are these factors associated with women’s and men’s time spent on housework tasks in the predicted ways? H1.1. The gender gap is expected to be narrower in gender-neutral tasks than in housework tasks that are traditionally gendered. H2.1. The gender gap in time is more likely to be explained by the availability of resources, relative or absolute, including time, for the tasks that are more gender-neutral compared to tasks that are still considered more typically feminine. H3.1 Both men and women are expected to do gender in all types of housework.  H2.2. Relative resources are expected to account for more of the gender gap than absolute resources for all gendered tasks than for more gender-neutral tasks. H3.2. Women who have more resources, relative or absolute, are expected to do less housework compared to women who have fewer resources, especially in housework tasks that are more gender-neutral.   H3.3. Men are expected to perform less domestic labour in all types of housework with the more resources that they provide, relative or absolute.   H3.4. Women are expected to do gender in the least enjoyable tasks still typically associated with feminine activities.  Because the relative resources perspective relies more on the relations of power reflecting the gender subordination and domination relative to the partner (Blood and Wolfe 1960; Heer 1963; Rodman 1967, 1972) compared to the absolute resources framework, which reflects only the resources of the respondent without taking into consideration the interrelation, I expect the following results:  Hypothesis 2.2. Relative resources are expected to account for more of the gender gap than absolute resources for all gendered tasks such as routine housework tasks, cooking and cleaning, than for more gender-neutral tasks. 47  The third question in this project is: (3) how factors that drive the changes in the gender gap in domestic chore time work to predict the variation within gender groups? Are these factors associated with women’s and men’s time spent on housework tasks in the predicted ways? Based on the previous research by Greenstein (2000), I propose the following hypotheses for testing: Hypothesis 3.1 Both men and women are expected to do gender in all types of housework. Based on the expectations of the economic perspectives, I expect that the result will show the following: Hypothesis 3.2. Women who have more resources, relative or absolute, are expected to do less housework compared to women who have less resources, especially in housework tasks that are more gender-neutral. Hypothesis 3.3. Men are expected to perform less domestic labour in all types of housework with the more resources that they provide, relative or absolute. Since the gender gap in participation in housework is reported to be converging due to the shift in gender roles (Gimenez-Nadal and Sevilla 2012; Guppy and Luongo 2015; Hook 2006; Kan et al. 2011; Marshall 2011; Robinson and Godbey 1997; Sayer 2005), I expect that even if I find that women’s participation in housework is not following the predictions of the economic perspectives, then: Hypothesis 3.4. Women are expected to do gender in the least enjoyable tasks still typically associated with feminine activities such as cleaning.   48  Chapter 3: Data and Methods  3.1 Datasets and Samples There are two main datasets that I use for this project. First, five time use cycles of the Canadian General Social Survey (GSS), 1986-2010, are the main datasets analyzed for Canada. Second, the American Time Use Survey (ATUS), 2003-2015, is the dataset for the US (Bureau of Labor Statistics 2015; Hofferth, Flood, and Sobek 2015).21 Both national surveys are based on time diaries. In time-use diaries, respondents reconstruct their previous day from memory and report it in the temporal sequence of each activity for 24 hours starting at 4 am in the morning.  Respondents can report the duration of activities in minutes. It is recorded as a multiple of 1 minute (no activity records seconds). However, because people usually think about their time in the multiples of 5 minutes, the distribution of the activity duration variable spikes every 5 minutes, available in the episode file for the time use surveys. Thus, 6.35% of all episodes in the GSS 2010 activities are of 5-minute duration, 9.14% of 10 minutes, 10.81% of 15 minutes, and 7.26% of 20 minutes, whereas the episodes of durations in between these modal values constitute below 1% of all recorded episodes.  The respondents are asked to report their activities in sequence and asked the related questions regarding the length, the location and who the respondent was with (Statistics Canada 2011). For example, the prompt for the respondents was the following for the GSS 2010, Cycle 24: “To find out exactly how people spend their time, we are going to ask about your activities over a 24-hour period. We will start at 4 in the morning because most people are asleep at that time. Let me give                                                    21 The American Time Use Survey Data Extract Builder (ATUS-X) is a user-friendly version of the ATUS, providing an easy step-by-step data extractor (Hofferth, Flood, and Sobek 2016). 49  you an example. Yesterday morning, I was asleep until 6. From 6 to 6:15 I got dressed while listening to the radio. From 6:15 until 6:25 I made breakfast. Then from 6:25 to 6:35 I ate breakfast with my spouse and son, while watching the news.22 You do not need to report activities of less than 5 minutes unless they involve travel or a change in the person you were with. Let’s begin. On [diary day], at 4:00 AM, what were you doing? …And then, what did you do?” (Social and Aboriginal Statistics Division 2009). Being the only data with detailed information on country-level daily time use with a representative sample, the time use surveys are the most apt measure for investigating participation in housework. There are a few advantages in using time diaries over other alternatives. Time diaries are less likely to reflect social desirability bias (Hofferth and Casper 2007; Kan et al. 2011) than surveys. Therefore, the present study has the potential to provide more accurate estimates to those proposed by the previous research (Brines 1994; Greenstein 2000), which were mainly based on self-reported measures of housework. Time diaries are considered to be more accurate than surveys because people tend to overestimate their contribution to housework when they are asked directly (Lee and Waite 2005; Marini and Shelton 1993; Robinson 1985).  3.1.1 General Social Survey This study compares time spent on household tasks by gender using the microfiles of the Canada-wide General Social Survey (GSS), cycles 2 (1986), 7 (1992), 12 (1998), 19 (2005), and 24 (2010) accessible through the Canadian Research Data Centre Network. The GSS research team interviewed non-institutional residents of Canada who were at least 15 years of age in 10                                                    22 The examples used can vary. 50  provinces and territories via telephone, using the RDD (random digit dialing) sampling technique. The response rates for the GSS time diaries are between 55.2% and 78.9%.23 The level for non-response is common for time use diaries where, generally, 60% of non-response is due to noncontact and 40% due to refusal (Abraham, Maitland, and Bianchi 2006). Differences between the sample and the population were observed with younger individuals and men being underrepresented in the unweighted sample (Statistics Canada 2011). Therefore, the final survey weights which I used included the non-response adjustment. Statistics Canada performed non-response adjustment based on two groups of non-responding telephone numbers: those with some information available (for example, where the household information was already collected prior to non-response) and those with no information. The non-response was handled by adjusting weights using the population distributions from external reference (Statistics Canada 2011) to account for those who did not respond. 3.1.2 American Time Use Survey For the American Time Use Survey (ATUS), 2003-2015 (Bureau of Labor Statistics 2015; Hofferth et al. 2015), the Census Bureau interviewed residents of the US who were at least 15 years of age in 50 states and one federal district. The sampling technique used is a stratified three-stage sample. The Current Population Survey (CPS) becomes the sampling frame for the ATUS and the selected members of households are interviewed 2-5 months after the completion of the CPS. The response rate for the ATUS time diaries are between 57.8 and 49.9%. The ATUS team adjusts for the nonresponse in the final weights which are used in the present study as well. The adjustment, however, only results in modest differences and the “set of estimates [with or without adjustment for nonresponse] are broadly similar” (Abraham et al. 2006). These                                                    23 Response rates were 78.9% (1986), 77% (1992), 59% (1998), 58.6% (2005), and 55.2% (2010).  51  differences are observed not with busier people but with people who have less connections to the community, mostly because it is difficult to reach them (Abraham et al. 2006).  3.1.3 Sample Selection I used several sample selection steps to produce the final analytic sample. I restrict my main sample to married or cohabiting individuals. Thus, in the GSS sample, 71% of men and 69% of women respondents report being married or living in common law. I chose not to include cohabitation as a separate marital status because two cycles of the GSS (out of five) did not include it as a distinct category. Cycles 2(1986) and 7(1992) grouped cohabitation with marriage. The results for the analyses and trends using the three-cycle sample of cohabiting individuals were indistinguishable in outcomes from those reported in this project for the full sample (see Appendix E Figure E1 for trends comparison). Similarly, in the ATUS, even though the cohabitation variable is available for the full sample, the results of analyses did not provide new information and were similar to those for the full sample. Additionally, the one-person households were dropped out of the analysis because the division of housework is only conceivable in households with both partners present in the house.  The final total sample of weighted data for the Canadian GSS includes 20,243.5 men’s person-days and 19,600.6 women’s person-days spanning the period of 1986-2010 (see Table 2) and 49,857.7 men’s person-days and 49,574.9 women’s person-days for the ATUS spanning the period of 2003-2015 (see Table 3). Personal weights were re-coded based on the original survey weights and scaled to the original sample size before the sub-setting.  For the estimation of variance in the Canadian sample, I used bootstrap weights, available for all cycles only with access to RDC. To avoid limitations associated with biases of the data 52  collection and data variation depending on the survey year due to year-specific characteristics and on the province due to significant differences in non-response across provinces24 (Statistics Canada 2011), I employ year and province 25 fixed-effects models with Heckman correction (Heckman 1976, 1979; Puhani 2000). 3.1.4 Specifics of Time Use Diaries In time use diaries, respondents reconstruct their previous day in a diary as a sequence of episodes of various activities starting from 4 am in the morning. The time of start is dictated by the context-driven point of the least variability. Thus, in North America, 4 am is the time of the least variability where most people are asleep. A number of studies were conducted to test the validity and reliability of time use diaries, most of them report that these are more reliable than traditional surveys  (Kan 2008; Kan and Pudney 2008; Lee and Waite 2005; Robinson 1985; Robinson and Bostrom 1994). Time use diaries usually collect data only for a single member of a household, which is one of the limitations of time use diaries (Lee and Waite 2005). Another limitation of time use diaries is that they rarely are able to capture multitasking, which might cause underestimation of time spent on various domestic tasks (Lee and Waite 2005). GSS 2010 collected information for secondary activities which captures multitasking, however, earlier surveys did not. However, the percentage of respondents reporting housework as secondary activity is rather small, 1.8% of all reported episodes (for comparison, 11.54% of all episodes reported ‘talking, conversation’ as the secondary activity). Moreover, Sullivan and Gershuny (2013) report that time spent multitasking on domestic work is not as much to influence the results obtained through analysis of primary activities.                                                    24 Although the weights adjust for non-response, they are rarely precise because there is always room for uncertainty in population distributions used for such adjustment. 25 Because of the number of states, I do not control for them. 53  3.1.5 Collection and Limitations of Time Use Diaries Due to the specific character of the time use diaries and their many advantages, it is necessary to keep in mind a few limitations. Even though more detailed information is collected on activities of individuals during a diary day, it is always just a snapshot of the range of their activities in general. These activities may only reflect the diary day and be not an accurate representation of their actual activities in other days, i.e. the diary recorded activities do not mean that on all other days, respondents perform the same types of activities as on the diary day. For instance, a respondent who did not report any housework during the diary day may actually perform a lot of housework on other days, if the diary day was on a weekday as opposed to being a weekend day. Second, completing time use diaries is very time consuming, which puts the quality of collected time use diaries into question. For instance, in the Cycle 24 of GSS, 9.6% of all diaries report 10 or less activities per day. This high percentage of low activity reports reflects the perception of how cumbersome it is to report all activities for time diaries. Therefore, time use diaries are at best only approximations of long-term time use of individuals.  3.1.6 Heckman Two-Step Correction for Selection Bias  I use Heckman’s adjustment to my model estimators to reduce the selection bias, which can result from the possible non-random character of the differences between the selected married and cohabiting sample compared to the non-married and non-cohabiting individuals that were not selected into the analysis and also those who did not report doing any housework. The Heckman selection corrects for not having a random sample. The correction includes two stages, first, the selection model (Equation 1.1.) estimates the probability of being “chosen” into the substantive model. Then the substantive model (Equation 1.2.)  estimators are adjusted using the 54  correction factor, the inverse Mills ratio, calculated from the selection model (Bushway, Johnson, and Slocum 2007). The selection equation is estimated with a probit, where 𝑌2  is dichotomous, Z – is the design matrix for the selection model, α – parameter vector, and δ – the error term of the selection model. In the substantive model (1.2), β is the parameter vector, X- the design matrix, 𝜌𝜀𝛿  – correlation between the error terms of the selection and substantive models, 𝜎𝜀𝜌𝜀𝛿 – the unknown parameter, and 𝜆(−𝛼𝑍) – the inverse Mill’s ratio. 𝑌2 =  𝛼𝑍 +  𝛿                                                                 (1.1)  𝑌1 =  𝛽𝑋 + 𝜎𝜀𝜌𝜀𝛿𝜆(−𝛼𝑍)                                                    (1.2)                                       The method allows for treating the error term in the substantive equation given the selection as an omitted variables problem to solve the problem of sample selection bias and related issues connected with model specification. The Heckman two-step estimator is more robust than the full information maximum likelihood estimator and can converge even in the absence of the exclusion restrictions, though, the presence of ‘valid’ exclusion restrictions is highly recommended (Bushway et al. 2007). By exclusion restrictions, researchers mean variables that are used like instrumental variables, which affect the selection process and not the substantive model (Bushway et al. 2007). These instrumental variables help to deal with endogeneity, specifically with the problem of omitted variables that affect the error term to be correlated with explanatory variables. Although the ‘exclusion restrictions’ (which are sometimes called ‘instruments’) are not necessary (Bascle 2008), it helps if such can be identified. Often, however, it is very difficult to find strong instruments that affect the selection but not the dependent variable in the substantive 55  model. I tested a few ideas that could work as potential ‘exclusion restrictions’, including time use data specific variables such as whether the diary was collected on the weekend (in the selection model but not in the substantive model) and the number of episodes. However, these variations for the selection model did not result in significant difference from the results that are presented here. Using other measures for resources, while having only the income transfer variable in the substantive model, is justified because women and men that are not married/cohabiting and not doing housework might be different from those who are and do housework in terms how much resources they have.  3.1.7 Comparing Canadian and American Samples  Tables 1 and 2 summarize the descriptive statistics for the main dependent variables of the two samples from the GSS and ATUS. We see many similarities between the two. Thus, in both Canada and the US, women spend almost twice as much time on housework as men. The biggest absolute gender difference is in cleaning tasks in both samples. Thus Canadian women on average spend 50 minutes more on cleaning compared to Canadian men, whereas American women spend 45.3 minutes more than American men. The gender gap in cooking is second to cleaning, and Canadian men spend 45.5 minutes less on cooking than Canadian women, and American men – 38.9 minutes. The smallest absolute difference between women and men is in shopping tasks. Thus Canadian women spend 13.9 minutes more on shopping on an average day, whereas American women – 17 minutes more than American men. If the absolute difference in routine housework can be considered as one of the measures of the equalitarian character of the gendered division of time allotted to a task then the shopping tasks are the most equalitarian, 56  which coincides with the idea that shopping is viewed more and more as gender-neutral activity (Baxter 2002; Craig et al. 2015). Table 2 Descriptive Statistics for the Main Variables for the Canadian GSS Variables Description Mean (Women) (N= 19600.6) Bootstrap SE Mean (Men) (N= 20243.5) Bootstrap SE Diff. in Means Dependent Variables       Domestic Tasks Domestic tasks combined (min.) 206.02 (0.348) 112.79 (0.280) 93.23***  Cooking Tasks Cooking and washing up (min.) 72.320 (0.146) 26.830 (0.082) 45.49***  Cleaning Tasks Cleaning, laundry, and mending (min.) 70.373 (0.195) 20.384 (0.112) 49.989***  Maintenance Tasks Maintenance and repair (min.) 6.045 (0.074) 22.214 (0.148) -16.169***  Shopping Tasks Shopping and services (min.) 57.281 (0.211) 43.362 (0.178) 13.919*** * p < .05, ** p < .01, *** p < .001 All results obtained using survey weights.  Table 3 Descriptive Statistics for the ATUS, 2003-2015 Variables Description Mean (Women) N=49574.9 SE Mean (Men) N=49857.7 SE Diff. in Means Dependent Variables         Domestic Tasks Domestic tasks combined (min.) 187.539 (0.909) 96.632 (0.773) 90.907***   Cooking Tasks Cooking (min.) 58.627 (0.386) 19.775 (0.241) 38.852***   Cleaning Tasks Cleaning, laundry, and mending (min.) 69.295 (0.605) 24.031 (0.412) 45.264***   Maintenance Tasks Maintenance and repair (min.) 2.272 (0.123) 12.842 (0.393) -10.57***   Shopping Tasks Shopping (min.) 57.345 (0.509) 39.985 (0.442) 17.360*** * p < 0.05, ** p < 0.01, *** p < 0.001. All results obtained using survey weights.  57   Figure 5 Time Spent on Cooking, Canada and US Figures 5-9 show the trends of the average time that married and cohabiting women and men spend on a diary day in Canada and in the US for housework tasks such as cooking, cleaning, shopping and maintenance. The last Figure 9 summarizes the trends between women and men for all time spent on housework. In 1986, Canadian women spent 91.6 minutes on cooking during an average day; by 2010, the time women spent decreased to 66.4 minutes. On the other hand, men increased their time on cooking from 16.7 minutes in 1986 to 30.5 minutes in 2010. The overall story, however, is the one of the gender gap stagnation. The trend in cooking (Figure 5) shows that the gender gap has stagnated in between 2005 and 2010 in Canada, and between 2003 and 2015 in the US. Similar patterns are evident in other domestic tasks (see Figures 6-8). 91.686.177.865.1 66.416.722.127.9 28.4 30.560.0 57.7 55.9 57.3 57.6 54.959.6 58.7 59.3 56.9 58.362.5 63.016.8 17.1 17.0 18.9 17.9 19.7 19.321.5 19.9 19.422.5 23.0 23.20.010.020.030.040.050.060.070.080.090.0100.01986 1992 1998 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Time Spent on Cooking, Canada and US (in minutes)Women (Canada) Men (Canada) Women (US) Men (US)58   Figure 6 Time Spent on Cleaning, Canada and US   Figure 7 Time Spent on Shopping, Canada and US 78.4 76.871.8 72.565.920.213.122.2 22.3 22.974.3 72.576.973.1 74.264.672.066.5 64.767.7 65.7 63.7 65.524.3 26.1 21.4 22.527.322.5 24.7 24.6 24.9 24.0 23.5 23.9 23.00.010.020.030.040.050.060.070.080.090.01986 1992 1998 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Time Spent on Cleaning, Canada and USWomen (Canada) Men (Canada) Women (US) Men (US)65.654.4 55.9 55.4 56.747.941.3 43.1 40.145.659.0 61.062.9 61.6 59.4 59.155.4 56.852.2 54.0 54.355.2 55.041.9 40.4 41.6 41.6 39.3 39.1 39.3 39.842.740.0 40.0 37.6 36.70.010.020.030.040.050.060.070.01986 1992 1998 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Time Spent on Shopping, Canada and USWomen (Canada) Men (Canada) Women (US) Men (US)59   Figure 8 Time Spent on Maintenance, Canada and US Figure 9 summarizes the trends in average time spent on all domestic tasks. The trends suggest that there is little discrepancy between the time spent on housework in Canada and in the US. The  resulting similarity may work as an indication of cultural affinity between two neighbouring nations (Esping-Andersen 1990).   Figure 9 Time Spent on Domestic Tasks, Canada and US 4.25.6 5.87.05.519.025.219.422.920.12.6 2.7 2.0 2.7 2.2 2.2 2.1 1.93.0 2.1 1.9 2.4 1.815.3 14.416.212.911.2 12.1 11.012.513.810.3 10.313.7 13.30.05.010.015.020.025.030.01986 1992 1998 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Time Spent on Maintenance, Canada and USWomen (Canada) Men (Canada) Women (US) Men (US)239.8222.9211.3200.0 194.6103.8101.6112.7 113.7 119.0195.9 197.7 193.4 189.1 179.1 180.2 185.498.4 96.2 95.7 94.3 101.2 96.4 98.2 96.30.050.0100.0150.0200.0250.0300.01986 1992 1998 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Time Spent on Domestic Tasks, Canada and USWomen (Canada) Men (Canada) Women (US) Men (US)60   Table 4 Independent and Control Variables, Canadian GSS Variables Description Mean (Women) (N= 19600.6) Bootstrap SE Mean (Men) (N= 20243.5) Bootstrap SE Diff. in Means Independent Variable        Income transfer (Δ) (-1,1) -1 – dependency, 1 – providership -0.226 (0.001) 0.421 (0.001) -0.647*** Control Variables         Paid Work Time spent in paid work, min. 165.683 (0.455) 261.303 (0.506) -95.62***  Personal Income Annual income, CAD 25492.91 (50.253) 51519.35 (324.98) -26026.4***          Full Time Reporting working full-time 43%  68%  -25%   Part Time Reporting working part-time 13%  4%  9%   Student Reporting student status 1.8%  1.1%  0.7%   Other Reporting having other than full time or part time occupation 43%  27%  16%   Own home Owns the dwelling or rents 81%  82%  -1%  Born in Canada 1=born in Canada; 0=otherwise 79%  78%  1%   Age Age of respondent in years 43.861 (0.021) 46.967 (0.020) -3.106***   Education (years) Education of respondent in years 13.533 (0.005) 13.479 (0.006) 0.054**   Children 1=respondent as children; 0=otherwise 53%  54%  -1%   Under 5 1=children under 5; 0=otherwise 18%  17%  1%   Household Size Number of people 3.166 (0.002) 3.142 (0.002) 0.024*  Leisure Time spent in leisure, min. 262.066 (0.328) 282.840 (0.312) -20.774***  Weekday 1=weekday; 0=Saturday or Sunday 72%  72%  0% English 1=English, 0=otherwise 55%  55%  0% 61  Variables Description Mean (Women) (N= 19600.6) Bootstrap SE Mean (Men) (N= 20243.5) Bootstrap SE Diff. in Means French 1=French, 0=otherwise 28%  28%  0% Chinese 1=Chinese, 0=otherwise 2.6%  2.4%  0.2% South Asian 1=South Asian, 0=otherwise 1.9%  2.3%  -0.4% Filipino 1=Filipino, 0=otherwise 0.9%  0.8%  0.1% * p < .05, ** p < .01, *** p < .001 (p values are for the adjusted Wald test and χ2 test for proportions). All results are weighted. . Table 5 Independent and Control Variables, ATUS Variables Description Mean (Women) N=49574.9 SE Mean (Men) N=49857.7 SE Diff. in Means Independent variables        Income transfer (-1,1) -1 – dependence, 1 – providership -0.434 (0.004) -.064 (0.005) -0.37*** Control variables                Paid Work Min spent on paid work on the diary day 171.552 (1.509) 268.177 (1.806) -96.625***   Personal Income In US dollars, per annum 21138.54 (171.116) 36414.45 (241.023) -15275.9***   Full Time Reporting working full-time 43%  68%  -25%***   Part Time Reporting working part-time 17%  7%  10%***   Other Reporting having other than full time or part time occupation 40%  25%  15%***   Own home Owns the dwelling or rents 82%  82%  0%   Born in the US 1=born in Canada; 0=otherwise 84%  84%  0%   Age Age of respondent in years 46.944 (0.096) 48.913 (0.102) -1.969***   Education (years) Education of respondent in years 13.851 (0.018) 13.801 (0.021) 0.05**   Children 1=respondent as children; 0=otherwise 46%  45%  1%   Under 5 1=children under 5; 0=otherwise 22%  22%  0%   Household Size Number of people 3.188 (0.009) 3.206 (0.010) -0.018* 62  Variables Description Mean (Women) N=49574.9 SE Mean (Men) N=49857.7 SE Diff. in Means   Weekday 1=weekday; 0=Saturday or Sunday 72%  72%  0%   White only 1=White, 0=otherwise 88%  87%  1%    Black only 1=Black, 0=otherwise 7%  8%  -1%   Asian only 1=Asian, 0=otherwise 4.4%  3.5%  0.9%   Native American only 1=Native American, 0=otherwise 0.7%  0.7%  0%   Hispanic 1=Hispanic, 0=Otherwise 13.1%  13.7%  -0.6% * p < 0.05, ** p < 0.01, *** p < 0.001 (adjusted Wald test and χ2 test for proportions)  3.2 Measures The following sections discuss the main measures used for the analysis of the participation in housework and the gender gap in the division of unpaid labour.  3.2.1 Dependent Variables The dependent variables are represented both by an aggregate measure of the time spent on all four domestic tasks combined and by each task individually: cooking, cleaning, maintenance, and shopping. Analyzing housework by separate tasks allows me to measure housework in two ways: by tasks and by time in minutes allotted to housework (Blair and Lichter 1991). In the GSS sample, cooking tasks are represented by meal preparation, baking, preserving food, home brewing, food (or meal) cleanup, and other related activities. Indoor cleaning, outdoor cleaning (garbage, snow removal, garage), laundry, ironing, folding, mending clothes/shoe care, dressmaking, and sewing (for self or household member) comprise the ‘cleaning category of the domestic tasks. Maintenance tasks include interior maintenance and repair, exterior maintenance 63  and repair of home, vehicle maintenance, and other home improvements. The shopping category covers a wide range of activities including everyday shopping, personal care services, government and financial services, adult medical care and dental care, other professional services, and travel to shopping and obtaining services. These tasks although very broad conceptually were chosen using the category already created by the Statistics Canada for shopping activities. The ATUS dependent variables were coded in the way to ensure similarity with the categories in the GSS. The full coding scheme can be found in Appendix A both for the GSS and the ATUS datasets.  Table 2 presents descriptive statistics for main domestic tasks. There is a significant difference between time spent by women and that by men for all tasks (adjusted Wald test returns significant p-values for all four tasks and their aggregation). On an average day, women spend 71.5 minutes on cooking, whereas men spend only 25. For the trend analysis and decomposition models zero values were treated as natural zeros, but for the regression models, all zero values for dependent variables were coded as missing. The choice is dictated by the nature of time use data because unlike traditional surveys time diaries focus only on one day and not on a more extended period of time which could allow a wider spread of non-zero values for housework measures. The Heckman models also employ the natural logarithms of the time variables to normalize the distribution of each. Alternative ways of measuring the value of housework participation is sometimes employed by macroeconomists. These measures include, for instance, imputing economic value of the housework produced (Colman 1998; Landefeld, Fraumeni, and Vojtech 2009). The measures achieved by imputing values, however, are problematic because they are highly imprecise and they ignore the institutional character of marriage and that goods 64  and services produced by housework are not that easily replaceable in the market. Moreover, these imputations must be, first and foremost, based on the estimates of time, captured by the time use surveys. My full coding references can be found in Appendix A. 3.2.2 Independent Variables Tables 3 and 4 summarize the descriptive statistics for the independent and control variables discussed in this section. The income transfer variable helps to test the relative resources argument and is adopted from the work of Brines (1994), who bases it on Sorensen and Mclanahan (1987), and is defined as (personal income – partner’s income)/ (personal income + partner’s income). The measure ranges between -1 and 1, where -1 is the highest level of dependency and +1 – the highest level of breadwinning.  The autonomy approach (as opposed to the relative resources) (Gupta 2007) is tested by personal income in CAD and home ownership (1= ‘owns home,’ 0= ‘otherwise’). An average Canadian man makes CAD 51,519.35 annually whereas an average Canadian woman earns CAD 25,492.91 in the GSS samples. Similarly, an average American man earns USD 36,414.45 and an average American woman – USD 21,138.54 per annum in the ATUS sample. Eighty-two percent of Canadians and of Americans report owning the home that they live in. There are four dummy variables indicating the employment status of the respondent as reported for the descriptive statistics in Table 4: full-time workers, part-time workers, students, and individuals with other employment status (retired, unemployed, individuals on parental leave, housewives and househusbands). I assume that the level of autonomy decreases in the order the employment status variables are presented above, corresponding to the average levels of personal income within each group - the average income of married and cohabiting full-time workers was the 65  highest and amounted to CAD 45,366, whereas the average income of married and cohabiting students was the lowest, CAD 15,210. The dollars in 1986, of course, are not the same as dollars in 2010. Although I do not adjust for inflation, some of the differences are absorbed by controlling for year fixed effects in the Heckman models.  To ensure comparability across cultural groups, I combined employment status variables into a single dummy variable distinguishing between working (full-time and part-time) and non-working individuals. The ‘0’ value for ‘Working’ variable represents students and individuals with ‘other’ employment status, thus the least autonomous individuals, while ‘1’ in the ‘Working’ variable represents the most autonomous men and women. As to the similarities and differences between Canadian and American samples, among both Canadians and Americans, 68% of men and 43% of women report being employed full-time. Thirteen percent of Canadian women and 17% of American women report part-time employment status, whereas 4% of Canadian men and 7% of American men do the same. The category ‘Student’ is not present in the ATUS sample and the category ‘Other’ is used as a reference which includes students as well.  The time availability argument is tested using the time spent on paid work. Paid work represents a time constraint which is imposed by the person’s involvement in an economic activity. An average Canadian man works approximately 261 minutes (around 4 hours) a day, whereas an average American man works 7 minutes more. Similarly, an average Canadian woman works 6 minutes less than an average American woman, 166 (2h46m) and 172 (2h52m) minutes respectively. Paid work represented by time spent on the job and unpaid work represented by time spent on housework, are important dimensions of everyday life. However, another dimension of daily life is leisure. The combined effect of leisure time variables 66  represents another constraint on the available time for the respondents. This experimental variable is, however, created only for the Canadian sample to simplify the analysis. An average Canadian men spends 283 minutes (4h43m) a day on combined leisure activities, whereas a Canadian woman spends, on average, 21 minutes less on leisure activities, 262 (4h22m) minutes. I did not create leisure variables for the ATUS sample because it would result in a number of discrepancies rendering the comparison between the samples close to impossible. Additionally, another variable related to this framework is the presence of children (Wight, Bianchi, and Hunt 2013), especially of children of 5 and above years of age because older children can help with housework. Fifty-three percent of Canadians and 45% of Americans report having children. Moreover, among Canadians, 18% report having children under 5 years of age, whereas 22% among Americans do the same. Because respondents can delegate the tasks to other adults within a household, I also test the time availability with the number of people in the household capped at 8 people for the GSS sample. Yet the average number of people in a household is 3 people both among Americans and Canadians, suggesting a traditional nuclear family with a single child.  As for the gender-centred argument, I take into consideration that gender attitudes change with time and generation (Baxter and Kane 1995; Marshall 2011). I use education and age in years trying to capture the age effects in gendered performances and those associated with differences in socialization as a result of education. 26  The average years of education for Canadians is 13.5 and for Americans is 13.8. Age is recorded in years. The average Canadian was younger than the average American. The average age for Canadian men was 47 as opposed                                                    26 The age variable, however, does not allow us to capture cohort effects. A test of period-cohort effects, which are outside of the scope of this discussion, can be an interesting endeavour for future research. 67  to 49 years of age for American men, the average age for Canadian women was 44 whereas the average age for American woman in the sample was 47. I also use selected lifestyle variables, labeling them as more ‘masculine’ if men were significantly more likely to be involved in the activity than women (after all, what is gender but its performance (Butler 1990)) and as more ‘feminine’ if women, on average, were significantly more likely to do the activity, using t-test results. Thus, I created two groups of leisure activities for the GSS sample as proxies for traditional gender attitudes: (arbitrarily) feminine and masculine, 27  and use them to analyze whether the involvement in such activities is able to provide an explanation for the gender gap in the selected domestic tasks.  All lifestyle variables are measured as time in minutes spent on an activity per diary day.  Other ways in which I intend to investigate more direct influences of gender than the economic frameworks allow are (1) by analyzing tasks individually, and (2) by different cultural and racial groups. 3.2.3 Control and Selection Variables Because there are usually considerable differences in diaries depending on the day of the week the diary was collected on, I also control for whether the diary day was completed for a weekday (1=weekday, 0=Saturday or Sunday). Time use diaries usually adjust in the weighting process for the weekday, thus resulting in equal distribution across the days of the week. In the analytical sample for the present study, a small yet negligible difference is produced, due to the employed                                                    27 ‘Masculine’ activities include:  engagement in professional, political organizations, attending fraternities and sororities, sports events, concerts, museums, clubs, doing sports, camping, performing, sightseeing, listening to the radio, watching TV, listening to CDs and music. ‘Feminine’ activities include: engagement in family-oriented,  Volunteer, religious organizations, studying for leisure, attending classical music concerts, spending time with friends, doing crafts, playing board games, and reading. 68  sub-setting and sampling procedures. Overall, 72% of the diaries were collected on weekdays and 28% on Saturdays and Sundays, indicating that survey weights are applied to daily distributions. This weighting is applied both to the GSS and the ATUS samples.  In the OLS models in GSS sample, I control for the province of residence because there is an expectation of province-level variation that was not accounted for by weighting procedures (Statistics Canada 2011). I do not control for state of residence in the OLS models for the US sample because including the 51 dummy variables for states partitions the sample into groups, some of which contain only a few observations. I also control for a few provincial-level variables in the analysis of the GSS data to account for structural-level differences. Thus I have added the provincial unemployment rate by year reported by the Economics and Statistics Branch  (Statistics Canada 2016a), marriage rates per 1000 people by province and by year (Statistics Canada 2015), and the percentage of women employed by province and by year (Ferrao 2010).  In the OLS models for the US sample, on top of the small sample issues, partitioning the sample of women or men with state dummies does not allow the estimation with the Heckman adjustment to converge in the models predicting time allocation to maintenance tasks. In other tasks, the coefficients for the relative economic contribution variables in the models with state dummies are similar to those reported here. Instead, I have merged the data on state unemployment rates from 2003 to 2015 from the online archives of the Bureau of Labour Statistics (Bureau of Labor Statistics 2016) and on marriage rates per 1000 people by state from the National Center for Health Statistics (National Center for Health Statistics 2016). I have calculated the percentage of women employed by state and by year using CPS datasets for the month of October for each year from 2003 to 2015 with the help of NBER’s Jean Roth’s CPS 69  Stata files (Roth 2016). These allowed me to have continuous measures controlling for the state-level institutional factors. Although cohabiting was shown to be a significant factor in defining how much housework women perform in the previous research as on individual level (Shelton and John 1993) and on national (Batalova and Cohen 2002), I have run the same analyses presented in the following chapters on the cohabiting sample of 1998-2010 GSS and found that the results are similar as for the full sample, deciding not to include cohabitation as a control variable for models. I have included the trends for 1998-2010 comparing the cohabiting subsample with the married subsample in Appendix E. The graph (Appendix Figure E1) shows that the time allocated to housework by cohabiting women and men is similar to that of married women and men. Year variables are dummy variables for each cycle of GSS and ATUS. Although there is little variation across years in the ATUS sample, I chose to control for years to preserve consistency with the GSS models. Weights are recoded to represent the original sample size from the GSS and the ATUS final weights. The details on the original GSS and the ATUS variables for each of the measures are presented in the Appendix A. 3.3 Models The decomposition models include testing the relative resources, time availability, autonomy and gender-centred frameworks separately. The full model is also used to analyze how much of the gender gap in select housework tasks can be explained by the factors introduced in all models. The decomposition models are done first to test frameworks in their ability to explain the gender gap in the domestic division of labour. These models are analyzed for the total sample and by main cultural groups in Canada: Anglo-, French, Chinese, South Asian, and Filipino 70  Canadians and by main racial groups in the US: those who report being White only, Black only, Asian only, and those who report being Native American. It is important to note, however, that sample sizes for cultural and racial groups are small. Moreover, the data deficiencies connected with the collection and recording of time use diaries, especially in the early years of the GSS, are a reminder to be cautious about generalizations regarding the cultural groups. However, with the use of personal weights for point estimates and bootstrap weights for variance estimation, some degree of certainty can be achieved.    The substantive models for the second stage of the Heckman procedure of testing the association between relative resources and gender-centred approaches to the time spent on domestic tasks among women and among men (the variation within gender groups) include the following variables: age, whether the respondent was born in Canada/the US, education, whether the respondent has children, whether the respondent has children under 5 years of age, household size, whether the diary day is a weekday, the measure for relative income and its power elements, cultural and racial group identification, the factorized year variables and province variables. The models also include the province-level control variables. The Heckman selection models included all independent variables and variables that might have influenced the selection, especially other measures for economic resources omitted in the main model to draw focus on one variable representing the bargaining such as personal income, partners’ income, time spent on paid work, whether the respondent is in the labour force, and home ownership.  This specification allows taking other variables measuring economic resources into account while not letting them confound the result for the main models. These other economic perspective testing 71  variables that were added to the selection models were also analyzed as controls in the substantive model but the results were similar with those reported in this project.28  3.3.1 Comparing Gaps: Decomposition Method The Blinder-Oaxaca decomposition method is used widely in the analysis of the separate contributions of characteristics to resulting wage differences (Chang and England 2011). The decomposition method allows an analyst to take the gender gap and allocate a portion of this to differences in education, experience, industry differences, or other factors between women and men. To my knowledge, it has not yet been employed extensively to the analysis of the gender gap in housework participation. With the help of the decomposition method, I am able to decompose the gender gap in housework into allocations based on resources, education, or other gender socialization factors. This technique provides a powerful tool to identity and quantify the contributions of various factors from competing frameworks to the gap between women’s and men’s contribution to domestic work. The decomposition procedure involves running separate regressions for each group, gender groups in the case of the present project, analyzing the contribution of factors to the groups’ time spent on housework, or any other variable of interest. The Blinder-Oaxaca pooled method includes the differentiation between the explained part of the gap and the unexplained part of the gap. The explained part represents the portion of the gap that can be accounted for by the factors presented in a tested model. The unexplained portion is                                                    28 The only one result that was different is for the association between relative resources and the time spent on shopping among women in models, which controlled for the time spent on paid work. The relative resources of women were found to be positively related to the time spent on shopping, that is, women increased their time in shopping with the increase of their economic contribution, when controlled for the time allotted to paid work. This pattern is characteristic of gender neutralization (Section 2.3.3.) and suggests that women are under pressure to increase their time spent on shopping proportionate to the increase of their economic contribution. 72  usually ascribed to discrimination or other unexplained differences. The definitions for explained and unexplained decomposition can be represented in the following way:  R = [E(X1) − E(X2)] ′β∗ + [E(X1) ′ (β1 − β∗) + E(X2) ′ (β∗ − β2)]  29                    (1)  Explained                                   Unexplained Where R is the mean difference between two groups (for examples, women and men, X1 design matrix for the first group, X2 design matrix for the second group, β1 slope parameters for the first group, β2 slope parameters for the second group, and β* is the non-discriminatory coefficients vector (Oaxaca and Ransom 1994; Jann 2008).30 The non-discriminatory coefficients are calculated based on which variant of the decomposition method is used. There are a few variations of the Blinder-Oaxaca decomposition method. The earlier developments of the method used the coefficients of one of the groups as reference, assuming that that group’s coefficients are non-discriminatory. Thus, for example, the actual gender gap would be compared to the gap that could be if women’s model coefficients were the same as men’s (Chang and England 2011). However, because very often there is no reason to assume that any group’s coefficients are non-discriminatory, the method improvements included some pooled types of calculation of the non-discriminatory coefficients (Jann 2008). Initially, the pooled decomposition employed more straight-forward ways of pooling, such as assigning equal weights to the coefficients (β* = WβA + (I-W)βB  where W=0.5) (Reimers 1983) or calculating them based on the number of observations in each group (Jann 2008). Later, the method included more complicated versions such as in                                                    29 Expected error term for the outcome variable of both groups (Y1 and Y2) is zero by the OLS assumptions. Thus expected errors for both regression models are E(ε1) = 0 and E(ε2) = 0. 30 β* = WβA + (I-W)βB; where W – relative weights given to coefficients of group A (Oaxaca and Ransom 1999). Ŵ=Ω=(XA’XA+XB’XB)-1XA’XA. This paper, therefore, uses a variant of the pooled method of constructing the non-discriminatory wage structures as proposed by Oaxaca and Ransom (1994), using a pooled model over both groups as the reference coefficients with including group variable (here: Female) as a control variable in the pooled model. Figure 10 represents such a situation where W=0.5. 73  Neumark (1988), Oaxaca and Ransom (1994), and as explained in Jann (2008). The decomposition method variant used in this project is the latest development available for Stata (Jann 2008). To analyze how much of the gender gap in the time spent on housework is produced by the differences in the men’s and women’s attributes and skills, the mean difference is multiplied by the slope from the non-discriminatory slope β*, since the slope tells us the rate of return to a unit change in the variable (see Figure 10). This produces the amount of the dependent variable which we can explain with the difference in the independent variable in focus. In other words, the explained part of the gap is ascribed to the differences in the factors measured by the independent variables. The unexplained part therefore, is the difference from the explained part to the highest point in the interval of the slope above and the difference to the lowest point in the interval from the slope below.31 This part is usually attributed to the effects of discrimination and unobserved variables (Jann 2008).                                                     31 For more detail, please see Elder, Goddeeris, and Haider (2010). 74   Figure 10 Decomposition Method To convert either of these two quantities to the percent of housework time explained by the mean difference in the variable, we take the product (mean difference times the slope) as the numerator in a ratio where the mean difference between the two sexes in housework time is the denominator. The resulting ratio (times 100) tells us what percent of the gap is explained by the sex difference on this factor. In the case of a series of dummy variables representing a concept, such as being in the labour force, the portions explained by mean differences on all the dummies in the set are summed.  The example above illustrates a simpler version of the decomposition method where average coefficients over both groups are used as the non-discriminatory coefficient β* (Reimers 1983). The analysis in this study uses a little more complex and less-biased version (Lee 2014) of calculating the ‘non-discriminatory’ coefficients called the pooled Oaxaca-Blinder decomposition method (Jann 2008). It also computes the two-fold decomposition with explained and unexplained parts as shown in (1) but the ‘non-discriminatory’ coefficients β* are pooled. Explained Unexplained Unexplained 75  The two gender groups that are compared are also included in the pooled model as an additional control variable (Jann 2008).  The decomposition analysis helps analyze how much the gender disparities on each independent variable explain the gender gap in the time spent on housework. Thus I compute the proportion of the gender gap explained by differences in factor slopes (Oaxaca 1973), that is, how much is explained by mean differences in factors (sometimes called “endowments”) (Jones and Kelley 1984; Oaxaca and Ransom 1999). First, I look at the influences of the factors slopes on the gender gap without controlling for other competing explanations, thus variables are grouped by the corresponding theoretical framework: the relative resources, time availability, autonomy, and gender-centred approach. The autonomy argument is tested also with only the personal income variable (including the main control variables). This choice is dictated by the fact that other autonomy variables, namely, the employment status variable are the cause of high multicollinearity (vif>10) in all models in which they are present, and thus it urges to analyze the influence of personal income as the main empirical proxy for testing the autonomy approach (Gupta 2007) using models where the labor force variable is omitted. The analysis of the factors separately by their respective framework is done because many of the measures are correlated and looking at them separately helps single out their influences on the outcome variables, on top of avoiding general multicollinearity problems. This is especially relevant to most of the variables measuring the economic dimension. They prove to be redundant when entered together into a single regression because of the high association among each other. Second, I look at the all competing explanations combined to see how much can be explained in total by all theoretical arguments tested here. All analyses above are done on five subsamples separately: Anglo-, 76  French, Chinese, South Asian, and Filipino Canadians. To do so I use Stata’s ‘oaxaca’ command with bootstrap standard errors for estimates. The bootstrap weights that are not available in PUMFs for 1986-1998 are obtained from the GSS datasets at the UBC Research Data Centre and results presented herewith are released with the Statistics Canada permission.  3.3.2 Ordinary Least Squares Models For the OLS models with the Heckman correction, explained above, I use the relative resources measure of income transfer (Brines 1994) and its quadratic and cubic terms for consecutive models. In the analysis of time-dairy data, OLS generates more unbiased estimates compared to Tobit (Stewart 2013). Significance of linear and cubic models suggests that the economic exchange theory is apt for explaining the division of housework, while the significance of models with the quadratic term over the models with the linear and cubic terms suggests that the gender-centred framework provides a better explanation of the data. Robust standard errors are used for these models.  77  Chapter 4: Decomposing the Gender Gap Gender norms reflect culture. The recent studies in the US (Hwang 2015) and other countries (Kan and Laurie 2016; Ting et al. 2015) confirm that gender norms within a cultural group change as the cultural context shifts with life events. Canada as a multicultural nation has many separate and at times conflicting identities where the intersectionality of gender and culture divides us but also makes us diverse. But are we actually so different?  How much does the gender gap vary based on our cultural identities? Do Asian Canadian wives assume more housework than French Canadian women? Are there differences in factors and frameworks that explain the gender gap in housework among one cultural group of Canadians with those that explain the gender gap among other groups? The studies on differences based on cultural belonging with regard to the gender differences in housework are limited (Kan and Laurie 2016; Sayer and Fine 2011; Ting et al. 2015) and there is a need to fill the research gap contrasting such disparities by individual domestic tasks. The present chapter addresses this shortcoming and analyzes the gender gap by cultural groups in routine and non-routine housework. 4.1 Immigration to Canada: Historic Perspective Being one of the largest immigrant receiving countries of the world, Canada is a greenhouse for cultural hybridity (Bhabha 1994). Since first settlements in the early 17th century and through the turbulent 20th century with its massive immigration mostly from the European continent, Canada has welcomed millions of immigrants throughout its history. In 2014, Canada accepted 260,404 new permanent residents (Immigration, Refugees and Citizenship Canada, 2015). Its current immigration policy is focused on skill-selectivity to ensure the integration of immigrants 78  and for the broader nation-building strategy (Reitz 2013). The long history of colonization and immigration shaped the form of today’s multicultural Canada.  Statistics Canada measures ethnic origin by asking respondents to self-identify ancestral origin. Such a definition of ethnic origin suffers from a number of limitations, including that it relies on the idea of ethnicity as based on an ‘objective’ ancestry (Satzewich and Liodakis 2007). Yet many Canadians identify their ancestral origin as solely ‘Canadian’ and the number of such self-identifications increases with each census since the introduction of this response option in 1981 (Satzewich and Liodakis 2007). Furthermore, the Statistics Canada measure of ethnic origin does not reflect how much the respondents actually identify with the chosen ethnic origin, how much the ancestry means to them, or does not easily accommodate the idea of multiple ancestry (Satzewich and Liodakis 2007). According to the 2011 National Household Survey (Statistics Canada 2016b),  31% of all single ethnic origin responses identified themselves as the ‘Canadian’ ethnic. Among single ethnic origin responses, the biggest ethnic identity other than Canadian was ‘English’ – 6.9%, followed by ‘Chinese’ – 6.4%, and ‘French’ – 6.1%. The combined share of all single ethnic origin responses within the South Asian region also constituted around 6.9% of all responses. The second largest group among East and South East Asian ethnic origin responses was ‘Filipino’ – 2.7%. ‘Canadian’ and ‘British Isles’ origin comprise the ‘Anglo-Canadian’ cultural group within the present analysis and represents the dominant cultural framework within the Canadian context. How different are cultural groups in Canada with regard to the gender gap in housework? This chapter tests the approaches within the economic theory and the gender-centred perspective as 79  they apply to cultural groups of Canadians. The choice of the perspectives is dictated by the fact that most research on the gendered division of housework emphasize two types of mechanisms influencing time use in households: economic exchange and gender display (Bianchi et al. 2000; Bittman et al. 2003; Brines 1994; Coltrane 2000; Greenstein 2000; Hook 2010; Shelton and John 1996). Using the outlined theoretical categorization, the following sections utilize the pooled Blinder Oaxaca decomposition method to analyze differences and similarities in the gender gap among various cultural groups in Canada. 4.2 Results: Time Spent on Housework Tasks Trends Regardless of their ethnicity, women continue doing most of the housework in Canada. Yet the ratio of all women’s time spent on overall domestic chores to all men’s decreased from 2.31 in 1986 to 1.64 in 2010, suggesting a slow convergence and move toward gender equality in unpaid work in Canada. Figure 11 shows the time spent on total housework by women and men between 1986 and 2010 and confirms that among Canadians, the mean time spent on overall housework forms the pattern of the gradual narrowing of the gender gap. Women spent less time on housework in 2010 than they did in 1986, and men did more housework in 2010 than they used to in 1986. In comparison, American women do approximately twice (2.006 times) as much housework as American men, thus the more recent gender gap in the US is a bit wider than in Canada. Moreover, the gender gap in America appears to be stagnant for the period between 2003 and 2015 (see Figure 12). It is likely, however, that this pattern of stagnation in the beginning of the millennium is also present in Canada. Thus the share of women’s housework decreased from 1.76 to only 1.64 in the period between 2005 and 2010 in Canada, which is 7% decrease and is less than the 11% decrease in the US, from 2.23 to 1.99 in the same period. The 80  patterns revealed in Figures 5-9 in the previous chapter analogously suggest that the same stagnation is equally characteristic to Canada.  Figure 11 Time Spent on Housework by Canadians, 1986-2010   Figure 12 Time Spent on All Housework by Americans, 2003-2015 0501001502002503001986 1992 1998 2005 2010Domestic Tasks, All CanadiansWomen Men0.050.0100.0150.0200.0250.02003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015Domestic Tasks, AmericansWomen Men81  As to the gender differences by individual tasks and by cultural groups, this section analyzes the trends in time spent on main domestic tasks by type: routine indoor tasks such as cooking and cleaning, a routine outdoor task such as shopping, and a non-routine task such as maintenance; and by cultural group: people who identify English as their primary language, people who identify French as their primary language, Chinese, South Asians, and Filipino in Canada and a few racial groups in the US, such as White, Black, Native American, and Asian. English and French speakers are the biggest groups in the Canadian GSS data, where the weighted samples consist of 21514.5 English-speaking and 10860.77 French-speaking women and men. Both groups form the main gender convergence pattern in domestic tasks. In the period from 1986 to 2010, the most precipitous change toward more egalitarian division of labour happened in cooking tasks. Thus the ratio of women’s participation to men’s decreased more than twice among both Anglo- and French Canadian groups, from 5.44 in 1986 to 2.06 in 2010 among Anglo-Canadians and from 4.98 in 1986 to 1.96 in 2010 among French Canadians. Figure 13 represents the trends in time spent on cooking between men and women among Anglo- and French Canadians from 1986 to 2010. The slope is steeper for French Canadian women who used to spend, on average, 9 minutes more than Anglo-Canadians in 1986 but almost reach parity with them in 2010, spending approximately 64 minutes on cooking on an average day.  The pattern is not as straight-forward for other cultural groups, partially because of the data collection, weight-assignment issues, and smaller sample sizes. Figure 14 shows the patterns in time spent in all domestic tasks among Chinese, South Asian, and Filipino Canadians. The most reliable data points are those collected in 2005 and 2010 because these two surveys were the most similar both in terms of questions and collection procedures. In this period, both Chinese 82  women and Chinese men show patterns of participation in housework different from the other two groups’ trends. Thus both Chinese women and men decreased their time spent on domestic tasks in between 2005 and 2010. Most of the decrease among Chinese men comes from the decrease in the time spent on shopping, Chinese men spent 59 minutes on shopping on an average day in 2005 and 44 minutes in 2010. Therefore, Chinese women conform with the general trend among all women and reduce their involvement in the domestic tasks, whereas Chinese men resist the general trend and also reduce their participation in housework, unlike the majority represented by the Anglo- and French Canadians. On the other hand, there is also resistance to general pattern among South Asian and Filipino Canadian men and conformity with gender norms among South Asian and Filipino women (see Figure 14), who increased their participation in housework between 2005 and 2010. South Asian and Filipino men who also increased their participation in domestic chores in between 2005-2010, conform with the general trend among Anglo- and French Canadians and breaking the societal gender norms. The increase in participation is evident in all domestic tasks, particularly in cooking, cleaning, and shopping (because of the sample sizes and low report rates of maintenance tasks, it is difficult to state with enough degree of confidence whether the patterns are the same for maintenance tasks). 83   Figure 13 Time Spent on Cooking Trends Among English and French Canadians, 1986-2010   The inconsistent pattern revealed among smaller cultural groups is a reminder to take the generalizations about the participation of the cultural groups in housework cautiously: the sample sizes and the data deficiencies connected with the collection and recording of time use diaries, especially in the early years of its inception, does not allow us enough degree of certainty to make generalizations. Yet some of the trends are still discernible, and with at least the data of the most recent years, the estimates are comparatively more reliable. These trends diverging from the general trends are not present in the racial groups in the US. All - Whites, Blacks, Native Americans, and Asian Americans – show similar trends of the stagnation of gender convergence in housework for the period between 2003 and 2015, where women still do more housework than men. The gender gap between women’s and men’s participation in housework is the narrowest among Black Americans, thus the share of Black American women’s involvement in housework exceeds men’s only by 1.99 times (compare to the average of 2.10) for the period between 2003 01020304050607080901986 1992 1998 2005 2010Cooking Tasks, English GroupWomen Men0204060801001986 1992 1998 2005 2010Cooking Tasks, French GroupWomen Men84  and 2015. Overall, the largest gender gap in the US can be observed among Asian Americans rather than among Blacks, White, or Native Americans. These findings suggest that racial minorities in the US may be somewhat traditional as suggested by the research on minorities (Kan and Laurie 2016; Sayer and Fine 2011; Ting et al. 2015). Similarly, in Canada, the gender gap is wider among Chinese, South Asians, and Filipino Canadians, or overall among Asian Canadians, rather than among the majority English and French Canadians.     Figure 14 Time Spent on Domestic Tasks Among Chinese (Left), South Asian (Centre), and Filipino (Right) Canadians, 1986-2010 Consistent among all Canadian groups, the gender gap is the narrowest among routine outdoor housework. The same is true about American racial groups, except the Asian group, where the gender gap is the narrowest in cleaning tasks rather than shopping. Overall, the data presented in Figures 13 and 14 allow us to observe some differences by cultural groups in the individual domestic tasks but the trends are overall similar for all cultural groups: women do less and men do more housework over time. There is a small reversal in trend 01002003004001986 1992 1998 2005 2010Domestic Task, Chinese CanadiansChinese WomenChinese Men01002003001986 1992 1998 2005 2010Domestic Tasks, South Asian CanadiansSouth Asian WomenSoouth Asian Men01002003004001986 1992 1998 2005 2010Domestic Tasks, Filipino CanadiansFilipino WomenFilipino Men85  for South Asian and Filipino women and Chinese men move away from the common pattern in the period between 2005 and 2010 but because of the small sample sizes these findings should not be considered as definitive. To address these yearly survey inconsistencies, models control for survey year and analyze the gender gap using the full sample. 4.3 Gender Norms and Same-sex couples Would gender be less relevant in doing family with a same-sex partner? How does the division of labour work in a household if traditional expectations are not applicable to the couple? Among couples where no traditional gendered roles are presumed such as in same-sex couples, gender roles are not expected to work in the same way as with heterosexual couples. Some researchers show that same-sex couples divide housework in a more egalitarian way (Kurdek 1993, 2006; Patterson 2000). Others find that the division of labour will still exist in same-sex couples depending on the resources and other power differentials established within a couple (Carrington 1999; Oerton 1997, 1998). Some differences were also found among lesbian couples as opposed to gay couples: lesbians were on average less likely to do housework and more likely to divide the housework equally between partners because they would deny the idea of being a housewife (Kurdek 2007; Oerton 1997). Goldberg (2013) argues even further to go beyond the idea of doing and undoing gender to understand the division of domestic labour and the notion of egalitarianism. The domestic work is more complicated than the perceived duality of femininity and masculinity and of gendered and non-gendered housework, it goes beyond our understanding of gendered roles but it is intertwined with the labour market, cultural, and other social dimensions. The analysis of housework division in same-sex couples, according to 86  Goldberg (2013), does little to improve our understanding of housework but adds to illustrate the complexities around it.  Table 6 compares the averages for time spent on various domestic tasks among same-sex couples with those of the averages among heterosexual couples in between 1998 and 2010.32 Women in same-sex partnerships do a little less housework than heterosexual women in general (6-7 minutes less on average), while men do as much housework in same-sex partnerships as they do in heterosexual couples. As to specific domestic tasks, women in same-sex partnerships do less of routine indoor tasks but more of outdoor routine and non-routine tasks, which are traditionally considered more gender-neutral or more masculine. Men in same-sex couples spend on average twice as much time on cooking than men on average, 48 minutes on an average diary day, yet for all outdoor routine and non-routine tasks they do a little less than heterosexual men in general. Overall, however, women even if they are in same-sex partnerships spend more time on housework tasks than men.  Table 6 Descriptive Statistics of Dependent Variables by Gender Between Same-Sex couples, 1998-2010, and All Couples, 1998-2010 Variables Description Mean (Women)a Bootstrap SE Mean (Men)b Bootstrap SE Dependent Variables         All Domestic Tasks Heterosexual 200.85 (0.342) 115.28 (0.295)  Same-sex 193.32 (5.016) 113.90 (2.761)   Cooking Tasks Heterosexual 68.66 (0.140) 28.89 (0.097)  Same-sex 46.11 (1.316) 48.10 (1.301)   Cleaning Tasks Heterosexual 70.17 (0.193) 22.47 (0.135)  Same-sex 52.22 (2.250) 24.90 (1.512)   Shopping Tasks Heterosexual 55.84 (0.180) 42.71 (0.181)  Same-sex 80.05 (3.898) 39.15 (1.722)   Maintenance Tasks Heterosexual 6.18 (0.130) 21.21 (0.155)  Same-sex 14.94 (1.433) 1.75 (0.443)                                                    32 The data on partners’ sex are not available for 1986 and 1992 GSS. 87  aN for heterosexual women = 13534.1 and N for women in same-sex partnerships = 67. bN for heterosexual men = 14264.6 and N for men in same-sex partnerships = 66.1. Even though on average, women do more housework than men even if they are in same-sex relationships, it does not necessarily mean that they do not share the housework in a more egalitarian way as, for example, Kurdek (2006) concludes. It might be that both women in the same-sex relationship do equally more housework under the pressure from the broader societal norms on what constitutes more ‘feminine’ roles. Kurdek (1993) also concludes that even though all same-sex couples divide housework in a more egalitarian way than heterosexual couples, on average, lesbians do more housework than gay partners. Thus, I conclude that same-sex couples’ involvement in housework is in no sense ‘gender empty’ (Oerton 1997) but lesbians and gays are in fact under the same pressure from the society to follow traditional roles in the housework as heterosexual women and men, where women do more housework on average than men, reaffirming the omnipresent coercion to comply with the belief in gender identity (Butler 1990). The results achieved herewith should not be considered as necessarily representative because of the very small sample size of same-sex couples (67 women and 66 men). Yet the relative consistency of the findings with the previous research (Kurdek 2007) provides us with the hope to reach a general idea of the differences between lesbian and gay couples in housework as the result of the broader societal pressures.   4.4 Results: Analyzing the Gender Gap in Housework The separate estimates for all, Anglo-, French, Chinese, South Asian, and Filipino Canadians are used to decompose the gap in the time spent on the four types of housework. Table 7 summarizes what share of the gender gap factors used in the models can explain for each type of domestic task, reporting both explained and unexplained portions of the gender gap. The overall 88  summaries presented in the table are based on full decomposition models, or Columns 6 in the Tables of the following subsections. The decomposition models can explain a third of the routine indoor housework, most of the routine outdoor tasks such as shopping, but cannot explain non-routine housework. Appendix B provides similar results in the US, where most of the gender gap can be explained in shopping followed by indoor routine tasks, and the least of the gender gap explained in maintenance.33 These results both in Canada and in the US suggest that women face more pressure in terms of spending time in indoor routine housework because the average, or the norm, shows higher normative performance of cooking and cleaning acts for women, while the frameworks proposed herewith are not suitable for the attempts at explaining the gender gap in non-routine housework such as maintenance.34 Among routine indoor tasks, the models provide explanation for a greater share of the time spent on cleaning than in cooking (39%), which is also true for the ATUS sample (see Appendix B, 20%), compared to only about a third of the time spent on cooking among the Canadian sample (see Table 7, 31%) and only a half of that in the United States (14%). Thus even though cleaning is considered a less enjoyable task, the portion of the gender gap that can be ascribed to discrimination and other unexplained variance is higher in cooking than in cleaning both in Canada (see Table 7) and in the US (see Appendix B). The models have the least strength in their ability to explain non-routine tasks (maintenance), which is unsurprising given that time use diaries are more appropriate for capturing routine activities due to their diurnal character. The negative numbers associated with maintenance tasks in Table 7 can be read as the amount that the time gap would increase if women had the same level as                                                    33 The actual percentages of the share explained cannot be compared to arrive at meaningful results because of the differences in measures and differences in models. However, the relative contributions of frameworks within the analysis of the ATUS data can be compared to results within the GSS data, which I report in here. 34 This is not surprising because most of the theoretical frameworks aim at explaining why women perform more of domestic chores and not at explaining why men do more of housework such as maintenance. 89  men on all factors in models. For instance, the negative 31% for the maintenance tasks among all Canadians indicates that if the gap between women’s and men’s factors such as for example differences in personal income and education would close, the gap in time that that they respectively spend on maintenance would actually increase.  Table 7 Explained and Unexplained Gender Gap in Domestic Tasks by Pooled Decomposition, in %35   All English French Chinese South Asian Filipino Cook explained 31*** (0.617) 33*** (0.690) 30*** (1.170) 34*** (2.964) 10 (4.406) 16 (5.522) unexplained 69*** (0.957) 67*** (1.161) 70*** (1.804) 66*** (5.784) 90*** (7.605) 84*** (9.232) Clean explained 39*** (0.843) 37*** (1.094) 42*** (1.629) 74*** (3.470) 22 (3.892) 49** (6.803) unexplained 61*** (1.297) 63*** (1.702) 58*** (2.339) 26 (5.535) 78*** (6.730) 51* (10.432) Shop explained 97*** (0.905) 80*** (1.128) 153*** (1.733) 26 (4.846) -737* (6.648) 311* (6.974) unexplained 3 (1.500) 20 (1.894) -53 (2.854) 74 (8.646) 837 (10.226) 211 (12.098) Main explained -31*** (0.624) -32*** (0.746) -29*** (1.501) -43 (2.776) 26 (1.556) -286 (9.920) unexplained 131*** (1.182) 132*** (1.426) 129*** (2.685) 143 (4.470) 74 (1.922) 386 (9.948) Bootstrap standard errors in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001 Most of the explanations that work for Anglo- and French Canadians work also for the explanation of the gender gap among other Canadians. Similarly, in the US, there are little to no differences between racial groups in what frameworks account for a greater share of the gender gap in types of housework (see Appendix B). The relative resources argument appears to hold for the indoor routine tasks such as cooking among all Canadians except Filipino Canadians, where the time availability framework can provide a better explanation. The relative resources can                                                    35 In many models, especially in full models of cultural groups where the initial sample size is not large enough, the effects of factors are amplified because the samples are sliced in such a way when for some levels of factors, there are not enough observations available to make reliable predictions. Yet I chose to keep all factors in the models even with smaller sample size to ensure consistency across all models. Many control variables also introduce noise specifically year and province/state-level variables. In the future research, these issues could be addressed with improving models. One direction could be using the 2SLS models instead of OLS. The complete data on decomposition outputs can be provided upon request (kamilakolpashnikova(at)gmail.com). 90  likewise provide the most explanation of the gender gap in cleaning among all Canadians except Filipino Canadians. In contrast, however, among Americans, the time availability framework can account for the biggest share of the gender gap in all types of routine housework (see Appendix B). The results are similar to those of Shelton (1992), where she finds that time constraints on American women especially those imposed by employment and parental statuses affect the persistence of the gender gap in housework. This discrepancy between Canadian and American results can be explained by contextual differences between Canadian and American societies such as, while in the former, the decision-making may rely more on the bargaining in the indoor routine work, individuals within the latter may be more prone to bargaining with their time rather than resources. Thus in America, the partner who has more time is expected to perform more housework. This more egalitarian decision-making process among Americans may also be only a reflection that the ATUS data is more recent than the GSS.   Contrary to the argument proposed by Gupta (2007), the autonomy argument does not hold more explanatory power for the gender gap among Canadians or Americans, yet it has the highest share in explaining the gender gap in cooking among Chinese and French Canadians but is only in the third place after the relative resources and the time availability framework. Similarly, among Asian Americans, the absolute resources can account for the biggest share compared to other racial groups in indoor routine housework but not more than the share of the time availability framework (see Appendix B). Overall, these results suggest that the decision-making in the North American households regarding the division of housework rarely rely on the autonomous decision by one of the partners, but rather is propelled by the bargaining process in Canada and time availability in the US.  91  Furthermore, the gender-centred framework factors measured in education and age effects can provide very little explanation to the gender gap in any type of housework among any group of Canadians and Americans. Gender discrimination together with unobserved variables measured as the unexplained part of the gender gap in housework can explain most of the differences in the time spent on cooking and maintenance among all Canadians, on cleaning among Anglo-, French, and South Asian Canadians. The share of the gap that can be ascribed to discrimination and unobserved variables is higher among South Asian Canadians for all routine housework. Similarly, the portion of the gender gap that can be ascribed to discrimination and unobserved variables is also higher among Black and Native Americans compared to other groups in the analysis, namely White and Asian Americans. Thus 85% of the gender gap in domestic tasks among Black and 86% of the gender gap among Native Americans cannot be accounted for by the factors of the four main frameworks tested in the present project (see Table in Appendix B). Overall, among routine housework, a bigger share of the gender gap can be ascribed to discrimination and other unobserved variables in indoor tasks such as cooking and cleaning rather than in outdoor tasks such as shopping both among Canadians and Americans. Table 7 shows that the models can explain most of the gender gap in shopping, especially among Filipino and French Canadians. The percentages over 100 mean that some factors have a negative influence on the explanations of the gap (leveling them up would increase the gap), making others to account for more than the total gap. It is worth noting that the gender gap is the narrowest in the shopping tasks in these two groups. The only group of Canadians where men are more likely to shop than women is South Asians, which can be interpreted as the remnant of the traditional association of shopping with men’s work rather than women’s work. These cultural 92  specifics can be illustrated with a quote about shopping among South Asian women in Britain borrowed from Hamlett et al. (2008): I wasn’t allowed to go in the shops in those days. It was only men who usually go in the shop, they don’t like their women going in front of everybody else, so it is his job. He used to bring everything that I needed. I just had to tell him if I was short of anything. (Bradford Heritage Recording Unit, 1994: 77). A little more than a third to two thirds of the gender gap variation can be explained by the differences in proposed factors in cleaning tasks and approximately one third of the variation in the gender gap in cooking. Chinese Canadians appear to have the most egalitarian views about cleaning tasks among the Canadian groups in the analysis because while having one of the narrowest gaps in the time spent on cleaning, the gender gap among the Chinese can be mostly attributed to the factors proposed by the current models (most of them from the economic domain and measurable). The proposed factors, though, cannot explain gender differences in participation in maintenance tasks, where a large percent of the differences is left unexplained.  4.4.1.1 Indoor Routine Tasks: Cooking Figure 15 summarizes the results for all theoretical frameworks separately as they apply to the five cultural groups in Canada: Anglo-, French, Chinese, South Asian, and Filipino Canadians. Summaries presented in figures of this Chapter are obtained from the decomposition tables. The results for the cultural groups are contrasted with the results for the total sample. The numbers associated with the bars in Figure 15 can be read as the amount that the time gap would increase (if the number is negative) or decrease (if the percentage is positive) if women had the same level as men on the factor that is being tested. For instance, the negative 1.8% for the relative 93  resources framework among Filipino Canadians indicates that if the gap between Filipino women’s and Filipino men’s relative contribution to household income would close, the gap in time that they respectively spend on cooking would actually increase. These negative effects mean that other factors would have to account for more than the total percentage of the explained difference if the explained portion of the gap is positive or that the unexplained part is more than 100% which is the case for this model among Filipino Canadians. The gender gap in cooking is left unexplained if relative economic contribution is used to test the association between resources and time spent on cooking among Filipino Canadians. The income transfer variable (referred to as the relative economic contribution when used as a concept) testing the relative resources argument is the highest single predictor of the gender gap in cooking among all cultural groups of Canadians except Filipino Canadians. Among Filipino Canadians, the time availability framework measured in time spent on paid work explains most of the gender gap in time spent on cooking. Considering that the time availability framework has the most explanatory power in the most egalitarian housework, shopping, it could be that the division of labour in cooking among Filipino Canadians is based on the most egalitarian processes among Canadian cultural groups. In fact, the proportional gender gap is the second smallest among Filipino Canadians only after the French Canadians. Thus in 2010, the share of French women’s time spent on cooking was 1.96 times bigger than French men’s, whereas the share of Filipino women was 2.02 of that of Filipino men’s. As the two main cultural groups, English and French Canadians, there are small differences in the explanatory power of the relative resources framework.  94  In Tables 8-12, I report results from the Blinder-Oaxaca decomposition to see how much each of the measures within tested frameworks account for the time spent on cooking among married and cohabiting Canadians. The tables summarize the percentage of the gender gap that can be accounted for by individual frameworks among cultural groups, while the final column with the full models shows the percentage explained when controlled for all other frameworks. All summaries within the decomposition tables in the Chapter are based on Blinder-Oaxaca decomposition models that control for all the main control variables and province-level variables, mentioned in the methods section.  Yet it makes sense to consider frameworks without controlling for other competing approaches because of the strong multicollinearity among economic factors. The issue of multicollinearity in the full models results in most of the variance explained in the time spent on cooking to be absorbed by one of the respective explanatory factors while underestimating the input of another. In the case of the GSS data, most of the variance explanation in time spent on cooking is absorbed by the income transfer variable, testing for the relative resources framework. Thus in the last column of Tables 8-9, the income transfer stays statistically significant and can still explain over 20% of the variation both among all Canadians except Filipino Canadians.  Time spent on paid work, representing the time availability perspective, is the second most potent predictor of the gender gap both for Canadians of all cultural groups except Filipino Canadians, where it plays the major role (see Figure 15). It can explain 15.3% of the gender gap in cooking among Anglo-Canadians and 14.8% among French Canadians (see Tables 8-9).  Among factors that test the autonomy framework, the strongest overall predictor is whether the respondent is in the workforce or not. It can explain 7.3% of the gender gap among English 95  Canadians and 6.9% among French Canadians, it is the strongest among the Chinese Canadians where it can explain 11.8% of the gender gap in cooking (see Tables 8-10). Personal income as a proxy for the autonomy approach appears as a significant predictor only for the French and Chinese Canadians. Thus the differences in personal income can explain 7.1% of the gender gap among French and 15.2% - among Chinese Canadians. Clearly, the autonomy approach to explanation of the gender gap in Canadian households does not hold in cooking: even though it is significant, other frameworks appear to be more potent in terms of their ability to explain gender differences in cooking. This finding casts doubt to the claims in Gupta (2007) regarding the autonomy explanation’s applicability to Canadian households and to cooking tasks in particular. In the full model that combines all frameworks together, personal income does not appear significant and is only significant among Anglo-Canadians, thus, suggesting that the effect of the main indicator for autonomy is absorbed by other factors and other competing arguments in explaining the gender gap among Canadians.   Figure 15 Percent Explained of the Gender Gap in Cooking by Different Frameworks On the other hand, the other factor used to test the autonomy framework, whether the respondent is in the labour force, appears to explain around 6.9% among French Canadians and 22.5 23.4 22.829.815.8-1.814.8 15.3 14.818.09.521.20.3 0.27.115.21.7 0.50.0-0.10.2 1.0-0.12.3-10.00.010.020.030.040.0All English French Chinese South Asians FilipinoPercent Explained of the Gender Gap in Time Spent on Cooking, CanadaRelative resources Time Availability Autonomy Gendered Processes96  up to 18.2% - among Filipino Canadians, the factor is a significant predictor of the time spent on cooking in all cultural groups except South Asian Canadians, where the association of labour force participation with the time spent on cooking is small (3.9%) and insignificant. The ‘working’ variable is not shown in Figure 15 but is presented in the model outputs in Tables 8-12. It is likely that for South Asian Canadians the decision-making relies more on joint decisions because the relative resources framework explains most of the gender gap, while the autonomy approach factors do not seem to be significantly related with the time spent on cooking.  The effects of age are significant but in the sense that the differences in age certainly do not add much to the explanation of the gender gap, instead the age differences would result in a wider gender gap if the age gap was to close.    97  Table 8 Percent Explained in Gender Gap in Time Spent on Cooking among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All English All English All English All English All English All English Relative Resources     Income Transfer 22.5*** 23.4***         18.7*** 21.3*** Time Constraint                  Paid Work   14.8*** 15.3***       12.7*** 13.4***      Leisure (total)   1.9*** 2.2***       2.4*** 3.1*** Children   -0.0 -0.1       0.1** 0.1* Under 5   0.0 0.0       0.0 0.0 Household Size   0.0 0.0       0.1** 0.3 Autonomy                  Personal Income    0.1 0.1     -0.0 -0.0      Working     7.3*** 7.3***     1.5*** 1.3*     Owns Home     -0.0 -0.1**     0.0 -0.1 Personal Income             Income       0.3 0.2     Gendered Processes Age         -1.2*** -1.3*** -1.0*** -1.3*** Education         0.0 -0.1** -0.0 -0.1 Leisure Men         1.1*** 1.1***   Leisure Women         -0.6*** -0.3   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 Total % explained 22.5*** 23.3*** 16.8*** 17.6*** 7.9*** 7.3*** 0.6* 0.1 -0.7** -0.7 31*** 33*** % unexplained 77.5*** 76.7*** 83.2*** 82.4*** 92.1*** 92.7*** 99.4*** 99.9*** 100.7*** 100.7*** 69*** 67*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001    98  Table 9 Percent Explained in Gender Gap in Time Spent on Cooking among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All French All French All French All French All French All French Relative Resources     Income Transfer 22.5*** 22.8***         18.7*** 20.9*** Time Constraint                  Paid Work   14.8*** 14.8***       12.7*** 12.6***      Leisure (total)   1.9*** 1.1*       2.4*** 1.4* Children   -0.0 0.0       0.1** 0.2* Under 5   0.0 -0.0       0.0 -0.0 Household Size   0.0 0.1       0.1** 0.3* Autonomy                  Personal Income    0.1 0.1     -0.0 -0.0      Working     7.3*** 6.9***     1.5*** 1.6     Owns Home     -0.0 0.0     0.0 0.0 Personal Income             Income       0.3 7.1***     Gendered Processes Age         -1.2*** -1.1*** -1.0*** -1.2* Education         0.0 0.2* -0.0 0.0 Leisure Men         1.1*** 0.4   Leisure Women         -0.6*** -0.6**   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 Total % explained 22.5*** 23.4*** 16.8*** 16.0*** 7.9*** 10.6*** 0.6* 7.5*** -0.7** -1.2* 31*** 30*** % unexplained 77.5*** 76.6*** 83.2*** 84.0*** 92.1*** 89.4*** 99.4*** 92.5*** 100.7*** 101.2*** 69*** 70*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001    99  Table 10 Percent Explained in Gender Gap in Time Spent on Cooking among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Chinese All Chinese All Chinese All Chinese All Chinese All Chinese Relative Resources     Income Transfer 22.5*** 29.8***         18.7*** 20.7 Time Constraint                  Paid Work   14.8*** 18.0***       12.7*** 10.1***      Leisure (total)   1.9*** -1.6       2.4*** 0.1 Children   -0.0 -0.1       0.1** 1.7 Under 5   0.0 0.4       0.0 -0.1 Household Size   0.0 0.0       0.1** -0.7 Autonomy                  Personal Income    0.1 0.1     -0.0 -0.0      Working     7.3*** 11.8***     1.5*** 0.7     Owns Home     -0.0 0.1     0.0 2.2 Personal Income             Income       0.3 15.2***     Gendered Processes Age         -1.2*** -5.4** -1.0*** 1.1 Education         0.0 1.0 -0.0 1.1 Leisure Men         1.1*** 0.1   Leisure Women         -0.6*** -3.0*   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 961 39844 961 39844 961 39844 961 39844 961 39844 961 Total % explained 22.5*** 30.7*** 16.8*** 16.2*** 7.9*** 16.7*** 0.6* 13.0*** -0.7** -7.9** 31*** 34*** % unexplained 77.5*** 69.3*** 83.2*** 83.8*** 92.1*** 83.3*** 99.4*** 87.0*** 100.7*** 107.9*** 69*** 66*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001     100  Table 11 Percent Explained in Gender Gap in Cooking among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All SA All SA All SA All SA English SA All SA Relative Resources     Income Transfer 22.5*** 15.8***         18.7*** 21.0** Time Constraint                  Paid Work   14.8*** 9.5***       12.7*** 8.4**      Leisure (total)   1.9*** 0.4       2.4*** -2.0 Children   -0.0 0.2       0.1** -0.0 Under 5   0.0 0.1       0.0 -0.0 Household Size   0.0 -0.3       0.1** -0.3 Autonomy                  Personal Income         0.1 0.1     -0.0 -0.0      Working     7.3*** 3.9     1.5*** -6.3*     Owns Home     -0.0 -0.3     0.0 0.1 Personal Income             Income       0.3 1.7     Gendered Processes Age         -1.3*** 0.1 -1.0*** -2.7 Education         -0.1** -0.1 -0.0 0.0 Leisure Men         1.1*** -2.5***   Leisure Women         -0.3 1.4   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 829 39844 829 39844 829 39844 829 21515 829 39844 829 Total % explained 22.5*** 15.8*** 16.8*** 10.2*** 7.9*** 5.1 0.6* 2.1 -0.7** -0.6 31*** 10 % unexplained 77.5*** 84.2*** 83.2*** 89.8*** 92.1*** 94.9*** 99.4*** 97.9*** 100.7*** 100.6*** 69*** 90*** P-values are reported for the coefficients. SA- South Asian. * p < 0.05, ** p < 0.01, *** p < 0.001   101  Table 12 Percent Explained in Gender Gap in Time Spent on Cooking among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Filipino All Filipino All Filipino All Filipino English Filipino All Filipino Relative Resources     Income Transfer 22.5*** -1.8         18.7*** -3.5 Time Constraint                  Paid Work   14.8*** 21.2***       12.7*** 13.8**      Leisure (total)   1.9*** 3.7       2.4*** -1.6 Children   -0.0 0.5       0.1** -0.1 Under 5   0.0 0.0       0.0 -0.0 Household Size   0.0 -1.9       0.1** -5.0 Autonomy                  Personal Income         0.1 -4.1     -0.0 -4.0      Working     7.3*** 18.2***     1.5*** 9.9     Owns Home     -0.0 1.1     0.0 0.3 Personal Income             Income       0.3 0.5     Gendered Processes Age         -1.3*** 0.1 -1.0*** -0.9 Education         -0.1** 2.3 -0.0 9.6* Leisure Men         1.1*** 2.3   Leisure Women         -0.3 -3.3*   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 337 39844 337 39844 337 39844 337 21515 337 39844 337 Total % explained 22.5*** -4.0 16.8*** 25.1*** 7.9*** 14.1* 0.6* 0.0 -0.7** 1.9 31*** 16 % unexplained 77.5*** 104*** 83.2*** 74.5*** 92.1*** 85.9*** 99.4*** 100.0*** 100.7*** 98.1*** 69*** 84*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   102  Moreover, the decomposition analysis shows that the age variable actually confounds the explanation of the gender gap rather than explaining it. The same story is evident for the effects of education. However, in the full model and only among Filipino Canadians, where I control for all theoretical frameworks, education appears to explain up to 9.6% of the gender gap. The gender gap among Filipino Canadians would decrease if women had the same education level as men. This is not applicable to other cultural groups of Canadians, where education plays a minor role in explaining the gender gap in cooking.  These results around the effects of age and education cast doubt on cultural capital related resource explanations and gender socialization explanations through education and the generational shift. Yet there is at least 67% (among Anglo-Canadians) and at most 90% (among South Asian Canadians) of the gender gap in cooking that cannot be explained by the factors discussed in the present project. It is likely that gender-related processes that happen regardless of the level of education, level of absolute or relative resources, and of the generational belonging affect the gender gap across all cultural groups in Canada. The greatest portion of the gap in cooking can be explained by the economic factors discussed here among Anglo-Canadians, while still 67% of the differences in cooking can be ascribed to insidious gender discrimination, unmeasured characteristics and other unobserved variance. These are disadvantages women face regardless of their cultural background, education level, economic resources, or age.  Overall, the results of the decomposition method show the relevance of the relative resources framework to the explanation of the gender gap in time spent on cooking more than other economic frameworks, including the autonomy approach, argued for by Gupta (2007). The 103  autonomy approach appears to be applicable to the explanation of the gender gap in two groups: French and Chinese Canadians, where the decision making process around who does cooking may rely also on autonomous decisions (Blood and Wolfe 1960), more so among Chinese Canadians than among French Canadians. Moreover, Filipino Canadians stand alone in their more egalitarian approach to cooking, where most of the division of this type of housework seems to depend on who has more time available. Thus among Filipino Canadians, a partner with more time available will be more likely to perform the housework. This finding comes in tandem with another idiosyncrasy of the cultural group – education seems to have more bearing among Filipino Canadians than among other Canadians, suggesting perhaps that there are more drastic differences in class as defined by education level among Filipino Canadians than any other cultural group when it comes to cooking. 4.4.1.2 Indoor Routine Tasks: Cleaning In many aspects the explanation of the gender gap in time spent on cleaning echoes that for the other indoor routine task – cooking. One difference from cooking is that a little bit more of the gender gap in cleaning tasks can be accounted for by the factors explored in this project. Thus the full model explains about 8% more of the gender gap, increasing from 31% in cooking to 39% in cleaning (see Table 7). For Chinese Canadians, the explanatory power of all factors combined increases from 34% in the gender gap in cooking to 74% - in cleaning tasks. Moreover, when only the sample of individuals who report cleaning tasks is considered, the overall ability of the frameworks to explain the gender gap increases from 39% to astonishing 76%, as shown in Table 13, which summarizes the gender gap among those Canadians who report any housework time on a diary day. Note that the difference between the explained gender 104  gap in cooking among all married and cohabiting Canadians (31%) and among those who report cooking on the diary day (33%) is minimal. The greater difference between the total sample of married and cohabiting Canadians and those among them who report doing any cleaning on the diary day is likely to be the result of the non-daily character of cleaning activities. Thus married and cohabiting Canadians, on average, spend 10 minutes more doing cleaning on weekends than on weekdays, and about half of them report doing any cleaning tasks on the diary day, thus some of the cleaning might happen on other days rather than diary day especially if it is not a Saturday or a Sunday. These are clearly the limitations of the time use data in capturing routine housework that does not necessarily occur on a daily basis. Table 13 Explained and Unexplained Gender Gap among Those Who Report Any Housework Time by Pooled Decomposition, in %  Cooking Cleaning Shopping Maintenance explained 33*** 76*** 94*** -40* unexplained 67*** 24 6 140*** * p < 0.05, ** p < 0.01, *** p < 0.001  Figure 16 summarizes the contribution of various main factors by theoretical approach to the explanation of the gender gap in cleaning. Similar to time spent on cooking, the gender gap in cleaning can mostly be accounted for by the relative resources argument for all groups of Canadians, except Filipino Canadians and to a small degree among Anglo-Canadians. Among Anglo-Canadians, gender differences can mostly be explained by the time availability framework followed by the relative resources framework. However, the difference between the two frameworks’ explanatory ability measured by income transfer and time spent on paid work is 105  miniscule – 22.4% for the time availability framework and 21.8% for the relative resources factor.  Likewise, as with results for the other routine indoor task – cooking – I find that the autonomy framework is also relevant only in two cultural groups of Canadians: French and Chinese. Thus the autonomy framework can explain 25% of the gender gap in cleaning among Chinese Canadians and 10% of the gender gap among French Canadians. I contend that these two cultural groups in Canada are more likely to employ the autonomous decision-making within households than other cultural groups among those analyzed by the present project. Autonomous decision-making process is akin to the idea of conflict management through avoidance in households (Buss and Schaninger 1983), which may suggest that French and Chinese Canadians prefer to avoid conflict with their partners through autonomous decision-making more often than other cultural groups who resolve conflict about sharing housework mostly through bargaining process. These findings suggest the existence of minor cultural differences in how the division of housework happens across cultural lines.   Figure 16 Percent Explained of the Gender Gap in Cleaning by Different Frameworks 24.6 21.828.852.630.716.822.4 22.4 23.030.523.940.20.4 0.29.925.45.0 6.10.0-0.30.5 0.8 0.2-1.5-10.00.010.020.030.040.050.060.0All English French Chinese South Asians FilipinoPercent Explained of the Gender Gap in Time Spent on Cleaning, CanadaRelative resources Time Availability Autonomy Gendered Processes106   Another notable cultural difference also observed in the analysis of the gender gap in cooking, is that for Filipino Canadians, it is the time availability framework that can explain the highest portion of the gender gap in cleaning. Considering that this type of reflection of the decisions made on the basis of the available time is common among the most egalitarian housework – shopping – I believe that results show that Filipino Canadians have one of the most egalitarian division of housework, primarily based on the available time. To some extent the same applies to Anglo-Canadians in their division of cleaning tasks within a household, since the time availability approach can account for a little more (22.4%) than the relative resources explanation (21.8%).  Tables 14-18 summarize the decomposition of the gender gap results in time spent on cleaning in detail by different cultural groups of Canadians, specifically comparing their results to overall pattern of the total sample (columns ‘All’ in the Tables). These tables are based on the decomposition coefficients and models controlling for the main control and province-level variables. For instance, Anglo-Canadians, being the biggest of the cultural groups in Canada, show results that resemble the overall pattern. Yet for the overall pattern, the relative resources framework can account for the biggest portion of the gender gap, whereas among English Canadians, the biggest portion of the gender gap is explained by the time availability framework.        107  Table 14 Percent Explained in Gender Gap in Time Spent on Cleaning among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All English All English All English All English All English All English Relative Resources     Income Transfer 24.6*** 21.8***         16.3*** 15.7*** Time Constraint                  Paid Work   22.4*** 22.4***       20.2*** 20.6***      Leisure (total)   3.3*** 3.3***       4.1*** 4.4*** Children   -0.0 -0.0       0.0 -0.0 Under 5   0.0 0.1**       0.1** 0.3*** Household Size   0.0 0.0       0.1* 0.3* Autonomy                  Personal Income    0.2 0.1     -0.0 -0.0      Working     7.9*** 7.0***     0.7 -0.1     Owns Home     -0.0 -0.1*     0.0 -0.1 Personal Income             Income       0.4 0.2     Gendered Processes Age         -1.1*** -1.1*** -0.3 -0.6 Education         0.0 -0.3*** -0.1** -0.3*** Leisure Men         2.3*** 2.4***   Leisure Women         -1.5*** -1.5***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 Total % explained 24.7*** 21.3*** 25.7*** 25.5*** 8.1*** 6.4*** 0.3 -0.4 -0.4 -1.0 39*** 37*** % unexplained 75.3*** 78.7*** 74.3*** 74.5*** 91.9*** 93.6*** 99.7*** 100.4*** 100.4*** 101.0*** 61*** 63*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001     108  Table 15 Percent Explained in Gender Gap in Time Spent on Cleaning among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All French All French All French All French All French All French Relative Resources     Income Transfer 24.6*** 28.8***         16.3*** 24.1*** Time Constraint                  Paid Work   22.4*** 23.0***       20.2*** 19.7***      Leisure (total)   3.3*** 2.9***       4.1*** 3.3*** Children   -0.0 0.0       0.0 -0.0 Under 5   0.0 -0.2**       0.1** -0.0 Household Size   0.0 0.1       0.1* 0.4* Autonomy                  Personal Income    0.2 5.8***     -0.0 -1.9      Working     7.9*** 6.8***     0.7 0.5     Owns Home     -0.0 0.0     0.0 0.0 Personal Income             Income       0.4 9.9***     Gendered Processes Age         -1.1*** -0.7* -0.3 -0.1 Education         0.0 0.5*** -0.1** -0.1* Leisure Men         2.3*** 2.0**   Leisure Women         -1.5*** -1.5***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 Total % explained 24.7*** 29.0*** 25.7*** 25.9*** 8.1*** 12.8*** 0.3 9.9*** -0.4 0.5 39*** 42*** % unexplained 75.3*** 71.0*** 74.3*** 74.1*** 91.9*** 87.2*** 99.7*** 90.1*** 100.4*** 99.5*** 61*** 58*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001     109  Table 16 Percent Explained in Gender Gap in Time Spent on Cleaning among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Chinese All Chinese All Chinese All Chinese All Chinese All Chinese Relative Resources     Income Transfer 24.6*** 52.6***         16.3*** 35.4 Time Constraint                  Paid Work   22.4*** 30.5***       20.2*** 18.0***      Leisure (total)   3.3*** 3.3       4.1*** 10.7* Children   -0.0 -0.4       0.0 -0.8 Under 5   0.0 0.3       0.1** -0.2 Household Size   0.0 -1.4       0.1* -4.1* Autonomy                  Personal Income    0.2 16.5**     -0.0 4.6      Working     7.9*** 13.6**     0.7 5.4     Owns Home     -0.0 1.8     0.0 4.9* Personal Income             Income       0.4 25.4***     Gendered Processes Age         -1.1*** -5.0 -0.3 6.4 Education         0.0 0.8 -0.1** 0.3 Leisure Men         2.3*** 3.8   Leisure Women         -1.5*** -3.6*   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 961 39844 961 39844 961 39844 961 39844 961 39844 961 Total % explained 24.7*** 55.4*** 25.7*** 33.5*** 8.1*** 30.3*** 0.3 23.9*** -0.4 -3.1 39*** 74*** % unexplained 75.3*** 44.6* 74.3*** 66.5*** 91.9*** 69.7*** 99.7*** 76.1*** 100.4*** 103.1*** 61*** 26 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001     110  Table 17 Percent Explained in Gender Gap in Cleaning among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All SA All SA All SA All SA English SA All SA Relative Resources     Income Transfer 24.6*** 30.7**         16.3*** 46.0** Time Constraint                  Paid Work   22.4*** 23.9***       20.2*** 16.8**      Leisure (total)   3.3*** 3.8       4.1*** -0.7 Children   -0.0 0.9       0.0 -0.6 Under 5   0.0 -0.6       0.1** 1.0 Household Size   0.0 0.2       0.1* 0.3 Autonomy                  Personal Income         0.2 1.2     -0.0 -5.7*      Working     7.9*** 15.0**     0.7 -5.9     Owns Home     -0.0 0.0     0.0 0.1 Personal Income             Income       0.4 5.0     Gendered Processes Age         -1.1*** -0.6 -0.3 -8.2 Education         0.0 0.2 -0.1** -0.8 Leisure Men         2.3*** -3.2   Leisure Women         -1.5*** 3.3   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 829 39844 829 39844 829 39844 829 21515 829 39844 829 Total % explained 24.7*** 29.7** 25.7*** 28.9*** 8.1*** 15.8** 0.3 4.9 -0.4 0.1 39*** 22 % unexplained 75.3*** 70.3*** 74.3*** 71.1*** 91.9*** 84.2*** 99.7*** 95.1*** 100.4*** 99.9*** 61*** 78*** P-values are reported for the coefficients. SA- South Asian. * p < 0.05, ** p < 0.01, *** p < 0.001   111  Table 18 Percent Explained in Gender Gap in Time Spent on Cleaning among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Filipino All Filipino All Filipino All Filipino English Filipino All Filipino Relative Resources     Income Transfer 24.6*** 16.8*         16.3*** 30.0* Time Constraint                  Paid Work   22.4*** 40.2***       20.2*** 30.8***      Leisure (total)   3.3*** -2.6       4.1*** -11.0 Children   -0.0 -1.0       0.0 0.0 Under 5   0.0 3.0       0.1** -0.7 Household Size   0.0 0.8       0.1* 2.6 Autonomy                  Personal Income         0.2 0.3     -0.0 -8.5      Working     7.9*** 25.9**     0.7 6.6     Owns Home     -0.0 4.2     0.0 2.9 Personal Income             Income       0.4 6.1     Gendered Processes Age         -1.1*** 0.4 -0.3 -0.5 Education         0.0 -1.5 -0.1** 5.1 Leisure Men         2.3*** 0.7   Leisure Women         -1.5*** -0.7   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 337 39844 337 39844 337 39844 337 21515 337 39844 337 Total % explained 24.7*** 7.8 25.7*** 37.7** 8.1*** 25.2* 0.3 1.6 -0.4 -5.1 39*** 49** % unexplained 75.3*** 92.2*** 74.3*** 62.3** 91.9*** 74.8** 99.7*** 98.4*** 100.4*** 105.1*** 61*** 51* P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   112  Notably among other factors, I find that being in the labour force is a significant predictor of the gender gap among Canadians, especially among Asian Canadians: Chinese, South Asian and Filipino groups. Being in the labour force can explain 25% of the gender gap among Filipino Canadians, 15% among South Asian Canadians and 14% among Chinese Canadians (Tables 16-18). By comparison, the participation in the labour force can explain only about 7% among Anglo- and French Canadians. Thus not having a job has a stronger implication in terms of involvement in cleaning tasks among Asian women than among Anglo- and French Canadian women. Additionally, the economic factors captured by the models are not the only contributors to the explanation of the gender gap in cleaning, more so than in the models for explaining the gender gap in cooking. The combined leisure variables have a higher impact on the explanation of the gender gap in cleaning compared to the gender gap in cooking, suggesting that men choose to do leisurely activities over cleaning more often than over cooking. Thus the combined leisure activities contribute 3.3% of the explanation of the gender gap in cleaning among all Canadians. In contrast, leisure variables can explain 1.9% of the gender gap in cooking among all Canadians. These differences in the effect of leisure variables on the time spent on housework tasks can be because most Canadians, on average, find cooking a more pleasant activity than cleaning. Considering that TV watching consumes the most portion of the time spent on leisurely activities and that men, on average, watch more TV than women, it can be said that instead of cleaning men often choose to watch television.  Taking into account that two thirds of the time spent on cleaning can be explained by all the factors in the models among Canadians who report doing any housework on the diary day, it appears that the economic perspectives – the relative resources, the autonomy argument, and the 113  time availability perspective – can explain most of the gender gap in time spent on cleaning, especially among Chinese Canadians, where the share of the gender gap explained is the highest. Yet cleaning is the domestic chore where the gender gap is widest: the share of women’s participation exceeded men’s by 2.88 times in 2010. I find that among the most unequal domestic tasks, the economic frameworks have a higher portion of explanatory power, specifically lying within the combination of the relative resources and time availability approaches. 4.4.1.3 Outdoor Routine Tasks: Shopping Factors analyzed in the present project can account for the highest portion of the gender gap in time spend on shopping compared to all types of housework (Table 7). This finding suggests that by leveling labour market opportunities for women with those of men, especially for all groups other than South Asian Canadians, we would be able to eradicate the gender gap in shopping.  Figure 17 summarizes the portion that individual factors from the main frameworks can account for in the gender gap in time spent on shopping. The figure shows that the time constraint framework, which uses time spent on paid work as a single proxy, can explain most of the gender gap in shopping time for all cultural groups in Canada except South Asian Canadians. Thus time spent on paid work can explain all of the gender gap in shopping among French, Chinese and Filipino Canadians. The share of the explained portion reporting more than 100% suggests that variables other than paid work (for which the decomposition models control for) contribute negatively and if the paid work equality between women and men would be achieved, the non-discriminatory distribution of the time spent on shopping would eradicate most of the gender gap in shopping if not all of it.  114  Tables 19-23 summarize the results of decomposition models, providing the percentage each of the main factors can account for in time spent on shopping between Canadian women and men. As to other factors within the time availability framework, time spent on overall leisurely activities can explain up to 47.2% of the gender gap in shopping among Filipino Canadians (see Table 23), 29.2% among French Canadians (see Table 20), and 13.8% among Anglo-Canadians (see Table 19). In addition to similar results in cooking and cleaning tasks, the gender gap in shopping among Chinese and South Asian Canadians cannot be ascribed to differences in leisurely activities to any significant portion (see Tables 21-22), unlike among the three other groups. It is also worth noting that leisure, especially that of screen time (most of the reported leisurely time is TV watching), depends heavily on individual cultural capital (Bourdieu 1980) and thus plays a role of a class signifier. These results provide evidence for the previous research establishing class differences in watching TV over other types of more high-brow leisurely activities (Bourdieu 1980; Lareau 2000; Molina, Campaña, and Ortega 2016).   Summarizing the results among various cultural groups for the three routine tasks, the findings show similarities among French and Chinese Canadians. Among them, the absolute measure of resources, personal income, contributes a high share of explanatory power for the gender gap in cooking, cleaning, and shopping. This finding indicates that the autonomous decision-making with regard to housework is more common among French and Chinese 115  Canadians than among other cultural groups in Canada.  Figure 17 Percent Explained of the Gender Gap in Shopping by Different Frameworks Two main differences are also noticeable between French and Chinese Canadians. First, participation in leisurely activities is a better predictor of the time availability explanation for the gender gap among French Canadians rather than among Chinese Canadians. This finding is consistent across all routine housework tasks. Leisure as a cultural activity is embedded in the lifestyles of French Canadians, particularly of the Quebec residents, for whom cultural leisure becomes a part of the expression of the Quebecois cultural identity (Smale 2010). Moreover, leisure has started to be considered an indispensable part of individual’s life in Quebec following the societal trends from 1971-2010 (Pronovost 2012). On the other hand, Chinese cultural values heavily relying on Confucianism and its principles of devotion, hard work, and diligence (Hu et al. 2014) may be able to explain the prevalence of the time constraint by work rather than leisurely activities among Chinese Canadians.  20.4 14.2 24.8 28.3-280.867.4103.3 82.5145.1351.6-341.1114.90.7 0.426.2 30.4-247.62.0-0.10.6-0.9 -6.513.9 11.5-400.0-300.0-200.0-100.00.0100.0200.0300.0400.0All English French Chinese South Asians FilipinoPercent Explained of the Gender Gap inTime Spent on Shopping, CanadaRelative resources Time Availability Autonomy Gendered Processes116  Second, factors lying within the economic approach account for a bigger portion of the gender gap in all types of routine housework among Chinese Canadians than among French (and other) Canadians. These findings indicate that economic factors predict more of the gender differences among Chinese Canadians than among other Canadians. Moreover, they suggest that a lesser portion of the gender gap in housework can be ascribed to discrimination and other unexplained factors among Chinese Canadians. Conversely, the frameworks that can account for the gendered division of shopping tasks among Filipino Canadians resemble the pattern among Anglo-Canadians. Thus for both Filipino and Anglo-Canadians, the time availability framework explains the biggest share of the gender gap in shopping while the autonomy framework measured by personal income does not have a significant explanatory power or has a negligible explanatory power, unlike among French and Chinese Canadians.  Gender roles in the Philippines are considered in general to be quite egalitarian historically (Piper and Roces 2003). Moreover, the Canadian immigration from the Philippines was mostly relying on Filipino women rather than men (Bonifacio 2013), making the first-generation immigrant Filipinas more likely to be employed in the first few years than Filipino men, who could be following their spouses into Canada. Yet this fact that more Filipino women than men migrated outside of the Philippines following economic opportunities as an explanation of the economic power of women over their partners, of course, can only apply to the first-generation immigrants, whereas the sample from the GSS includes not only the first generation but everyone who self-identifies as a Tagalog-speaker or reports being born in the Philippines.   117  Table 19 Percent Explained in Gender Gap in Time Spent on Shopping among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All English All English All English All English All English All English Relative Resources     Income Transfer 20.4** 14.2*         -31.5*** -40.3*** Time Constraint                  Paid Work   103.3*** 82.5***       108.0*** 89.5***      Leisure (total)   18.4*** 13.8***       19.7*** 17.2*** Children   0.1 0.3*       -0.5** -0.6** Under 5   0.0 0.3***       0.4*** 0.8*** Household Size   -0.0 -0.0       -0.1 -0.2 Autonomy                  Personal Income    0.1 0.1     0.3** 0.3***      Working     20.9*** 17.3***     -8.7*** -5.6*     Owns Home     0.0 -0.2     -0.0 0.1 Personal Income             Income       0.7 0.4     Gendered Processes Age         -8.7*** -7.8*** -5.0*** -3.7** Education         -0.1 0.6** 0.3** 0.8** Leisure Men         9.0*** 8.1***   Leisure Women         -5.4 -4.7***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 Total % explained 21.6*** 14.0* 122.3*** 97.9*** 22.1*** 17.2*** 1.5** 0.3 -4.6*** -3.4* 97*** 80*** % unexplained 78.4*** 86.0*** -22.3** 2.1 77.9*** 82.8*** 98.5*** 99.7*** 104.6*** 103.4*** 3 20 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   118  Table 20 Percent Explained in Gender Gap in Time Spent on Shopping among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All French All French All French All French All French All French Relative Resources     Income Transfer 20.4** 24.8         -31.5*** -29.3 Time Constraint                  Paid Work   103.3*** 145.1***       108.0*** 157.2***      Leisure (total)   18.4*** 29.2***       19.7*** 33.4*** Children   0.1 -0.0       -0.5** -0.2 Under 5   0.0 -0.4       0.4*** -0.1 Household Size   -0.0 -0.1       -0.1 -0.5 Autonomy                  Personal Income    0.1 6.0     0.3** 8.7      Working     20.9*** 27.1***     -8.7*** -17.7**     Owns Home     0.0 -0.0     -0.0 -0.3 Personal Income             Income       0.7 26.2***     Gendered Processes Age         -8.7*** -9.1*** -5.0*** -8.1* Education         -0.1 -0.9* 0.3** -0.9* Leisure Men         9.0*** 11.4***   Leisure Women         -5.4 -5.6***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 Total % explained 21.6*** 25.0 122.3*** 174.3*** 22.1*** 33.3*** 1.5** 25.6*** -4.6*** -3.8 97*** 153*** % unexplained 78.4*** 75.0* -22.3** -74.3** 77.9*** 66.7** 98.5*** 74.4*** 104.6*** 103.8*** 3 -53 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   119  Table 21 Percent Explained in Gender Gap in Time Spent on Shopping among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Chinese All Chinese All Chinese All Chinese All Chinese All Chinese Relative Resources     Income Transfer 20.4** 28.3         -31.5*** -33.9 Time Constraint                  Paid Work   103.3*** 351.6***       108.0*** 110.2***      Leisure (total)   18.4*** -24.4       19.7*** 15.1 Children   0.1 -8.2       -0.5** -22.1* Under 5   0.0 5.4       0.4*** -0.9 Household Size   -0.0 4.7       -0.1 3.5 Autonomy                  Personal Income    0.1 20.1     0.3** 10.6      Working     20.9*** 17.0     -8.7*** -34.6**     Owns Home     0.0 3.2     -0.0 -3.4 Personal Income             Income       0.7 30.4     Gendered Processes Age         -8.7*** -57.2 -5.0*** -4.6 Education         -0.1 -6.5 0.3** -3.4 Leisure Men         9.0*** 43.4*   Leisure Women         -5.4 -90.4***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 961 39844 961 39844 961 39844 961 39844 961 39844 961 Total % explained 21.6*** 33.6 122.3*** 339.5*** 22.1*** 33.2 1.5** 24.9 -4.6*** -100.8* 97*** 26 % unexplained 78.4*** 66.4 -22.3** 239.5 77.9*** 66.8 98.5*** 75.1 104.6*** 200.8 3 74 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   120  Table 22 Percent Explained in Gender Gap in Shopping among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All SA All SA All SA All SA English SA All SA Relative Resources     Income Transfer 20.4** -280.8         -31.5*** -111.3 Time Constraint                  Paid Work   103.3*** -341.1***       108.0*** -888.7***      Leisure (total)   18.4*** -88.5*       19.7*** -227.0* Children   0.1 8.5       -0.5** 0.8 Under 5   0.0 -8.0       0.4*** -4.0 Household Size   -0.0 2.3       -0.1 16.9 Autonomy                  Personal Income         0.1 -223.3*     0.3** -98.4      Working     20.9*** -56.4     -8.7*** 341.6*     Owns Home     0.0 -11.7     -0.0 2.2 Personal Income             Income       0.7 -247.6**     Gendered Processes Age         -8.7*** 57.0 -5.0*** 139.1 Education         -0.1 13.9 0.3** 7.1 Leisure Men         9.0*** 31.0   Leisure Women         -5.4 -42.0   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 829 39844 829 39844 829 39844 829 21515 829 39844 829 Total % explained 21.6*** -2.7 122.3*** -396.4*** 22.1*** -156.7 1.5** -107.8 -4.6*** 102.6 97*** -737* % unexplained 78.4*** 102.7 -22.3** 496.4** 77.9*** 256.7 98.5*** 207.8 104.6*** -2.6 3 837 P-values are reported for the coefficients. SA- South Asian. * p < 0.05, ** p < 0.01, *** p < 0.001   121  Table 23 Percent Explained in Gender Gap in Time Spent on Shopping among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Filipino All Filipino All Filipino All Filipino English Filipino All Filipino Relative Resources     Income Transfer 20.4** 67.4         -31.5*** 22.4 Time Constraint                  Paid Work   103.3*** 114.9***       108.0*** 326.4***      Leisure (total)   18.4*** 47.2*       19.7*** 173.5* Children   0.1 2.1       -0.5** -4.9 Under 5   0.0 22.6**       0.4*** -9.1 Household Size   -0.0 4.0       -0.1 13.1 Autonomy                  Personal Income         0.1 -12.1     0.3** -47.5      Working     20.9*** 67.6     -8.7*** 21.7     Owns Home     0.0 15.3     -0.0 -82.1 Personal Income             Income       0.7 2.0     Gendered Processes Age         -8.7*** 1.1 -5.0*** -31.5 Education         -0.1 11.5 0.3** 39.0 Leisure Men         9.0*** 36.0*   Leisure Women         -5.4 19.2   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 337 39844 337 39844 337 39844 337 21515 337 39844 337 Total % explained 21.6*** -57.7 122.3*** 171.6*** 22.1*** -5.2 1.5** -70.6 -4.6*** 39.7 97*** 311* % unexplained 78.4*** 157.7 -22.3** -71.6 77.9*** 105.2 98.5*** 170.6 104.6*** 60.3 3 211 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001  122  On the other hand, South Asian Canadians are unique among the cultural groups analyzed here, especially in terms of the gendered division of shopping tasks. Shopping tasks among South Asians are traditionally considered masculine (Hamlett et al. 2008), and accordingly, among South Asian Canadians, men reported more time spent on shopping than women. This finding clearly distinguishes South Asian Canadians as having divergent cultural norms around outdoor routine housework. Overall, unlike for all other cultural groups, the frameworks proposed here are not able to provide satisfactory explanation of the gender gap in shopping tasks among South Asian Canadians because they aim at explaining why women do more housework than men, rather than the other way around. The frameworks cannot account for shopping activities of South Asian Canadians because shopping housework is associated more with men rather than women. This contention is confirmed by the negative sign of the explained part for shopping for all frameworks among South Asian Canadians (see Figure 17), except the gender-centred approach factor, which reports positive percent explained. However, these results for the gender-centred framework factor (years of education) are not on the statistically significant level and might have occurred due to chance.  4.4.1.4 Non-Routine Tasks: Maintenance Figure 18 summarizes the percent of the gender gap in maintenance tasks explained by the main factors within the tested frameworks. It shows that the factors cannot explain satisfactorily the overall gender differences in time spent on maintenance activities. Thus none of the major explanatory frameworks of the gender gap appear to be statistically significant (see also Tables 24-28), which makes the estimates unreliable and may be caused by chance.  123  All in all, however, I can report that most of the gender gap in time spent on maintenance comes from the differences in the gendered processes. In other words, the differences between women and men in how much time they devote to maintenance are due to the gendered character of the task and do not depend on the economic factors. Specific factors of the framework, such as those measured by age and ‘feminine’ leisurely activities, contribute significantly to accounting for the gender gap in maintenance. These results for the decomposition of all analyzed factors are provided in Tables 24-28.  For the total sample the relative resources framework can explain most of the gender differences when we control for all other factors from other frameworks. Likewise, for the Anglo-Canadian subsample, the relative resources framework can provide with about 11.7% of the explanation of the gender gap in the full model (see Table 24). Among French Canadians, the autonomy argument, specifically testing the influence of labour force participation, provides 8% of the explanation to the gender gap in maintenance in the full model (see Table 25). Although in Figure 18 the relative resources framework seems to provide a high share of explanation of the gender gap in maintenance among South Asian Canadians. In fact, if we look at the detailed results in Table 27, the decomposition coefficients are not statistically significant and may be caused by chance. The resulting insignificant estimates, however, can be also due to the small sample size. Overall, the models presented herewith are not suited to test non-routine housework such as maintenance. However, this is not due to the fact of the non-routine character of the task but because of its traditional association more with men’s work because I find similar results for the shopping tasks among South Asian Canadians, where shopping is traditionally more associated 124  with men. Explanations for the traditionally masculine tasks should be sought outside economic factors and the limited gender-centred approach factors presented in the study. It might be that for this type of housework, more cultural and gendered explanations are more suitable. The limitations of the present project in explanatory ability as applied to the gender gap in maintenance may also be the result of the idiosyncrasies of time use data, not capturing non-routine activities more precisely. Yet considering that the frameworks manage to account for most non-daily routine housework such as cleaning and shopping, it is more likely that these results around maintenance tasks show that the tested theoretical frameworks perform well at capturing factors responsible for the housework traditionally associated with women by the merit of distinguishing themselves also from the phenomena that cannot be accounted for by the frameworks, such as housework traditionally associated with men.36  Figure 18 Percent Explained of the Gender Gap in Maintenance by Different Frameworks                                                      36 This ability of the frameworks, therefore, works not unlike the divergent validity of a construct, where we can test the validity by contrasting it with what it is not, with what it must diverge from. 1.8-0.86.9-83.960.4-65.5-35.9 -34.4 -38.2-26.6-50.5-25.9-0.2 -0.3 -0.3-28.810.7-25.30.00.5-0.34.5 2.6-21.3-100.0-80.0-60.0-40.0-20.00.020.040.060.080.0All English French Chinese South Asians FilipinoPercent Explained of the Gender Gap inTime Spent on Maintenance, CanadaRelative resources Time Availability Autonomy Gendered Processes125  Table 24 Percent Explained in Gender Gap in Time Spent on Maintenance among Anglo-Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All English All English All English All English All English All English Relative Resources     Income Transfer 1.8 -0.8         12.3** 11.7* Time Constraint                  Paid Work   -35.9*** -34.4***       -39.1*** -36.6***      Leisure (total)   -9.1*** -8.9***       -9.1*** -11.7*** Children   -0.1 -0.4**       -0.1* 0.7*** Under 5   -0.0 -0.1*       -0.0 -0.4** Household Size   -0.0 -0.0       -0.0 -0.3 Autonomy                  Personal Income    -0.2 -0.3     -0.2* -0.3      Working     -4.3*** -5.1***     5.1*** 2.4     Owns Home     0.0 0.6***     -0.1 0.8*** Personal Income             Income       -0.2 -0.3     Gendered Processes Age         2.6*** 1.5** -0.3 -1.5 Education         -0.0 0.5** 0.1* 0.8** Leisure Men         -8.2 -8.6***   Leisure Women         3.9*** 4.2***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 39844 21515 Total % explained 1.9 -0.4 -45.0*** -43.8*** -4.3*** -4.3** 0.0 0.2 -1.5* -1.9* -31*** -32*** % unexplained 98.1*** 100.4*** 145.0*** 143.8*** 104.3*** 104.3*** 100.0*** 99.8*** 101.5*** 101.9*** 131*** 132*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   126  Table 25 Percent Explained in Gender Gap in Maintenance among French Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All French All French All French All French All French All French Relative Resources     Income Transfer 1.8 6.9         12.3** 13.3 Time Constraint                  Paid Work   -35.9*** -38.2***       -39.1*** -43.2***      Leisure (total)   -9.1*** -9.3***       -9.1*** -11.1*** Children   -0.1 -0.0       -0.1* -0.0 Under 5   -0.0 0.5**       -0.0 0.1 Household Size   -0.0 0.1       -0.0 0.6 Autonomy                  Personal Income    -0.2 -1.8     -0.2* 0.6      Working     -4.3*** -2.3     5.1*** 8.0**     Owns Home     0.0 -0.0     -0.1 -0.3* Personal Income             Income       -0.2 -0.3     Gendered Processes Age         2.6*** 2.9*** -0.3 1.8 Education         -0.0 -0.3 0.1* -0.2 Leisure Men         -8.2 -8.9***   Leisure Women         3.9*** 4.5***   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 39844 10861 Total % explained 1.9 7.3 -45.0*** -46.9*** -4.3*** -4.2 0.0 -0.2 -1.5* -1.6 -31*** -29*** % unexplained 98.1*** 92.7*** 145.0*** 146.9*** 104.3*** 104.2*** 100.0*** 100.2*** 101.5*** 101.6*** 131*** 129*** P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001   127  Table 26 Percent Explained in Gender Gap in Maintenance among Chinese Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Chinese All Chinese All Chinese All Chinese All Chinese All Chinese Relative Resources     Income Transfer 1.8 -83.9         12.3** -102.4 Time Constraint                  Paid Work   -35.9*** -26.6**       -39.1*** -11.2      Leisure (total)   -9.1*** 11.1       -9.1*** -13.6 Children   -0.1 -1.0       -0.1* 1.5 Under 5   -0.0 -1.8       -0.0 0.6 Household Size   -0.0 2.6       -0.0 0.2 Autonomy                  Personal Income    -0.2 -17.4     -0.2* 0.1      Working     -4.3*** -13.9     5.1*** -2.9     Owns Home     0.0 2.3     -0.1 7.6 Personal Income             Income       -0.2 -28.8*     Gendered Processes Age         2.6*** 18.6* -0.3 24.6* Education         -0.0 -4.5 0.1* 5.3 Leisure Men         -8.2 13.5   Leisure Women         3.9*** -2.2   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 961 39844 961 39844 961 39844 961 39844 961 39844 961 Total % explained 1.9 -86.3 -45.0*** -15.5 -4.3*** -26.4 0.0 -26.4 -1.5* 33.2* -31*** -43 % unexplained 98.1*** 186.3* 145.0*** 115.5* 104.3*** 126.4* 100.0*** 126.4* 101.5*** 66.8 131*** 143 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001    128  Table 27 Percent Explained in Gender Gap in Maintenance among South Asian Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All SA All SA All SA All SA English SA All SA Relative Resources     Income Transfer 1.8 60.4         12.3** 58.6 Time Constraint                  Paid Work   -35.9*** -50.5       -39.1*** -80.9      Leisure (total)   -9.1*** -32.7*       -9.1*** -39.9 Children   -0.1 -3.0       -0.1* 1.1 Under 5   -0.0 1.8       -0.0 -1.7 Household Size   -0.0 2.5       -0.0 0.8 Autonomy                  Personal Income         -0.2 9.8     -0.2* 5.4      Working     -4.3*** 8.9     5.1*** 42.3     Owns Home     0.0 -0.7     -0.1 0.1 Personal Income             Income       -0.2 10.7     Gendered Processes Age         2.6*** 4.8 -0.3 27.1 Education         -0.0 2.6 0.1* -2.3 Leisure Men         -8.2 -7.7   Leisure Women         3.9*** -13.4   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 829 39844 829 39844 829 39844 829 21515 829 39844 829 Total % explained 1.9 52.8 -45.0*** -86.0 -4.3*** 14.1 0.0 7.2 -1.5* -16.5 -31*** 26 % unexplained 98.1*** 47.2 145.0*** 186.0 104.3*** 85.9 100.0*** 92.8 101.5*** 116.5 131*** 74 P-values are reported for the coefficients. SA- South Asian. * p < 0.05, ** p < 0.01, *** p < 0.001    129  Table 28 Percent Explained in Gender Gap in Maintenance among Filipino Canadians, 1986-2010, Pooled Oaxaca-Blinder Decomposition  (1) (2) (3) (4) (5) (6)  Relative Resources Time Availability Autonomy Personal Income Gendered Processes Full Model  All Filipino All Filipino All Filipino All Filipino English Filipino All Filipino Relative Resources     Income Transfer 1.8 -65.5         12.3** -164.4 Time Constraint                  Paid Work   -35.9*** -25.9       -39.1*** -4.4      Leisure (total)   -9.1*** -78.0       -9.1*** -97.7 Children   -0.1 0.1       -0.1* 0.1 Under 5   -0.0 -12.6       -0.0 3.4 Household Size   -0.0 -0.1       -0.0 1.1 Autonomy                  Personal Income         -0.2 -9.9     -0.2* 78.7      Working     -4.3*** 14.8     5.1*** -40.2     Owns Home     0.0 -41.4     -0.1 7.4 Personal Income             Income       -0.2 -25.3     Gendered Processes Age         2.6*** -0.5 -0.3 -3.8 Education         -0.0 -21.3 0.1* -61.1 Leisure Men         -8.2 -21.1   Leisure Women         3.9*** -27.9   Control Variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes N 39844 337 39844 337 39844 337 39844 337 21515 337 39844 337 Total % explained 1.9 -110.4 -45.0*** -146.7 -4.3*** -71.4 0.0 -62.3 -1.5* -101.2 -31*** -286 % unexplained 98.1*** 210.4 145.0*** 246.7 104.3*** 171.4 100.0*** 162.3 101.5*** 201.2* 131*** 386 P-values are reported for the coefficients. * p < 0.05, ** p < 0.01, *** p < 0.001  130  4.4.1.5 Gender Gap Decomposition Discussion The economic approach can explain a considerable portion of the gender gap in the household division of labour. Thus the relative resources framework can explain a sizeable share of the gender differences in time spent on cooking and cleaning among all Canadians with the exception of Filipino Canadians. The time availability framework, especially time spent on labour market activities, explains most of the gender gap in time spent on shopping among all Canadians except South Asian Canadians, where the traditional association of shopping activities with men appears to persist.  The implication of this is that leveling opportunities in the labour market allowing women to engage in economic activities as much as men would close the present gender gap in shopping and some of the gender gap in indoor routine housework. These findings regarding a sizeable share that economic factors play in the gender gap come hand in hand with the gender-centred explanation of the division of housework (Bittman et al. 2003; Brines 1994; Greenstein 2000). The gender explanation can explain the observed segregation in individual housework tasks. However, because this approach lays too much emphasis on gender norms outside of the control of an individual or a nation, it runs the risk of shifting unwarrantedly the responsibility away from the public domain and policy makers, potentially stalling the progress toward greater gender equality. Yet the current findings also confirm that gender differentials in economic factors can also account for a big portion of the gender gap in routine housework (approximately a third in indoor tasks and all of the gender gap in shopping) can still be attributed to gender discrimination and other unobserved variation particularly among South Asian Canadians. Because the economic factors are still significant in accounting for the gender gap in time spent 131  on routine housework, these results could be useful for policy makers by showing how much can be achieved by equalizing the opportunities of women to those of men on the labour market. Nevertheless, the shortest answer to the question posed in the present chapter about whether there is a cultural gap in housework is no. In all cultural groups, most of the routine indoor work is done by women. There is little difference between Canadians in how wide the gender gap is in the select domestic tasks and in the factors that explain the gap, with a few minor exceptions. Even though the cultural perspective might to some extent provide more explanation in non-routine housework such as maintenance or in cooking if we assume that main cultural factors concerning gender attitudes were not adequately captured by the models in the present project, the economic exchange factors are enough to explain most of the gender differences in shopping and cleaning, especially among those who report doing any housework activities on the diary day. Moreover, factors that can explain most of the gender gap are quite similar for all cultural groups analyzed in the present Chapter.  One policy implication of these findings is that most of the gender gap might be ameliorated through policy actions regarding the leveling out of the labour market opportunities for women and men without discrimination by ethnic origin. For instance, because the time availability framework can explain most of the gender gap in shopping activities, the gender gap in shopping time would completely close in most of the major Canadian cultural groups if hypothetically men and women were to spend similar periods of time on economic activities. The pertinence of the relative resources argument in cooking and cleaning also suggests that there might still be power relations at work that disadvantage the partners with lower contribution to the household income 132  and pressure them to contribute more of their time on routine indoor activities. This consequence applies equally to all Canadians. A few limitations are necessary to discuss with regard to the results. Due to constraints imposed by the idiosyncrasy of the time diary data, particularly in the distribution of the activity variables, decomposition analysis relies heavily on the samples who report housework on the diary day, yet it does not mean that respondents do not perform any housework on other days than the diary day. Such limitation of the time diaries favours routine activities that are more likely to happen daily such as cooking while other activities which do not always happen everyday but are still routine, such as cleaning and shopping, might not reflect as accurately as results for daily routine housework. This limitation can be overcome by focusing on the samples that report doing any housework on the diary day especially for the non-daily routine housework activities with sampling bias correction. The future research with time use data could adjust for the selection bias in a way that could account for the idiosyncrasies of those who report doing housework and those who do not in a more thorough fashion. It would be useful because men who report doing housework might be different from men who did not report any housework.  Another, more technically challenging, strategy would be to use imputation methods for men who did not report doing housework on the diary day because it might be a weekday rather than a Saturday where those same men might have reported more housework than they did on the diary day. Such imputation would not only require a rigorous machine learning model but also more observations and features (variables) to be able to accurately predict time potentially spent on housework on other days. Other directions future research could take, are in the development and adoption of more sophisticated measures for class to probe if economic and leisure activities 133  could be further fleshed out. This direction is especially interesting to analyze leisure activities in more detail such as screen time, which seems to have a great impact on the gender gap in housework in Canada. It might well be that it is the lifestyle differences that dictate the time allocation to certain leisurely activities over housework.     134  Chapter 5: Gendered Division of Labour: Relative Resources or Gender-centred Approach? 5.1 Routine Indoor Housework and Relative Resources In this chapter, I analyze gender differences in more detail than the decomposition method would allow, bringing the analysis back toward the more conventional approach in the field of domestic labour. This chapter investigates the predictive power of the relative resources framework for time spent on the four housework tasks with the focus on routine indoor tasks. The framing of this chapter is guided by the results of the gender gap decomposition. The decomposition analysis showed that tested frameworks are more reliable for routine housework rather than non-routine chores such as maintenance activities. Furthermore, the relative resources approach showed its superior predictive power in the more gendered housework such as the routine indoor tasks. As a consequence, the analysis in this chapter will pay more attention to how much of the time spent on housework can be predicted by the relative resources of partners and whether the predictions go in the direction laid out by the expectations of the economic approach. The previous chapter established that there is still a considerable gap between men and women in housework. However, the progress over the last few decades reveals the overall trend toward a more equal division of labour within the Canadian household. Results pointing at the gradual convergence are consistent with analogous studies conducted in Canada by Marshall (2011) and in other Western countries (Chesters 2011; Gimenez-Nadal and Sevilla 2012; Hook 2010; Kan et al. 2011). Chapter 4 also highlighted the frameworks which are more capable of explaining the gender differences in routine housework. The relative resources framework can 135  explain the greatest share of routine indoor housework with minor cultural nuances; whereas the time availability framework is the most appropriate choice when accounting for the gender gap in routine outdoor tasks. One limitation of the decomposition analysis is that it does not allow for any conclusion on what the exact association between tested factors and the participation of women and men in housework is. To overcome this limitation, the present Chapter examines involvement in housework activities based on the factors that are revealed to be able to account for the gender gap.  Therefore, in the following sections, I address the mechanisms behind men’s and women’s participation in routine housework in more detail analyzing the Canadian GSS, 1986-2010, and comparing the results against ATUS, 2003-2015. In this intellectual pursuit, I try to answer the following questions: whether the time spent on housework in Canada is accounted for by tested factors in the way predicted by the theoretical models? The mechanisms, lying behind these associations are analyzed within a unified framework following Brines (1994) using my own extensions presented in the theoretical part (Chapter 2). Overall, the present chapter establishes that Canadian women still do gender in most of the routine tasks, whereas Canadian men are in fact showing a pattern inconsistent with the prevalent gender norms – they ‘undo’ gender in all routine housework.  5.1.1 Results Testing Specifications on Aggregate Housework Table 29 summarizes the results testing linear, quadratic, and cubic specifications of the association between relative resources and logged time spent on housework among all Canadian women and men between 1986 and 2010. All coefficients in the Table are estimated with the Heckman correction and the robust standard errors are reported to adjust for heteroscedasticity 136  for the three theoretical models. Models with cubic terms test the cumulative disadvantage specification within the bargaining approach (see Equation (2)), models with the quadratic term test the gender processes explanation (see Equation (3)), and models with only linear terms (see Equation (4)) in each section of the table test the simple linear form of the bargaining perspective.  ln(𝑇) =  𝛽1∆ + 𝛽2∆2 +  𝛽3∆3 + 𝐵𝑋 +  𝜀                                      (2) ln(𝑇) =  𝛽1∆ + 𝛽2∆2 + 𝐵𝑋 +  𝜀                                              (3) ln(𝑇) =  𝛽1∆ + 𝐵𝑋 +  𝜀                                                   (4), where T – time spent on a housework task; 𝛽1 – are the coefficients in the Column 4 of Table 29, representing the robust regression coefficients with the Heckman adjustment for the linear term of the income transfer variable; 𝛽2 – are the robust regression coefficients with the Heckman adjustment for the quadratic term (Column 3 in Table 29) of the income transfer variable; 𝛽3 – are the robust regression coefficients with the Heckman adjustment for the cubic term (Column 2 in Table 29) of the income transfer variable; ∆ - the income transfer variable; B – regression coefficients for control variables and the constant; and X – the design matrix without the income transfer variable. While both linear and cubic models are consistent with the premises of the bargaining perspective, the latter attempts to model the consequences of cumulative disadvantage, when the effects on the participation in housework for the sole breadwinners or fully dependent individuals are assumed to be more extreme than for dual earners, reflecting in more precipitous change on the edges of the cubic curve. The model with the quadratic term, on the other hand, 137  represents the functional form of the effect of the relative resources under the gender processes explanation, such that it goes contradictory to the predictions of the economic approach.  Table 29 Models on Logged Time Spent on Housework for Canadian Women and Men, 1986-2010  Income Transfer3 (Δ3) Income Transfer2 (Δ2) Income Transfer (Δ) ρ Log likelihood -2 log (L1/L2) (df)  Cooking Women -0.112 (0.058) 0.094** (0.031) 0.031 (0.048) -0.921*** -18566.067 5.51(1) *    0.110*** (0.030) -0.053* (0.021) -0.921*** -18568.820 20.91(1) ***     -0.080*** (0.019) -0.922*** -18579.275   Cooking Men -0.080 (0.097) 0.195** (0.066) -0.027 (0.073) -0.888*** -19899.054 1.10(1)    0.163** (0.055) -0.074 (0.044) -0.888*** -19899.603 14.53(1) ***     0.013 (0.034) -0.888*** -19906.866   Cleaning Women -0.008 (0.086) 0.140** (0.047) 0.037 (0.071) -0.924*** -19144.799 0.01(1)    0.141** (0.044) 0.031 (0.032) -0.924*** -19144.806 15.73(1) ***     -0.008 (0.029) -0.925*** -19152.671   Cleaning Men -0.041 (0.147) 0.107 (0.093) 0.145 (0.120) -0.861*** -11606.622 0.12(1)    0.093 (0.081) 0.119 (0.067) -0.861*** -11606.684 2.22(1)     0.163** (0.055) -0.861*** -11607.792   Shopping Women -0.019 (0.092) 0.127** (0.049) 0.062 (0.074) -0.927*** -16742.899 0.06(1)    0.130** (0.047) 0.048 (0.033) -0.927*** -16742.929 11.44(1) ***     0.015 (0.030) -0.928*** -16748.649   138   Income Transfer3 (Δ3) Income Transfer2 (Δ2) Income Transfer (Δ) ρ Log likelihood -2 log (L1/L2) (df)  Shopping Men -0.227 (0.125) 0.163 (0.085) 0.183 (0.085) -0.926*** -16349.428 5.33(1) *    0.068 (0.069) 0.049 (0.055) -0.926*** -16352.093 1.60(1)     0.087* (0.042) -0.926*** -16352.893   Maintenance Women 0.305 (0.378) 0.264 (0.224) -0.139 (0.294) -0.945*** -3045.735 0.82(1)    0.201 (0.203) 0.069 (0.139) -0.944*** -3046.144 1.27(1)     0.001 (0.121) -0.946*** -3046.781   Maintenance Men -0.137 (0.294) 0.327 (0.229) -0.168 (0.195) -0.969*** -8088.658 0.42(1)    0.258 (0.156) -0.239 (0.136) -0.969*** -8088.869 4.90(1)     -0.079 (0.087) -0.969*** -8091.319   Unstandardized coefficients; Robust standard errors in parentheses. * p < .05, ** p < .01, *** p < .001. All models control for being born in Canada, age, education, having children, having children under 5, household size, weekday, ethnic origin, unemployment rate, marriage per 1000, female employment rate, years, and provinces.  The results for Canadian men differ from those found for American men based on the study of Brines (1994). Canadian married and cohabiting men are ‘undoing’ gender in cooking, and with some reservations in cleaning and shopping. The gender-centred model results show the positive sign for the quadratic term in all cases (B=0.163, p<.01 for cooking, B=0.093, p=0.247 for cleaning, and B=0.068, p=0.322 for shopping), while adding the cubic term does not improve the model fit significantly in all cases, except in shopping but in shopping, the cubic specification looks like the quadratic (gender) specification. The addition of the quadratic term improves the 139  model fit only for testing the association between relative resources and time spent on cooking. The results in Table 29 confirm that men who make less money than their partners are more likely to cook, just like women in the same situation. However, it is the positive linear model that provides sufficient model fit for the explanation of the logged time of men spent on cleaning, emphasizing that men increase their participation in cleaning housework with the increase in their relative resources. These results negate the expectations laid out by the economic framework and suggest that other mechanism than bargaining are at work for the explanation of the men’s participation in cleaning.  I find that men who earn more than their partners actually also contribute to routine housework more than men who earn as much as their partners, contradicting the findings of Brines (1994) among American men.37 However, this finding can also reflect the effect of a historical change since 1985, when the data in Brines’ study was collected, and the existence of cultural differences between the US and Canada. Yet the Canadian data for 1986 show the same results as the consequent years in Canada, albeit not statistically significant in 1986. These findings suggest the ‘undoing’ gender explanation for the involvement of Canadian men in housework in 1986 as well as in the consequent years.  On the other hand, I find that Canadian women still conform with the gender norms and ‘do’ gender in all routine housework, particularly in cleaning and shopping, which can be confirmed by model specifications using the quadratic term. The specifications provide a sufficient fit for the data and adding of the cubic term does not improve the model fit significantly. All quadratic specifications of models for routine housework among Canadian women report positive                                                    37 The findings of the present study still hold when controlled for personal income or time spent on paid work in the substantive model (see 3.1.6. on the Heckman adjustment explanation) as well.  140  coefficients for the respective quadratic terms (B=8.277, p<0.001 for cooking, B=16.685, p<0.001 for cleaning, and B=15.858, p<0.001 for shopping). These results are consistent with the findings among American women by Greenstein (2000), who conducted his study using the National Survey of Families and Households. 5.1.2 Results for Individual Housework Tasks 5.1.2.1 Cooking Table 29 summarizes the results for the regression coefficients with Heckman correction against time spent on the four types of housework including cooking. The analysis employs the sample of all Canadian married and cohabiting women and men, controlling for age, education, having children, having children under 5, household size, weekday, ethnic origin, being born Canadian, provincial unemployment rate, marriage per 1000 residents, female employment rate, years, and provinces. Likelihood-ratio tests comparing models (last column of the table) allow me to conclude that the cubic model improves the model fit for explaining the association between the relative economic contribution and logged time spent on cooking for women over the quadratic and linear specifications. In particular, the addition of the quadratic term improves the model fit significantly over the model with the linear term only (χ2 (1)= 20.91, p <0.001) and the cubic specification improves the model fit over the quadratic term model (χ2 (1)= 5.506, p <0.05).  The more women earn compared to their partners, the less they cook, as it would appear from the cumulative disadvantage model that the cubic specification stands for. The cubic predictions for the time spent on cooking are plotted, based on the results of the regression analysis, in the graph in the left-most pane in Figure 14. The graph shows that even though it is the cubic curve, 141  it resembles the quadratic specification, except for the breadwinning women where the curve reverses its direction a little. Thus, the economic approach is capable of predicting cooking time for women who earn less than their partners. However, as soon as a woman starts earning more than her partner, the expectation is that she will start compensating for the transgression of the traditional gender norms and cook more often. Yet if she starts earning closer to 100% of the family income, she would spend a little less time on cooking. There are no significant differences between cultural groups in Canada when Anglo-Canadians are used as the reference group. The only difference that is worth noting is that South Asian Canadian women spend significantly more time on cooking than Anglo-Canadian women (see Table 31). For men, a similar pattern emerges out of the GSS data. The gender-centred model provides a sufficient fit for the association between logged time spent on cooking and the relative economic contribution. Thus, technically speaking, only the addition of the quadratic term improves the model fit (χ2 (1)= 14.526, p <.001) over the linear model, whereas adding the cubic term does not improve the model fit significantly over the quadratic model (χ2 (1)= 1.098, p = 0.295). The cubic model specification for women (see Figure 19) and quadratic specification for men (see Figure 20) both fall within the gender-centred explanation.  142   Figure 19 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cooking (Y), Canadian Women  Figure 20 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cooking (Y), Canadian Men Yet the explanations for the observed patterns differ. Therefore, while Canadian women are ‘doing’ gender in cooking, except the dependent and sole breadwinning women, the Canadian men show patterns of behaviour that are opposite to the normative gender display and represent an instance of ‘undoing’ gender. Thus dependent and breadwinning men cook more often than men who earn as much as their partners. 143  With the purpose of comparing to results among American women and men, Table 30 summarizes the regression coefficients for the relative economic contribution variables with the Heckman correction against the logged time spent on cooking for the US sample. Likelihood-ratio tests comparing models demonstrate that the cubic model for American women provides the best fit for explaining the association between the relative economic contribution and logged time spent on cooking for women. Thus, the cumulative disadvantage model does not improve the model fit with the addition of the cubic term over the gender display model (χ2 (1)= 1.04, p =.308), whereas the addition of the quadratic term does (χ2 (1)= 12, p < .001). Yet similarly with the Canadian results, the cubic specification resembles quadratic except for the sole breadwinning women (see Figure 21). Thus, the best explanation for the women’s participation in cooking is provided by the gender-centred perspective. Both dependent women and women who earn more than their partners cook more compared to women who earn as much as their partners, suggesting that the balance of power is essential for disencumbering women and for reaching more egalitarian division of cooking tasks. For the American men, the gender display model provides a sufficient fit for the association between logged time spent on cooking and the relative economic contribution. Thus, the addition of the quadratic term improves the model fit (χ2 (1)= 16.7, p < .001) over the linear model, whereas adding the cubic term does not (χ2 (1)= 0.18, p >.05) improve the model fit significantly over the quadratic model. The gender display of the men’s participation in cooking is, however, not associated with the traditional masculine gender display reported by Brines (1994). In fact, the pattern that American men, just like Canadian men, is similar to that associated with the gender performance of women, allowing me to conclude that American men are also ‘undoing’ 144  gender in cooking. Like dependent women, dependent American men are devoting more of their time to cooking than men who earn as much as their partners. Likewise, breadwinning men also cook more often than men who earn as much as their partners. The predicted models for women and men can be graphically represented as in Figure 21, where the patterns of women and men are similar. Yet it is worth noting that the absolute time spent on cooking for an average woman still significantly exceeds that of an average man both in Canada and in the US.   Figure 21 Predicted Association between Relative Economic Contribution and Logged Time Spent on Cooking, American Women (Upper) and American Men (Bottom) There are a few possible explanations for this gender transgression of North American men. First, because higher earning men in general are more concerned about food (Szabo 2013, 2014), 145  they tend to invest their time into cooking more than men who earn less. Second, because men who earn more might also have more free time on their hands, therefore, freeing that time for health and leisurely pursuits such as cooking. The health explanation is intertwined with the class and lifestyle dietary differences explanation (Darmon and Drewnowski 2008; James et al. 1997). The lifestyle explanation suggests that for men, cooking is still considered a leisure activity rather than housework (van Hooff 2011; Lupton 2000; Neuman, Gottzén, and Fjellström 2015).  The leisure explanation for men’s cooking is sometimes referred to as the ‘masculine’ explanation (Szabo 2014), therefore, breadwinning men actually provide a neutralizing explanation for their non-normative gender behaviour as opposed to the neutralized behaviour common to ‘doing’ gender. Additionally, data show that American men (and Canadian men) in the dependent positions do not shy away from cooking. This is similar to the situation with Nordic countries specifically in the Scandinavian region (Neuman et al. 2015). In Canada, the ‘undoing’ gender explanation for men’s participation in cooking as work activity originated from qualitative research (Szabo 2013, 2014). In the interviews with men living in the Toronto area, Szabo (2014) concludes that these men see “their cooking as a way of nurturing others, expressing love and care, or creating ‘home’, responsibilities primarily attributed to women”.  These results, however, should be used with caution because the main part of men ‘undoing’ gender comes from men who are in the dependent positions doing more housework (see Figure 20) rather than from breadwinning men, where the involvement in cooking is more equal for all breadwinning men. This similarity of women’s and men’s patterns of association of time spent on cooking and relative resources indicates that we might be looking at the gender convergence of factors that define the participation in this type of housework.  146   Table 30 Year Fixed Effects Models on Log of Time Spent on Housework for American Women and Men, 2003-2015  Income Transfer3 (Δ3) Income Transfer2 (Δ2) Income Transfer (Δ) ρ Log likelihood -2log(L1 /L2) (df)  Cooking Women -0.051 (0.068) 0.082* (0.033) 0.015 (0.057) -.947*** -23902.57 1.04(1)   0.084* (0.033) -0.025 (0.022) -.947*** -23903.09 12(1) ***    -0.027 (0.022) -.947*** -23909.09   Cooking Men -0.030 (0.101) 0.128** (0.044) -0.012 (0.093) -.938*** -26724.59 0.18(1)   0.125** (0.043) -0.039 (0.027) -.938*** -26724.68 16.7(1) ***    -0.031 (0.027) -.936*** -26733.03   Cleaning Women -0.090 (0.103) 0.122* (0.052) 0.124 (0.086) -0.895*** -21451.17 1.42(1)   0.129* (0.051) 0.055 (0.034) -0.895*** -21451.88 12.18(1) ***    0.048 (0.033) -0.898*** -21457.97   Cleaning Men -0.313 (0.172) 0.208** (0.078) 0.306 (0.157) -0.798*** -15907.16 7.12(1) **   0.182* (0.075) 0.036 (0.046) -0.796*** -15910.72 12.32(1) ***    0.045 (0.045) -0.772*** -15916.88   Shopping Women -0.296*** (0.090) 0.012 (0.046) 0.234** (0.074) -0.941*** -20544.38 20.58(1) ***   0.025 (0.045) 0.003 (0.029) -0.941*** -20554.67 0.62(1)     0.003 (0.029) -0.941*** -20554.98    Income Income Income ρ Log -2log(L1 /L2)  147  Transfer3 (Δ3) Transfer2 (Δ2) Transfer (Δ) likelihood (df) Shopping Men -0.060 (0.104) -0.022 (0.044) 0.077 (0.097) -0.907*** -25177.34 0.66(1)   -0.027 (0.043) 0.023 (0.026) -0.907*** -25177.67 0.76(1)    0.021 (0.026) -0.907*** -25178.05   Maintenance Women -0.584 (0.515) -0.094 (0.361) 0.469 (0.556) -0.712*** -2356.387 1.90(1)   -0.004 (0.433) 0.021 (0.525) -0.719*** -2357.339 0.00(1)    0.020 (0.459) -0.717*** -2357.34   Maintenance Men 0.039 (0.315) -0.252 (0.134) -0.004 (0.291) -0.947*** -8480.562 0.03(1)   -0.249 (0.131) 0.031 (0.084) -0.947*** -8480.578 7.39(1) ***    0.018 (0.086) -0.947*** -8484.275   Unstandardized coefficients; Robust standard errors in parentheses * p < .05, ** p < .01, *** p < .001. All models control for being born in the US, age, education, having children, having children under 5, household size, being Black, Native American, Asian, or other non-White, weekday, years, and state level variables. The likelihood ratio test χ2 for linear model compares to the model with controls only.   5.1.2.2 Cleaning In Table 29, the results show that among the theoretical frameworks testing the association between relative resources and time spent on housework, it is the gender-centred approach that can also explain women’s participation in cleaning. However, the linear bargaining perspective is enough for explaining the association between the relative economic contribution of men and their participation in cleaning tasks. For women, the model fit improves with the addition of the quadratic term (χ2 (1)= 15.73, p < 0.001), while it does not with the inclusion of the cubic term 148  (χ2 (1)= 0.014, p = 0.906). This provides evidence in favor of the gender-centred explanation over the bargaining approach in the association of women’s relative economic contribution and the logged time spent on cleaning. For men, neither quadratic nor the cubic term improve the model fit significantly over the model with only income transfer (χ2 (1)= 2.216, p = 0.137 for the quadratic specification over the linear one and χ2 (1)= 0.124, p = 0.725 for the cubic specification over the quadratic one). Thus, the best fit for men’s participation in cleaning is provided by the linear variation of the model specifications.  Figure 22 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), Canadian Women 149   Figure 23 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), Canadian Men Yet the linear form in the results for the association of relative resources with the participation of men in cleaning is different from that expected under the economic framework and bargaining perspective because the coefficient for the relative resources measure is positive. This positively increasing line is graphically represented in the right-most pane of the Figure 23. If these results were reflecting the involvement of women in cleaning, it would be possible to interpret that with the increase in the share women provide to the household income, they also increase their time spent on cleaning to compensate for the gender norm violation in the labour market. However, this cannot be applied to the involvement of men in cleaning. One possible explanation lies in class and lifestyle differences rather than gender neutralization.  Men who earn more money than their partners might also in general make more money than average, thus having different lifestyles than the average Canadian men who earn as much as or less than their partners. Yet the fact that this pattern of association of breadwinning men’s relative resources and their time spent on housework does not change when controlled for 150  absolute resources (personal income) together with the fact that the effects of education are small and negative (Tables 31-34) cast doubt on the class explanation of the observed phenomenon. The ability to outsource appears to have very little effect on the explanation of cleaning among breadwinning men and women in Canada because both breadwinning men and women seem to spend more time on housework than those who earn as much as their partners. This result is similar to findings on limited consequences of outsourcing for the housework in France (Windebank 2007), the US (Killewald 2011), and the UK (Sullivan and Gershuny 2013). Here the resource-based explanation of the power of the partners with more resources to buy themselves out of housework does not apply.  As to Canadian women, since research shows that breadwinner-dependent couples suffer higher likelihood of divorce than dual-earner couples (Cooke 2006), divorce by itself might be a feared prospect that drives breadwinning women to perform more of cleaning tasks at home. On the other hand, the dependent women represent a demographic group where the outsourcing explanation might hold (Craig and Baxter 2016; Gupta 2006b). Women who are in the dependent position and contribute less than a half of what their partners do but whose household income is CAD 150,000 or over spend 79 minutes on an average day cleaning, while women in the same position but whose household income is CAD 35,000, spend about 94 minutes on the same task. Thus higher income households with dependent women might spend some of their income on outsourcing cleaning tasks but the difference of 15 minutes a day is too small to come to any definitive conclusions. Cleaning is not only likely to be outsourced by wealthy households to the outside workers, but also to household children in households with older children. Thus, the presence of children is significantly decreased in models exploring the cleaning activities for women with children of 5 years of age or older, while it increases participation for women’s time 151  spent on cooking (see Tables 31 and 32). However, no sufficient differences were found in participation in cleaning tasks based on cultural group belonging both among women and men. Figures 22 and 23 represent graphically the association between Canadian women’s and men’s participation in cleaning and their relative economic contribution. The cubic and quadratic specifications for women’s participation in cleaning look identical and both coincide with the predictions of the gender-centred approach. The pattern for breadwinning women who do more housework than women who earn as much as their partners is consistent with the normative gender display explanation, where breadwinner women try to neutralize (Sykes and Matza 1957) their non-normative gender behaviour in the labour market.   Figure 24 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), American Women 152   Figure 25 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Cleaning (Y), American Men To compare the Canadian sample with the ATUS sample, in Table 30, the results for the American women and men are presented and the association between relative economic contribution and their participation in cleaning tasks. For American women, the best fit is provided by the quadratic model specification, similar to the results among Canadian women. The addition of the cubic term does not improve the model fit over the model with the quadratic term (χ2 (1)= 1.42, p =.233). Graphically the curve representing the association between relative resources and time spent on the housework resembles the quadratic specification (see Figure 24) because the local maximum is on the edge of the domain of the income transfer variable (close to 1). Adding the quadratic term improves significantly the model fit over the model with just income transfer (χ2 (1)= 12.18, p < .001) and the quadratic term of the income transfer is significant  (B=0.129, p<.05). Thus the results for American women are similar to Canadian – they also ‘do’ gender in cleaning. Thus, the ATUS data show that women who contribute more to the family budget are also involved in more cleaning.  153  This finding casts doubt on the outsourcing explanation of time spent on housework among breadwinning women in cleaning tasks (Gupta 2006a) because women who earn more money than their partners for their families, on average, also devote more of their time on cleaning. The average income for women who provide for more than 25% of the family income compared to their partners is around USD$ 43,469 per annum, whereas for women whose partners contribute for more than 25% to the family income than the women, the average is around USD$ 5,858. Therefore, on average, breadwinning women in Canada also have higher absolute income. Thus the ability to outsource has likely a very little effect on the explanation of cleaning among breadwinning women in the US. As to the American men, the best explanation is provided by the cubic specification. Thus, the addition of the cubic term improves the model fit significantly over the model with the quadratic term (χ2 (1)= 7.12, p =.008), while the coefficient for the cubic term of income transfer is not significant (B=-0.313, p>.05). Yet the association overall represents more similarities with the pattern displayed by women in cleaning tasks, except for the highest group of breadwinning men (see Figure 25). Thus American men except those who earn much more than their partners tend to “undo” gender similarly to their participation in cooking. This is a little different from what is observed among Canadian men – Canadian men who earn less than their partners do less of housework than men who earn more (represented by the positive linear association). Thus for dependent men the pattern of involvement in the cleaning tasks is that of ‘doing’ gender, while men ‘undo’ gender if they earn more than their partners. 5.1.2.3 Shopping and Maintenance Together with the results for the association of relative resources and time spent on cooking and cleaning, Table 29 also summarizes results for models for Canadian women’s and men’s 154  time spent on other household tasks, namely shopping and maintenance. These models include all the demographic and control variables. Similar to cleaning tasks, women ‘do’ gender in shopping but neither the bargaining nor the gender display perspective were able to provide enough evidence to explain the participation of women in maintenance. Thus the quadratic specification improves the model fit over the model with the quadratic term in shopping for women (χ2 (1)= 11.44, p < .001), while adding the cubic term does not improve the model fit (χ2 (1)= 0.06, p =0.807). Figure 26 summarizes how the association between relative resources and women’s time spent on shopping would look like for each of the specifications. The cubic and quadratic specifications for the shopping tasks look identical and represent the pattern characteristic for women ‘doing’ gender. Thus not only dependent women do more shopping but also breadwinning women spend more time on shopping to compensate for their non-normative gender behaviour in the labour market. Despite the fact that shopping is one of the most egalitarian tasks, the results show that even here the power imbalances are at work and women are pressured to conform to societal gender norms, especially Canadian women of Filipino descent, who seem to involve in the shopping activities significantly more than the reference Anglo-Canadian group (B=0.402, p<.05) (see Table 33). As for Canadian men, they ‘undo’ gender in shopping and they bargain in maintenance tasks. Among all Canadian men, French Canadian men seem to spend significantly more time on shopping than the majority represented by the Anglo-Canadian men (see Table 33). In particular, the cubic specification can explain the time spent by men on shopping and the quadratic – on maintenance. The quadratic specification does not improve the model fit over the model with the linear term for explaining time spent on shopping among men (χ2 (1)= 1.6, p = 0.206). The cubic term the model fit improves significantly (χ2 (1)= 5.33, p <0.05). Yet if the association between 155  relative resources and men’s time spent on shopping is represented graphically then the patterns are more indicative of men’s ‘undoing’ gender (see Figure 27). Dependent men spend more time on shopping and breadwinning men also do more shopping compared to men who earn as much as their partners. The left-most pane on Figure 27 that represents the cubic specification indicates that sole breadwinning men actually show a bit of a reversal where they start spending less time on shopping.  Canadian men bargain in maintenance tasks. For the time spent by men on maintenance tasks, it is the quadratic term that improves the model fit (χ2 (1)= 4.9, p <0.05) but not the cubic term (χ2 (1)= 0.422, p =0.516). However, Figure 28, that summarizes the association between relative resources and men’s time spent on maintenance, confirms that only dependent men perform more maintenance while for dual earners and breadwinning men the expectation of spending time on maintenance is generally low. This finding allows me to conclude that it is bargaining that is at work for men participating in maintenance which is also due to the character of this type of housework – consistent with the premises of men ‘doing’ gender. Men who contribute less to the economic resources of the household devote more of their time to the maintenance tasks – asserting their masculinity in the traditionally masculine housework to neutralize for gender norm violation in the labour market. Conversely, men who contribute more to the household income are less likely to take on the maintenance tasks in a household because men who perform according to the expectations of the gender norms do not need to compensate otherwise. Moreover, even though there are no discernable differences among women based on cultural group belonging, Canadian men of Chinese (B=1.048, p<.05), South Asian (B=1.381, p<.01), and Filipino (B=1.375, p<.01) descent are more likely to spend time on maintenance 156  tasks than the Anglo-Canadian men, suggesting a significant emphasis on masculine tasks among Asian Canadians.  Figure 26 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), Canadian Women  Figure 27 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), Canadian Men As for the American women and men, Table 30 presents the summary of the results for models associating the relative economic contribution and time spent on other household tasks such as shopping and maintenance. These models control for all demographic variables, year and state level factors. Neither the bargaining nor the gender-centred perspective were able to 157  provide enough evidence to explain the participation of American women in maintenance and of American men in shopping. As for the latter tasks, it is the cubic specification that provides the best fit for women’s time spent on shopping Among models analyzing women’s time, adding the quadratic term does not improve the model fit for shopping (χ2 (1)= 0.62, p=.431) and the cubic term improves the model fit significantly (χ2 (1)= 20.58, p<.001) (see Table 30). Yet the predicted pattern of the cubic curve (see Figure 29) resembles that for ‘doing’ gender until the point where it reverses for the breadwinning women who contribute 50% more than their partners. Thus in shopping, both Canadian and American women ‘do’ gender. For American men, neither quadratic nor the cubic specification improve the model fit significantly. If we look at the graph of the specification, however, (see Figure 30) it is clear that this pattern is similar to that of Canadian men’s participation in cleaning – American men increase their time spent on shopping with more relative contribution they provide to their households. This finding is contrary to the expectations of the economic approach and suggests possible class differences in shopping activities among American men. It is distinct from the dependent Canadian men who do more shopping with the increase of their dependency but similar to breadwinning Canadian men (see Figure 27). Adding the quadratic term improves the model fit significantly for the association between the relative economic contribution and men’s time spent on maintenance. However, for maintenance, the explanation for the participation of American women lies in factors unrelated to relative economic contribution of the partners and is an expected outcome because the frameworks discussed here are found to be more appropriate to explaining why women do more housework than men rather than men’s participation in male-dominated housework tasks such as maintenance or shopping among South Asian Canadians. 158   Figure 28 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Maintenance (Y), Canadian Men  Figure 29 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), American Women 159   Figure 30 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Shopping (Y), American Men  Figure 31 Predicted Association between Relative Economic Contribution (X) and Logged Time Spent on Maintenance (Y), American Men 5.2 Discussions and Chapter Conclusions The present chapter explored women’s and men’s participation in individual housework tasks in its association with relative resources. The trends in domestic work discussed in Chapters 3 and 4 reveal distinct patterns by the three major routine housework types. Thus, over the period of 1986-2010, in the routine indoor tasks such as cooking and cleaning, women, on average, 160  reduced their involvement. In contrast to women, men showed an increase in all routine indoor domestic tasks involvement. The two main explanations were tested for the observed patterns: the bargaining and the gender-centred perspectives. Using the theoretical framework proposed by Brines (1994) and with the extensions discussed in Chapter 2, I find that there are considerable similarities in the patterns of women’s and men’s participation by individual housework tasks. Furthermore, employing year and province fixed-effects models with the Heckman correction for selection bias, I find that for cooking (and to some extent, shopping and cleaning), men produce a new pattern of gendered behaviour in Canada – they are ‘undoing’ gender. Thus men who contribute a greater share of the household income spend more time on routine housework than men whose economic contribution is on the similar level to their partners’.  In contrast, women in Canada “do” gender in all routine housework, consistent with the findings of Greenstein (2000) for American women. To summarize, Brines’ (1994) ideas and the “doing” gender perspective are more applicable to women in Canada and specifically to their involvement in routine housework tasks rather than men. Brines’ (1994) framework provides a strong theoretical idea to testing the power relationships in national households, considering that the relative resources framework is able to capture most of the gender gap in indoor routine housework. This chapter accounts for the possibility that at least two main extensions of the present theoretical stance in the analysis of the gendered division of unpaid labour. First, the gender-centred perspective can be extended to the patterns uncharacteristic of the normative gender expectations as the “undoing” gender explanation (Deutsch 2007), as portrayed by the findings of this chapter in cooking and shopping among Canadian men. Second, the theoretical framework should account for diversity in explanations for different groups by relative economic 161  contribution: breadwinners, equal-earners, and dependents, as suggested by the findings in the present Chapter. For instance, the slight change in direction for the association of relative resources (1) with time spent on cooking among principal breadwinning Canadian women (see Figure 19) and (2) with time spent on shopping among principal breadwinning Canadian men (see Figure 27) indicate the necessity to look at the sole breadwinners as a category of their own. Thus what applies to principal breadwinners might not apply to dependent individuals. For instance, Canadian women who are the main breadwinners in a household do gender in routine housework tasks to compensate for the gender norm violation in the labour market, but women who depend on their partners do not, since their pattern is consistent with the expectations of the economic framework. Their time commitment to the task increases with the increase of their dependency on their partners as the economic approach predicts.  There are a few directions that the present research should explore in the future. First, the theoretical extension of Brines (1994) to include (1) the ‘undoing’ gender perspective with regard to routine and non-routine tasks and consequently, (2) the analysis of the gender norms regarding the participation in certain housework tasks, which I propose in Chapter 2, contributes to the observed patterns among men doing routine housework in Canada and is confirmed by the same observed pattern among American men doing routine indoor housework. Furthermore, future research could benefit from accounting for distinct factors explaining power-dynamics in dual-earner couples compared to breadwinner-dependent couples. That is, the findings of the present Chapter indicate that women tend to spend the least amount of time on housework when their resources match those of their partners. Egalitarianism in the conjugal power in terms of economic dependency breeds equality in time spent on housework.   162  Table 31 Year and Province Fixed-Effects Models on Log of Time Spent Cooking for Women and Men with Heckman Adjustment, 1986-2010  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) -.080*** .031 -.053*  -.013 -.027 -.074 (.019) (.048) (.021)  (.034) (.073) (.044) Δ2  .094 .110***   .195** .163**   (.031) (.030)   (.066) (.055) Δ3  -.111    -.080    (.058)    (.097)  Control Variables             Born in Canada -.114*** -.108** -.110***  -.154** -.147** -.147**  (.031) (.031) (.031)  (.046) (.046) (.046)      Education (years) -.021*** -.020*** -.020***  -.018** -.018** -.018**  (.004) (.004) (.004)  (.006) (.006) (.006)      Age .004*** .004*** .004***  .001 .001 .001  (.001) (.001) (.001)  (.001) (.001) (.001)      Children .058 .060 .058  -.085 -.080 -.080  (.037) (.037) (.037)  (.057) (.057) (.057)      Under 5 .076** .073** .073**  -.078* -.082* -.082*  (.022) (.022) (.022)  (.036) (.036) (.036)      Household Size .035* .032* .032*  .051* .046 .046  (.015) (.015) (.015)  (.024) (.024) (.024)      Weekday -.179*** -.178*** -.178***  -.219*** -.218*** -.219***  (.023) (.023) (.023)  (.033) (.033) (.033)         163   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m French -.026 -.022 -.022  -.016 -.016 -.016  (.036) (.036) (.036)  (.061) (.061) (.061) Chinese .140 (.076) .135 (.076) .137 (.076)  .043 (.124) .041 (.124) .043 (.124) South Asian .452*** (.085) .453*** (.085) .453*** (.085)  .230 (.164) .235 (.163) .234 (.163) Filipino .198 (.117) .208 (.116) .211 (.117)  .121 (.171) .102 (.173) .104 (.173) Province level controls Yes Yes Yes  Yes Yes Yes      Year variables Yes Yes Yes  Yes Yes Yes      Province vars. Yes Yes Yes  Yes Yes Yes Constant 4.808*** 4.732*** 4.714***  4.804*** 4.791*** 4.790***  (.470) (.471) (.471)  (.776) (.775) (.775) ρ -.922*** -.921*** -.921***  -.888*** -.888*** -.888*** σρ -.851 -.848 -.849  -1.047 -1.048 -1.048 N(uncensored) 13099(11230) 13099(11230) 13099(11230)  12509(6856) 12509(6856) 12509(6856) Log likelihood -18579.275 -18566.067 -18568.815  -19906.866 -19899.054 -19899.603 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001    164  Table 32 Year and Province Fixed-Effects Models on Log of Time Spent on Cleaning for Women and Men with Heckman Adjustment, 1986-2010  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) -.008 .037 .031  .163 .145 .119 (.029) (.071) (.032)  (.055) (.120) (.067) Δ2  .140** .141**   .107 .093   (.047) (.044)   (.093) (.081) Δ3  -.008    -.041    (.086)    (.147)  Control Variables             Born in Canada -.131** -.126** -.126**  -.017 -.013 -.013  (.046) (.046) (.046)  (.068) (.068) (.068)      Education (years) -.026*** -.024*** -.024***  -.029** -.029** -.028**  (.006) (.006) (.006)  (.009) (.009) (.009)      Age -.005*** -.005*** -.005***  .002 .002 .002  (.001) (.001) (.001)  (.002) (.002) (.002)      Children -.222*** -.222*** .222***  -.132 -.129 -.130  (.055) (.054) (.054)  (.087) (.086) (.087)      Under 5 -.036 -.036 -.036  -.172** -.175** -.174**  (.033) (.033) (.033)  (.057) (.057) (.057)      Household Size .034 .030 .030  -.029 -.031 -.031  (.023) (.023) (.026)  (.036) (.036) (.036)      Weekday -.240*** -.242*** -.242***  -.139** -.139** -.139**  (.033) (.033) (.033)  (.051) (.051) (.051)         165   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m French -.016 (.057) -.010 (.057) -.010 (.057)  .018 (.101) .017 (.101) .017 (.101) Chinese -.164 (.122) -.169 (.121) -.168 (.121)  -.191 (.164) -.199 (.163) -.197 (.163) South Asian -.083 (.134) -.082 (.134) -.082 (.134)  .191 (.250) .189 (.250) .187 (.250) Filipino .301 (.173) .316 (.173) .316 (.173)  .179 (.362) .164 (.368) .165 (.367) Province level controls Yes Yes Yes  Yes Yes Yes      Year variables Yes Yes Yes  Yes Yes Yes      Province vars. Yes Yes Yes  Yes Yes Yes Constant 6.271*** 6.141*** 6.140***  5.990*** 5.959*** 5.959***  (.705) (.704) (.704)  (1.286) (1.285) (1.285) ρ -.925*** -.924*** -.924***  -.861*** -.861*** -.861*** σρ -1.140 -.1.136 -1.136  -1.125 -1.123 -1.123 N(uncensored) 13099(8196) 13099(8196) 13099(8196)  12509(2712) 12509(2712) 12509(2712) Log likelihood -19152.671 -19144.799 -19144.806  -11607.792 -11606.622 -11606.684 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001    166  Table 33 Year and Province Fixed-Effects Models on Log of Time Spent on Shopping for Women and Men with Heckman Adjustment, 1986-2010  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) .015 .062 .048  .087* .183* .049 (.030) (.074) (.033)  (.042) (.093) (.055) Δ2  .127** .130**   .163 .068   (.049) (.047)   (.085) (.069) Δ3  -.019    -.227    (.092)    (.125)  Control Variables             Born in Canada -.155** -.149** -.150**  -.150** -.150** -.147**  (.046) (.046) (.046)  (.053) (.053) (.053)      Education (years) -.050*** -.049*** -.049***  -.048*** -.049*** -.048***  (.007) (.007) (.007)  (.007) (.007) (.007)      Age .003* .003 .003  -.001 -.001 -.001  (.001) (.001) (.001)  (.002) (.002) (.002)      Children -.004 -.004 -.004  .092 .091 .093  (.058) (.058) (.058)  (.074) (.074) (.074)      Under 5 .127*** .122** .122**  -.041 -.042 -.041  (.036) (.036) (.036)  (.046) (.046) (.046)      Household Size -.027 -.031 -.031  -.039 -.040 -.040  (.025) (.025) (.025)  (.031) (.031) (.031)      Weekday -.154*** -.157*** -.157***  -.114** -.113*** -.114**  (.035) (.035) (.035)  (.042) (.042) (.042)         167   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m French .029 (.065) .034 (.065) .034 (.065)  .167* (.072) .169* (.072) .168* (.072) Chinese .139 (.123) .133 (.124) .133 (.124)  .126 (.150) .120 (.150) .126 (.150) South Asian .059 (.146) .058 (.146) .058 (.146)  .287 (.170) .288 (.170) .285 (.169) Filipino .386 (.203) .402* (.202) .402* (.202)  .007 (.219) .006 (.218) .003 (.219) Province level controls Yes Yes Yes  Yes Yes Yes      Year variables Yes Yes Yes  Yes Yes Yes      Province vars. Yes Yes Yes  Yes Yes Yes Constant 5.612*** 5.523*** 5.519***  5.711*** 5.697*** 5.706***  (.752) (.754) (.754)  (.962) (.962) (.962) ρ -.928*** -.927*** -.927***  -.926*** -.926*** -.926*** σρ -1.153 -1.151 -1.151  -1.257 -1.257 -1.257 N(uncensored) 13099(6259) 13099(6259) 13099(6259)  12509(4670) 12509(4670) 12509(4670) Log likelihood -16748.649 -16742.899 -16742.929  -16352.893 -16349.428 -16352.093 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001   168  Table 34 Year and Province Fixed-Effects Models on Log of Time Spent on Maintenance for Women and Men with Heckman Adjustment, 1986-2010  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) .001 -.139 .069  -.079 -.168 -.239 (.121) (.234) (.139)  (.087) (.195) (.136) Δ2  .264 .201   .327 .258   (.224) (.203)   (.229) (.156) Δ3  .305    -.137    (.378)    (.294)  Control Variables             Born in Canada -.099 -.106 -.095  -.117 -.101 -.101  (.213) (.213) (.213)  (.125) (.125) (.125)      Education (years) -.034 -.032 -.031  -.020 -.020 -.020  (.032) (.032) (.032)  (.015) (.015) (.015)      Age .006 .006 .006  -.006 -.006 -.006  (.006) (.006) (.006)  (.003) (.003) (.003)      Children -.388 -.392 -.390  -.026 -.020 -.021  (.241) (.241) (.241)  (.131) (.130) (.131)      Under 5 -.054 -.053 -.056  -.043 -.050 -.050  (.161) (.160) (.160)  (.087) (.087) (.087)      Household Size .064 .056 .058  -.024 -.031 -.031  (.121) (.121) (.121)  (.055) (.055) (.055)      Weekday .149 .135 .140  .008 .007 .008  (.143) (.143) (.143)  (.080) (.080) (.080)         169   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m French .325 (.265) .328 (.265) .326 (.266)  .097 (.131) .097 (.131) .097 (.131) Chinese .801 (.675) .798 (.661) .778 (.666)  1.084* (.476) 1.048* (.468) 1.066* (.467) South Asian 1.361 (.777) 1.379 (.771) 1.388 (.768)  1.386** (.459) 1.381** (.457) 1.375** (.457) Filipino (omit.) (omit.)  (omit.)  1.489* (.708) 1.514* (.708) 1.523* (.708) Province level controls Yes Yes Yes  Yes Yes Yes      Year variables Yes Yes Yes  Yes Yes Yes      Province vars. Yes Yes Yes  Yes Yes Yes Constant 10.020** 9.920*** 9.846**   7.313*** 7.283*** 7.302***  (3.709) (3.713) (3.722)  (1.844) (1.845) (1.842) ρ -.946*** -.945*** -.944***  -.969*** -.969*** -.969*** σρ -2.068 -2.053 -2.044  -1.908 -1.907 -1.908 N(uncensored) 13099(589) 13099(589) 13099(589)  12509(1691) 12509(1691) 12509(1691) Log likelihood -3046.7806 -3045.735 -3046.1438  -8091.3189 -8088.6582 -8088.8692 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001  170  Chapter 6: Conclusions  This dissertation provides a novel, detailed analysis of the gender gap in the household division of labour by employing the Blinder-Oaxaca decomposition method followed by the analysis of the association of relative resources with the participation in housework among women and men. The latter, more conventional in the field up to this day, provides findings congruent with much of the extant literature. Employing the time use diaries from the Canadian General Social Survey accessed at the UBC Research Data Centre, I explored (1) whether the gender convergence of housework was achieved in various individual housework tasks, including routine indoor, routine outdoor, and non-routine tasks; and (2) what theoretical perspective, particularly whether it is one corresponding with the economic approach expectations or the gender-centred perspective, could explain the gender gap and (3) whether the association of women’s and men’s resources with their participation in individual housework tasks result in the predicted patterns. The present project, therefore, facilitated a more thorough study into whether women continue to face disadvantage with respect to the time spent on diverse household tasks. Specifically, the findings show not only the role of power balance and resources for participation in housework but also to answer more elusive questions such as whether this association can be accounted for by specific resources and what the processes of household decision-making underlying them are between married and cohabiting women and men in Canada. As to the first question, the key finding is that although there is a discernible change in Canada toward a more equalitarian division of household labour since 1986, within the last 18 years, there is a slight stagnation evident both in Canada and in the US. Overall, women still do 171  more housework than men, even when we compare the average time spent on housework between women and men in same-sex partnerships. The decomposition analysis aimed at answering the second question of the present dissertation and uncovered meaningful variations within types of housework and bore results suggesting that the least portion of the gender gap can be ascribed to discrimination or other unobserved variance in routine outdoor tasks such as shopping than in other types of housework. Furthermore, a number of mechanisms were found to be at work to produce this gender gap in the individual types of domestic chores. Thus the time availability framework can explain most of the observed gender gap for more egalitarian tasks such as shopping. In other words, a good portion of the gender gap can be accounted for by who has more time to do shopping for the household and that the bargaining in shopping tasks happens with the time resources rather than with monetary means. On the other hand, for indoor routine housework, which is traditionally associated more with women’s work and where there is still a considerable gender gap, the relative resources framework provides the highest share of the explanatory power and thus suggests that the power imbalance and bargaining with economic resources are the processes maintaining the gender gap in cooking and cleaning.  This finding has a few immediate implications. First and foremost, Gupta’s (2007) contention regarding the reliance of the household division of labour on autonomy cannot be valid for indoor routine housework overall and is supported only in a few cultural groups of Canadians, namely among French and Chinese Canadians. The greater share that autonomous resources can explain for the gender gap among French and Chinese Canadians suggests that the cultural norms of the French and Chinese Canadians may rely more on autonomous decisions within the housework domain and the avoidant type of conflict resolution in partnerships (if the assumptions of the autonomy perspective are correct, see Gupta (2007)) rather than bargaining 172  and open power struggle as among other Canadian cultural groups. Second, Chapter 5 shows unequivocally that partners that have equal resources and thus equal economic power share housework in a way that women spend the least time on routine housework. It is, however, essential to have power balance within (heterosexual) partnerships by ensuring that women reach pay equity with men to attain more gender-equal division of housework on the societal scale. 6.1 Addressing the Hypotheses As to the specific hypotheses raised in Chapter 2, I find support for some and did not have enough evidence to reject others, all of which I summarize below in Table 35.  Table 35 Summary of the Results Associated with Hypotheses Question 1. Convergence has mostly occurred in gendered housework. Question 2. The gender gap in gender-neutral housework tasks is not explained by the same frameworks as the gendered housework. Question 3. All dependent women and men bargain. Breadwinning women do gender. Breadwinning men undo gender. H1.1. The gender gap is narrower in gender-neutral tasks. H2.1. The economic factors can explain a greater portion of the gender gap in gender-neutral tasks. H3.1 Women do gender, men don’t.  H2.2. Power relations as captured by the relative resources perspective matters more in gendered housework. H3.2. Breadwinning women who have more resources do more housework.   H3.3. Breadwinning men do more housework than men who contribute equally as their partners.   H3.4. Women do gender in all routine housework  The analysis of trends in housework reveal support for the expectations of the Hypothesis 1.1. Thus I find that the gender gap in Canada is actually wider in tasks that are traditionally considered more gendered such as cooking and cleaning compared to shopping, which is regarded as more gender-neutral. Regardless of their cultural group belonging, Canadian women continue to do most of the housework. Yet the ratio of all women’s time spent on overall 173  domestic chores to all men’s time decreased from 2.31 in 1986 to 1.64 in 2010, indicating a slow convergence and move toward gender equality in unpaid work in Canada. In comparison, American women do approximately twice (2.006 times) as much housework as American men, implying that the gender gap in the US might be a little wider than in Canada if we disregard the measurement differences between the GSS and ATUS samples. Yet the gender gap in America appears to be stagnant for the period between 2003 and 2015. This pattern of stagnation is also present in Canada in all types of housework since 2000.  One of the measures of the egalitarian character of a task is the absolute difference in how much time women and men spend on it – the smaller the difference, the more egalitarian the task. Measuring gender equality in such a way makes shopping tasks seem the most egalitarian because women and men spend almost similar amounts of time on shopping tasks. The biggest absolute difference between women’s and men’s time spent on housework and thus the indicator of the highest gender inequality is in cleaning tasks, which are also self-reported by Canadians as the least enjoyable. These results confirm the predictions of the Hypothesis 1.1, the narrowest gender gap is in shopping which is coincidentally considered the most gender-neutral. The confirmation of the Hypothesis 1.1 is found for all cultural groups with the exception of South Asian Canadians, where men spend more time on shopping because shopping is considered culturally a more masculine task.  Hypothesis 2.1 states that the gender gap is more likely to be accounted for by the availability of resources, relative or absolute, for the tasks that are more gender-neutral such as shopping compared to tasks that are still considered more traditionally feminine – such as routine indoor housework tasks (cooking and cleaning). The decomposition analysis of the gender gap showed that in fact, as expected, the approaches lying within the economic theory can explain most if not 174  all of the gender gap in shopping, while they can only explain about a third of the gender gap in cooking and cleaning.  If the unexplained portion of the gender gap can be ascribed to factors other than gender discrimination, including cultural determinants with regard to gender attitudes which might not be adequately measured in the models, then there is room for unexplained variance that is due to such cultural factors and gender discrimination in the gender gap in cooking and to some extent in cleaning. Thus I can conclude that in fact, the results herein confirm the hypothesis and show that the more egalitarian tasks can be accounted for by the economic factors lying within the economic framework. Moreover, factors that can explain most of the gender gap are quite similar for all cultural groups analyzed: Anglo-, French, Chinese, South Asian, and Filipino Canadians. This result of invariability of the gender gap determinants across cultural groups suggests negligible cultural differences in terms of the factors accounting for the gender gap in domestic work. One implication is that most of the gender gap, at least in the tasks such as shopping and cleaning, will be narrowed down through policy actions leveling labour market opportunities for women with those of men. Nevertheless, the current findings also confirm that outside of the economic factors, a big share of the gender gap in routine housework can still be accounted for by gender discrimination or other unobserved variance especially among South Asian Canadians, where the gender gap remains mostly unexplained by all factors proposed by the present project. A considerable part of the gender gap in daily routine housework tasks such as cooking and traditionally masculine tasks such as maintenance cannot be explained by the economic factors brought about by the present project, leaving a big share of the gap unexplained. The economic 175  factors are enough to account for the most of the gender differences in shopping and cleaning, especially among those who report doing any housework activity on the diary day. The fact that economic factors play a significant role in accounting for the gender gap in routine housework casts doubt on the pervasive gender-centred explanation for the division of domestic labour (Brines 1994; Greenstein 2000). The gender explanations focusing on advocating changes in norms and attitudes shift the emphasis of policymakers away from labour market inequalities, potentially stalling the progress toward greater gender equity that can be achieved through leveling economic opportunities for women.  According to Hypothesis 2.2., relative resources of a partner are expected to explain a greater portion of the gender gap than absolute resources in gendered tasks such as routine housework tasks (cooking and cleaning) than in more gender-neutral tasks. The findings of Chapter 4 confirm that in fact, the relative resources argument can explain the biggest share of the gender gap in time spent on cooking and cleaning compared to other theoretical approaches. The predominance of the relative resources framework also suggests that the power imbalance between partners disadvantages those partners who contribute less to household income and pressure them to allot more of their time to routine indoor housework. On the other hand, the relative resources approach does not explain as much as the time availability framework for the time spent by partners on shopping, where time constraints can account for most of the gender gap. Thus the gender gap in shopping time would possibly completely close in most of the major Canadian cultural groups if women spent the same amount of time on labour market activities as men.  The short answer to the question whether there is a cultural gap in the explanation of the time spent on housework is no. Yet there are a few exceptions. Thus participation in routine 176  housework among Filipino Canadians can be explained by the time availability framework, unlike among other cultural groups in Canada, where the reign of the relative resources approach is uncontested. This finding suggests that among Filipino Canadians, the decision-making about who should do more housework depends on who has time to cook or clean rather than on partners’ relative resources and thus Filipino Canadians bargain with time and not with monetary resources. South Asian Canadians stand alone in their traditional association of shopping activities with men’s work rather than women’s. It means that the interpretation of the gender gap in shopping among South Asian Canadians is different than for other Canadians.  Hypothesis 3.1 states that both men and women are expected to ‘do’ gender in all types of housework. Extending the theoretical specifications for testing the association between relative resources and time spent on housework proposed by Brines (1994), I arrived at the conclusion that there are considerable similarities in the patterns of women’s and men’s participation by individual housework tasks. Yet the interpretations of these patterns differ by gender group. As expected, women in Canada ‘do’ gender in all routine types of housework, consistent with the findings of Greenstein (2000) among American women. Women, therefore, still face considerable pressure from society to conform to the predominant gender norms. In contrast, I find that for cooking and to some extent, for shopping and cleaning, breadwinning men produce a new pattern of gendered behaviour in Canada – they are ‘undoing’ gender. Thus men who contribute a greater share of the household income spend more time on routine housework than men whose economic contribution is on the similar level as their partners’. In other words, West and Zimmerman (1987) ideas of the ‘doing’ gender are more applicable to women rather than men in Canada, specifically to their involvement in routine housework tasks. Because the relative resources framework is capable of accounting for a sizeable portion of the gender gap in 177  indoor routine housework, Brines’ (1994) theoretical specifications still represents a strong framework for testing the power relationships concerning the division of labour within households, specifically as it pertains to the bargaining processes of households’ decision making. In sum, Hypothesis 2.3 is confirmed for women but not for men, indicating that the gender norms in Canada are disproportionately disadvantaging women relative to men. As to Hypothesis 3.2.: women who have more resources, relative or absolute, are expected to do less housework compared to women who have less resources, especially in housework tasks that are considered more gender-neutral. The changes toward a more egalitarian society with regard to the participation of women in all spheres of activities lead to expectations that women nowadays are more competitive and less likely to assume housework. Our society portrays women as the main drivers of societal change with regards to gender and one would expect to observe such change within the Canadian household. However, I find that breadwinning women are more likely to spend more time on housework than women who contribute as much as their partners to the household resources. Therefore, Canadian breadwinning women are actually ‘doing’ gender and tend to compensate for their gender norm violation in the labour market. Power that these women obtain over their partners by breadwinning in no way appears to exonerate them from doing housework. In fact, the more they earn, the more housework they do, in all indoor routine housework disregarding whether the task is more gender-neutral or not. Thus the findings of the present project allow me to reject the Hypothesis 3.1. The situation where women spend the least amount of time on housework is found to be when the economic resources that partners bring into the family match. Equality in economic power means more equal division of housework. 178  Hypothesis 3.3. states that men are expected to participate less in all types of housework proportionately to the amount of resources that they provide, relative or absolute. The economic approach predicts that men would do less housework with increased contribution of resources to the household. However, the present project uncovers a completely new phenomenon, where breadwinning men do more housework compared to men who earn as much as their partners. This pattern is evident in all routine housework but more so in cleaning and shopping. Moreover, according to the gender-centred perspective, men in dependent positions who provide less of the household resources than their partners are expected to do less housework because they ‘do’ gender and allot less time to housework proportionately to their gender norm violation, where society traditionally still expects them to be breadwinners. Thus the less men bring to the table, the less they are expected to do housework. Yet the findings of the present project insist that men do more housework in dependent positions, it is especially evident in cooking and shopping tasks but not so much in cleaning chores. The relative normalization of gender norm violation among Canadian men reveals that men in Canada are freer than women to transgress societal expectations. According to the Hypothesis 3.2., women are expected to do gender in the least enjoyable tasks such as cleaning. The present project reveals the predicted pattern of women’s participation in cleaning consistent with the premises of the gender-centred approach. Thus, the results show that breadwinning women are expected to spend more time on cleaning than women who earn as much as their partners. But this result is equally valid for all types of routine housework, not only for the least enjoyable ones. Once again, the findings confirm that women face harsher societal pressures to conform with gender norms, the more they deviate. 179  6.2 Strengths and Limitations, Implications, and Future Research The present project provided at least two main extensions to the current theoretical development in the analysis of the gendered division of housework. First, the pooled Blinder Oaxaca decomposition analysis revealed striking differences in how much the economic framework can account for in the explanation of the individual types of domestic chores. Thus while the economic factors, particularly lying within the time availability framework, are capable of accounting for the most of the gender gap in shopping, they can explain up to a third of the gender gap in indoor routine housework. Therefore, the findings of the present project provide evidence that there is a difference in mechanisms through which the division of labour occurs in individual tasks within Canadian households echoed in the variety of how much of the gender gap the analyzed frameworks can account for.  Second, the gender-centred perspective has a major blind spot which was revealed by the results of the present dissertation – it ignored the patterns uncharacteristic to the normative gender expectations such as those of the ‘undoing’ gender approach (Butler 2004; Deutsch 2007). Accordingly, I discover the ‘undoing’ gender pattern in cooking and shopping among dependent Canadian men and in cooking, cleaning, and shopping among breadwinning men.  Third, the present project establishes that the current theoretical frameworks would benefit immensely from accounting for different categories of relative economic contributors: breadwinners, equal-earners, and dependants. Thus the explanations proposed by a framework may apply to principal breadwinners but not to dependent individuals. For instance, breadwinning Canadian women ‘do’ gender in routine housework tasks because they increase their participation in housework with the increase of their relative resources, contrary to the 180  predictions of the economic approach. This observation can be explained by the desire to compensate for the gender norm violation in the labour market. On the other hand, dependent women do not ‘do’ gender in cooking and do not employ gender neutralization techniques because their pattern is consistent with the expectations of the economic framework. Dependent women increase their time spent on cooking with the decrease in their relative resources. Moreover, it appears from the results of the current project that the most egalitarian division of labour, in general, occurs in equal-earning households where both partners earn equally. This implication casts doubt on the over-reliance on the analysis of absolute measures of economic resources as viable predictors of the household division of labour because it is clear that housework is divided more equally if both partners contribute similar amounts of resources to the household, suggesting that economic power equity is vital for the egalitarian decision-making regarding housework.  I also believe that this project provides at least two other important implications for the gendered division of housework of a more technical character. First, I show that the conventional analysis focusing on explaining variance among women and men does not adequately study the actual gender gap in time spent on domestic labour and the factors that contribute to it. Previous research in the area analyzed the association of factors with the variance within a gender group and did not analyze the actual contribution of the factors widening or narrowing the gender gap. Second, a lot of information about mechanisms behind the division of unpaid labour is lost when housework is lumped into one category without considering at least the major types. The current analysis reveals that there are distinct patterns idiosyncratic to types of housework, indicative of the task segregation by gender group and asserting the importance of measuring housework by 181  task (Blair and Lichter 1991). These peculiarities provide evidence that explanations and mechanisms behind the division of housework depend on the type of domestic chore. More significantly, the implication is that the frameworks are distinctly more applicable to explaining ‘women’s work’ rather than participation in tasks traditionally associated with men.  It is also important to acknowledge a few limitations of the analysis. Because of the characteristics of time diary data, which are collected for one day and do not reflect longer periods of times, the present results rely heavily on the samples who report housework on the diary day. Yet it does not mean that respondents do not perform any housework on days other than the diary day. This limitation might have underestimated the significance of some factors if their effect is weakened by the time use data collection limitations. Time diaries collection favours routine activities which are more likely to happen daily, for instance, cooking but not so much of the less routine tasks such as cleaning. Conversely, other activities, which people do not always do everyday but which are still routine like cleaning and shopping, might not have estimates as reliable as those for daily routine housework. The limitation can be overcome by focusing on the samples that report doing any housework on the diary day especially for the non-daily routine housework activities utilizing the sampling bias correction. Having this limitation in mind, I analyzed the association of relative resources with time spent on housework by gender group using the Heckman correction technique.  Another limitation is that both Canadian and American time use surveys collect diaries from only one person in a household. Even though limited information is available on the spouse of the respondent, for most of the part, the detailed information on the time use and particularly, on the time spent on housework by both partners could benefit the research on housework because it 182  would provide more information on how partners actually share the housework in households rather than providing information on aggregate, gender group level. There are surveys that include information on housework for both partners such as NSFH and PSID (Brines 1994; Greenstein 2000; Lee and Waite 2005), however, those are ‘stylized’ surveys and are less reliable than time use surveys because of the inherent social desirability bias. One of the future directions this research could take to remedy this limitation is to collect time use data on both partners in households where there are couples, similarly as in Australia (Birch, Le, and Miller 2009). This, however, will require additional funding to time use data collection on federal level. Furthermore, the time spent on housework is but one way of measuring housework. There are other methods of measuring housework, especially those focused on imputing the economic value of housework (Colman 1998; Landefeld et al. 2009). These latter studies are important for policy makers, showing how much women contribute to the economy of a nation, while their work still remains undervalued and invisible. Since 1971 the Government of Canada (Statistics Canada) has taken steps to measure the economic value of unpaid labour (Hamdad 2003; Prince Edward Island Advisory Council on the Status of Women 2003; Stone and Pelletier 1998). These measures facilitate our understanding on how much (at least approximately) the domestic work contributes to our economy. All of these estimates of economic value, however, rely on the time use diaries and time use surveys. Therefore, to be able to impute the economic value of domestic labour, first and foremost, we need to have accurate estimates of the time spent on housework. Yet the collection of time use diaries in Canada is not conducted yearly and due to the lack of funding, the 2016 GSS time-use survey (which is not yet released at the time that I write my dissertation) runs the risk of being the last time use survey conducted by Statistics Canada. I urge 183  the policy makers to understand the importance of time use surveys for the analysis of the invisible economy of unpaid labour and of gender inequality perpetuated by it.  As to the methodological implications, one of the ways that research of housework should proceed methodologically is to allow the collection of data which would help time use diaries to adjust for the selection bias more precisely than the present data allowed. It is important to develop more empirical work with regard to the adjustment for the selection bias because individuals who report doing housework might be different from individuals who did not report any housework in the predictable ways that can be captured by additional variables. As I noted at the end of Chapter 4, another way to circumvent the selection bias is to use imputation for those who did not report doing housework on the diary day, perhaps employing machine learning techniques. As the future research direction, I plan to focus on implementation of the developments of machine learning techniques to imputation in time use diaries. The complication with such imputation would be that we need more observations and more consistent longitudinal data, this would require consistent funding of the time use team of Statistics Canada and perhaps establishing collaboration initiatives with population centres in Maryland and Minnesota, involved in the development of the American Time Use Survey. Another direction the future research could take, is in the development and adoption of more detailed measures for social class to probe if the changes in attitudes and behaviours around housework happen unequally for people of different stations. This direction is especially interesting apropos the analysis of leisure activities in more detail. It might well be that it is the lifestyle differences that dictate time allocation to certain leisurely activities over housework.   184  One of the characters in Edna Ferber’s novels exclaims that “Housework's the hardest work in the world. That's why men won't do it,” but she may be wrong today – men in Canada and America are starting to do what women do when it comes to routine housework. Dependent men spend more time on housework, ‘undoing’ their traditional gender, and breadwinning men do more housework driving the societal change. The shift toward more egalitarian society is here. But why we, women, have to be the perennial damsels in distress waiting to be saved by Princes Charming? The Canadian society is in dire need of a revolution in its expectations laid disproportionately on women. We need to extend opportunities to safely violate gender norms for all women.   185  Bibliography  Abraham, Katharine G., Aaron Maitland, and Suzanne M. Bianchi. 2006. “Nonresponse in the American Time Use Survey.” Public Opinion Quarterly 70(5):676–703. Acock, Alan C., and David H. Demo. 1994. Family Diversity and Well-Being. Thousand Oaks, CA: Sage. Alwin, Duane F., and Ryan J. McCammon. 2004. “Generations, Cohorts, and Social Change.” Pp. 23–50 in Handbook of the Life Course, edited by Jeylan T. Mortimer and Michael J. Shanahan. New York, NY: Springer. Antill, John K., Jacqueline J. Goodnow, Graeme Russell, and Sandra Cotton. 1996. “The Influence of Parents and Family Context on Children’s Involvement in Household Tasks.” Sex Roles 34(3/4):215–36. Artis, Julie E., and Eliza K. Pavalko. 2003. “Explaining the Decline in Women’s Household Labor: Individual Change and Cohort Differences.” Journal of Marriage and Family 65(3):746–61. Barnett, Rosalind C., and Yu-Chu Shen. 1997. “Gender, High- and Low-Schedule-Control Housework Tasks, and Psychological Distress: A Study of Dual Earner Couples.” Journal of Family Issues 18(4):403–28. Bascle, Guilhem. 2008. “Controlling for Endogeneity with Instrumental Variables in Strategic Management Research.” Strategic Organization 6(3):285–327. Batalova, Jeanne A., and Philip N. Cohen. 2002. “Premarital Cohabitation and Housework: Couples in Cross-National Perspective.” Journal of Marriage and Family 64(August):743–55. 186  Baxter, Janeen. 2002. “Patterns of Change and Stability in the Gender Division of Household Labour in Australia, 1986-1997.” Journal of Sociology 38(4):399–424. Baxter, Janeen, and Belinda Hewitt. 2013. “Negotiating Domestic Labor: Women’s Earnings and Housework Time in Australia.” Feminist Economics 19(1):29–53. Baxter, Janeen, and Emily W. Kane. 1995. “Dependence and Independence: A Cross-National Analysis of Gender Inequality and Gender Attitudes.” Gender & Society 9(2):193–215. Becker, Gary S. 1965. “A Theory of the Allocation of Time.” The Economic Journal 75(299):493–517. Becker, Gary S. 1981. A Treatise on the Family. Cambridge, MA: Harvard University Press. Berk, Sarah Fenstermaker. 1985. The Gender Factory: The Apportionment of Work in American Households. New York: Plenum. Bhabha, Homi K. 1994. The Location of Culture. London and New York, NY: Routledge. Bianchi, Suzanne M., and Melissa A. Milkie. 2010. “Work and Family Research in the First Decade of the 21st Century.” Journal of Marriage and Family 72(3):705–25. Bianchi, Suzanne M., Melissa A. Milkie, Liana C. Sayer, and John P. Robinson. 2000. “Is Anyone Doing the Housework? Trends in the Gender Division of Household.” Social Forces 79(1):191–228. Bianchi, Suzanne M., John P. Robinson, and Melissa A. Milkie. 2006. Changing Rhythms of American Family Life. New York, NY: Russell Sage Foundation. Birch, Elisa Rose, Anh T. Le, and Paul W. Miller. 2009. Household Divisions of Labour: Teamwork, Gender and Time. Palgrave Macmillan. Bittman, Michael, Paula England, Liana Sayer, Nancy Folbre, and George Matheson. 2003. “When Does Gender Trump Money? Bargaining and Time in Household Work.” American 187  Journal of Sociology 109(1):186–214. Blair, Sampson Lee, and Daniel T. Lichter. 1991. “Measuring the Division of Household Labor: Gender Segregation of Housework Among American Couples.” Journal of Family Issues 12(1):91–113. Blau, Francine D., Mary C. Brinton, and David B. Grusky, eds. 2006. The Declining Significance of Gender? New York: Russell Sage Foundation. Blood, Robert O., and Donald M. Wolfe. 1960. Husbands and Wives, the Dynamics of Married Living. New York: The Free Press. Bonifacio, Glenda Tibe. 2013. Pinay on the Prairies: Filipino Women and Transnational Identities. Vancouver, BC: UBC Press. Bourdieu, Pierre. 1980. Distinction: A Social Critique of the Judgement of Taste. Cambridge, MA: Harvard University Press. Breen, Richard, and LP Cooke. 2005. “The Persistence of the Gendered Division of Domestic Labour.” European Sociological Review 21(1):43–57. Brines, Julie. 1994. “Economic Dependency, Gender, and the Division of Labor at Home.” American Journal of Sociology 100(3):652. Browne, Jude. 2004. “Resolving Gender Pay Inequality? Rationales, Enforcement and Policy.” Journal of Social Policy 33(4):553–71. Bureau of Labor Statistics. 2015. American Time Use Survey. Washington, DC. Retrieved April 15, 2015 (http://www.bls.gov/tus/). Bureau of Labor Statistics. 2016. “Regional and State Employment and Unemployment Archived News Releases.” Retrieved September 9, 2016 (http://www.bls.gov/schedule/archives/laus_nr.htm#2003). 188  Bushway, Shawn, Brian D. Johnson, and Lee Ann Slocum. 2007. “Is the Magic Still There? The Use of the Heckman Two-Step Correction for Selection Bias in Criminology.” Journal of Quantitative Criminology 23(2):151–78. Buss, W.Christian, and Charles W. Schaninger. 1983. “The Influence of Sex Roles on Family Decision Processes and Outcomes.” Advances in Consumer Research 10:439–44. Butler, Judith. 1990. Gender Trouble: Feminism and the Subversion of Identity. New York, NY: Routledge. Butler, Judith. 2004. Undoing Gender. New York, NY: Routledge. Carrington, Christopher. 1999. No Place like Home: Relationships and Family Life among Lesbians and Gay Men. Chicago: University of Chicago Press. Chang, Chin F., and Paula England. 2011. “Gender Inequality in Earnings in Industrialized East Asia.” Social Science Research 40(1):1–14. Chang, Mariko Lin. 2000. “The Evolution of Sex Segregation Regimes.” American Journal of Sociology 105(6):1658–1701. Charles, Maria, and Karen Bradley. 2009. “Indulging Our Gendered Selves? Sex Segregation by Field of Study in 44 Countries.” AJS; American journal of sociology 114(4):924–76. Chesters, J. 2011. “Gender Convergence in Core Housework Hours: Assessing the Relevance of Earlier Approaches for Explaining Current Trends.” Journal of Sociology 49(1):78–96. Chetty, Raj, Nathaniel Hendren, Patrick Kline, and Emmanuel Saez. 2014. “Where Is the Land of Opportunity? The Geography of Intergenerational Mobility in the United States.” Quarterly Journal of Economics (September):1–71. Colman, Ronald. 1998. The Economic Value of Unpaid Housework and Child Care in Nova Scotia. 189  Coltrane, Scott. 2000. “Research on Household Labor: Modeling and Measuring the Social Embeddedness of Routine Family Work.” Journal of Marriage and Family 62(4):1208–33. Coltrane, Scott, and Masako Ishii-Kuntz. 1992. “Men ’S Housework : A Life Course Perspective.” Journal of Marriage and Family 54(1):43–57. Coltrane, Scott, Ross D. Parke, and Michele Adams. 2004. “Complexity of Father Involvement in Low Income Mexican American Families.” Family relations 53(2):179–89. Cooke, Lynn Prince. 2006. “‘Doing’ Gender in Context: Household Bargaining and Risk of Divorce in Germany and the United States.” American Journal of Sociology 112(2):442–72. Coverman, Shelley. 1985. “Explaining Husbands’ Participation in Domestic Labor.” The Sociological Quarterly 26(1):81–97. Coverman, Shelley, and Joseph F. Sheley. 1986. “Change in Men’s Housework and Child-Care Time, 1965-1975.” Journal of Marriage and Family 48(2):413–22. Craig, Lyn, and Janeen Baxter. 2016. “Domestic Outsourcing, Housework Shares and Subjective Time Pressure: Gender Differences in the Correlates of Hiring Help.” Social Indicators Research 125(1):271–88. Craig, Lyn, Abigail Powell, and Judith E. Brown. 2015. “Co-Resident Parents and Young People Aged 15–34: Who Does What Housework?” Social Indicators Research 121(2):569–88. Cunningham, Mick. 2005. “Gender in Cohabitation and Marriage: The Influence of Gender Ideology on Housework Allocation Over the Life Course.” Journal of Family Issues 26(8):1037–61. Cunningham, Mick. 2007. “Influences of Women’s Employment on the Gendered Division of Household Labor Over the Life Course: Evidence From a 31-Year Panel Study.” Journal of family issues 28(3):422–44. 190  Darmon, AdamDrewnowski Nicole, and Nicole Drewnowski. 2008. “Does Social Class Predict Diet Quality?” The American Journal of Clinical Nutrition 87(5):1107–17. Davis, Shannon N., and Theodore N. Greenstein. 2013. “Why Study Housework? Cleaning as a Window Into Power in Couples.” Journal of Family Theory & Review 5(2):63–71. Demo, David H., and Alan C. Acock. 1993. “Family Diversity and the Division of Domestic Labor : How Much Have Things Really Changed?” Family Relations 42(3):323–31. Deutsch, Francine M. 2007. “Undoing Gender.” Gender and Society 21(1):106–27. DiPrete, T. A., and Claudia Buchmann. 2013. The Rise of Women: The Female Advantage in Education and What It Means for American Schooling. New York: Russell Sage Foundation. Dotti Sani, Giulia Maria. 2016. “Undoing Gender in Housework? Participation in Domestic Chores by Italian Fathers and Children of Different Ages.” Sex Roles 74(9–10):411–21. Edlund, Jonas, and Ida Oun. 2016. “Who Should Work and Who Should Care? Attitudes towards the Desirable Division of Labour between Mothers and Fathers in Five European Countries.” Acta Sociologica 59(2):151–69. Ehrenreich, Barbara. 2001. Nickel and Dimed: On (Not) Getting By America . New York, NY: Metropolitan Books. Elder, Todd E., John H. Goddeeris, and Steven J. Haider. 2010. “Unexplained Gaps and Oaxaca–Blinder Decompositions.” Labour Economics 17(1):284–90. England, Paula, and George Farkas. 1986. “Household Formation, Marriage, Divorce.” Pp. 31–101 in Households, Employment, and Gender: A Social, Economic, and Demographic View. England, Paula, and Barbara Stanek Kilbourne. 1990. “Markets, Marriages, and Other Mates: The Problem of Power.” Pp. 163–89 in Beyond the marketplace: Rethinking economy and 191  society, edited by Roger Friedland and A.F. (Sandy) Robertson. Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capitalism. Cambridge, UK: Polity Press. Evertsson, Marie. 2014. “Gender Ideology and the Sharing of Housework and Child Care in Sweden.” Journal of Family Issues 35(7):927–49. Evertsson, Marie, and Magnus Nermo. 2004. “Dependence within Families and the Division of Labour: Comparing Sweden and the United States.” Journal of Marriage and the Family 66(5):1272–86. Ferrao, Vincent. 2010. Paid Work. Ottawa, ON. Ferree, Myra Marx. 1990. “Beyond Separate Spheres: Feminism and Family Research.” Journal of Marriage and Family 52(November):866–84. Gershuny, Jonathan. 2000. Changing Times: Work and Leisure in Postindustrial Society. Oxford, New York: Oxford University Press. Gerson, Judith M., and Kathy Peiss. 1985. “Boundaries, Negotiation, Consciousness: Reconceptualizing Gender Relations.” Social Problems 32(4):317–31. Gimenez-Nadal, Jose Ignacio, and Almudena Sevilla. 2012. “Trends in Time Allocation: A Cross-Country Analysis.” European Economic Review 56(6):1338–59. Goffman, Erving. 1976. “Replies and Responses.” Language in Society 5(3):257. Goldberg, Abbie E. 2013. “‘Doing’ and ‘Undoing’ Gender: The Meaning and Division of Housework in Same-Sex Couples.” Journal of Family Theory & Review 5(2):85–104. Greenstein, Theodore N. 2000. “Economic Dependence, Gender, and the Division of Labor in the Home: A Replication and Extension.” Journal of Marriage and Family 62(May):322–35. 192  Guppy, Neil, and Nicole Luongo. 2015. “The Rise and Stall of Canada’s Gender-Equity Revolution.” Canadian Review of Sociology 52(3):241–65. Gupta, Sanjiv. 2006a. “Her Money, Her Time: Women’s Earnings and Their Housework Hours.” Social Science Research 35(4):975–99. Gupta, Sanjiv. 2006b. “The Consequences of Maternal Employment During Men’s Childhood for Their Adult Housework Performance.” Gender & Society 20(1):60–86. Gupta, Sanjiv. 2007. “Autonomy, Dependence, or Display? The Relationship between Married Women’s Earnings and Housework.” Journal of Marriage and Family 69(2):399–417. Hakim, Catherine. 1995. “Five Feminist Myths about Women’s Employment.” The British journal of sociology 46(3):429–55. Hakim, Catherine. 1998. “Developing a Sociology for the Twenty-First Century: Preference Theory.” The British Journal of Sociology 49(1):137–43. Hamdad, Malika. 2003. Valuing Households’ Unpaid Work in Canada, 1992 and 1998: Trends and Sources of Change. Ottawa, ON. Hamlett, J., A. R. Bailey, A. Alexander, and G. Shaw. 2008. “Ethnicity and Consumption: South Asian Food Shopping Patterns in Britain, 1947--75 1.” Journal of Consumer Culture 8(1):91–116. Heath, Anthony. 1976. Rational Choice and Social Exchange: A Critique of Exchange Theory. Cambridge, UK: Cambridge University Press. Heckman, James J. 1976. “The Common Structure of Statistical Models of Truncation, Sample Selection and Limited Dependent Variables and a Simple Estimator for Such Models.” Annals of Economic and Social Measurement 5(4):475–92. Heckman, James J. 1979. “Sample Selection Bias as a Specification Error.” Econometrica 193  47(1):153–61. Heer, David M. 1963. “The Measurement and Bases of Family Power: An Overview.” Marriage and Family Living 25(2):133–39. Heisig, J. P. 2011. “Who Does More Housework: Rich or Poor?: A Comparison of 33 Countries.” American Sociological Review 76(1):74–99. Hobson, Barbara. 1990. “No Exit, No Voice: Women’s Economic Dependency and the Welfare State.” Acta Sociologica 33(3):235–50. Hochschild, Arlie Russell, and Anne Machung. 1989. The Second Shift. New York, NY: Viking Penguin. Hofferth, Sandra L., and Lynne M. Casper. 2007. Handbook of Measurement Issues in Family Research. Mahwah, NJ: Lawrence Erlbaum Associates. Hofferth, Sandra L., Sarah M. Flood, and Matthew Sobek. 2015. American Time Use Survey Data Extract System: Version 2.5 [Machine-Readable Database]. College Park, Maryland, and Minneapolis, Minnesota. Hofferth, Sandra L., Sarah M. Flood, and Matthew Sobek. 2016. “American Time Use Survey Data Extract Builder.” Retrieved June 29, 2016 (https://www.atusdata.org/atus/).  Hofferth, Sandra L., and Frances Goldscheider. 2010. “Family Structure and The Transition to Early Parenthood.” Demography 47(2):415–37. Hofferth, Sandra L., and Yoonjoo Lee. 2016. “Family Structure and Trends in US Fathers’ Time with Children, 2003 – 2013.” Family Science 6(1):318–29. van Hooff, Jenny H. 2011. “Rationalising Inequality: Heterosexual Couples’ Explanations and Justifications for the Division of Housework along Traditionally Gendered Lines.” Journal of Gender Studies 20(1):19–30. 194  Hook, Jennifer L. 2004. “Reconsidering the Division of Household Labor: Incorporating Volunteer Work and Informal Support.” Journal of Marriage and Family 66:101–17. Hook, Jennifer L. 2006. “Care in Context: Men’s Unpaid Work in 20 Countries, 1965-2003.” American Sociological Review 71(4):639–60. Hook, Jennifer L. 2010. “Gender Inequality in the Welfare State: Sex Segregation in Housework, 1965-2003.” AJS; American journal of sociology 115(5):1480–1523. Hu, Qiao et al. 2014. “East Is East and West Is West and Never the Twain Shall Meet: Work Engagement and Workaholism across Eastern and Western Cultures.” Journal of Behavioral and Social Sciences 1(1):6–24. Hwang, Jisoo. 2015. “The Second Shift: Assimilation in Housework Time among Immigrants.” Review of Economics of the Household 1–19. Immigration, Refugees and Citizenship Canada. 2015. Facts and Figures 2014 – Immigration Overview: Permanent Residents. Ishii-Kuntz, M., and Scott Coltrane. 1992. “Predicting the Sharing of Household Labor: Are Parenting and Housework Distinct?” Sociological Perspectives 35(4):629–47. James, W. P., M. Nelson, A. Ralph, and S. Leather. 1997. “Socioeconomic Determinants of Health. The Contribution of Nutrition to Inequalities in Health.” BMJ (Clinical research ed.) 314(7093):1545–49. Jann, Ben. 2008. “A Stata Implementation of the Blinder-Oaxaca Decomposition A Stata Implementation of the Blinder-Oaxaca Decomposition.” The Stata Journal 8(4):453–79. Jäntti, Markus et al. 2006. “American Exceptionalism in a New Light : A Comparison of Intergenerational Earnings Mobility in the Nordic Countries, the United Kingdom and the United States.” IZA Discussion Paper (1938):1–40. 195  Jones, F. L., and Jonathan Kelley. 1984. “Decomposing Differences Between Groups: A Cautionary Note on Measuring Discrimination.” Sociological Methods & Research 12(3):323–43. Kan, Man Yee. 2008. “Measuring Housework Participation: The Gap Between ‘stylised’ questionnaire Estimates and Diary-Based Estimates.” Social Indicators Research 86(3):381–400. Kan, Man Yee, and Heather Laurie. 2016. “Gender , Ethnicity and Household Labour in Married and Cohabiting Couples in the UK.” Institute for Social & Economic Research. Kan, Man Yee, and Stephen Pudney. 2008. “Measurement Error in Stylized and Diary Data on Time Use.” Sociological Methodology 38(1):101–132. Kan, Man Yee, O. Sullivan, and J. Gershuny. 2011. “Gender Convergence in Domestic Work: Discerning the Effects of Interactional and Institutional Barriers from Large-Scale Data.” Sociology 45(2):234–51. Kay, Fiona M., and John Hagan. 1998. “Raising the Bar: The Gender Stratification of Law-Firm Capital.” American Sociological Review 63(5):728–43. Killewald, Alexandra. 2011. “Opting Out and Buying Out: Wives’ Earnings and Housework Time.” Journal of Marriage and Family 73(2):459–71. Kurdek, Lawrence A. 1993. “The Allocation of Household Labor in Gay, Lesbian, and Heterosexual Married Couples.” Journal of Social Issues 49(3):127–39. Kurdek, Lawrence A. 2006. “Differences between Partners from Heterosexual, Gay, and Lesbian Cohabiting Couples.” Journal of Marriage and Family 68(2):509–28. Kurdek, Lawrence A. 2007. “The Allocation of Household Labor by Partners in Gay and Lesbian Couples.” Journal of Family Issues 28(1):132–48. 196  Landefeld, J.Steven, Barbara M. Fraumeni, and Cindy M. Vojtech. 2009. “Accounting for Household Production: A Prototype Satellite Account Using the American Time Use Survey.” Review of Income and Wealth 55(2):205–25. Lareau, Annette. 2000. “Social Class and the Daily Lives of Children: A Study from the United States.” Childhood 7(2):155–71. Lee, Myoung Jae. 2014. “Reference Parameters in Blinder-Oaxaca Decomposition: Pooled-Sample versus Intercept-Shift Approaches.” Journal of Economic Inequality 13(1):69–82. Lee, Yun-Suk, and Linda J. Waite. 2005. “Husbands’ and Wives’ Time Spent on Housework: A Comparison of Measures.” Journal of Marriage and Family 67(May):328–36. van der Lippe, Tanja. 2010. “Women’s Employment and Housework.” Pp. 41–58 in Dividing the Domestic: Men, Women, and Household Work in Cross-National Perspective, edited by Judith Treas and Sonja Drobnic. Stanford, CA: Stanford University Press. Lloyd, Kim M., and Scott J. South. 1996. “Contextual Influences on Young Men’s Transition to First Marriage.” Social Forces 74(3):1097–1119. Lundberg, Shelly, and Ra Pollak. 1996. “Bargaining and Distribution in Marriage.” The Journal of Economic Perspectives 10(4):139–58. Lupton, D. 2000. “‘Where’s Me Dinner?’: Food Preparation Arrangements in Rural Australian Families.” Journal of Sociology 36:172–86. Manser, Marilyn, and Murray Brown. 1980. “Marriage and Household Decision-Making: A Bargaining Analysis.” International Economic Review 21(1):31. Marini, Margaret Mooney, and Beth Anne Shelton. 1993. “Measuring Household Work: Recent Experience in the United States.” Social Science Research 22:361–82. Marshall, Katherine. 1986. “Converging Gender Roles.” Women (75):1–13. 197  Marshall, Katherine. 2009. The Family Work Week. Marshall, Katherine. 2011. “Generational Change in Paid and Unpaid Work.” Canadian Social Trends (11):11–24. McElroy, Marjorie B., and Mary Jean Horney. 1981. “Nash-Bargained Household Decisions: Toward a Generalization of the Theory of Demand.” International Economic Review 22(2):333–49. Meissner, Martin, Elizabeth W. Humphreys, Scott M. Meis, and William J. Scheu. 1975. “No Exit for Wives: Sexual Division of Labour and the Cumulation of Household Demands.” The Canadian review of sociology and anthropology 12(4):424–39. Mincer, Jacob, and Solomon Polachek. 1974. “Family Investments in Human Capital : Earnings of Women.” Pp. 397–431 in Economics of the Family: Marriage, Children, and Human Capital, edited by Theodore W. Schultz. Chicago: University of Chicago Press. Molina, José Alberto, Juan Carlos Campaña, and Raquel Ortega. 2016. “What Do You Prefer for a Relaxing Time at Home: Reading, Watching TV or Listening to the Radio?” Applied Economics Letters 4851(April):1–7. Nakhaie, Reza. 2002. “Class, Breadwinner Ideology, and Housework among Canadian Husbands.” Review of Radical Political Economics 34:137–57. National Center for Health Statistics. 2016. “Marriages and Divorces.” National Vital Statistics System. Retrieved August 10, 2016 (http://www.cdc.gov/nchs/nvss/marriage-divorce.htm). Neuman, Nicklas, Lucas Gottzén, and Christina Fjellström. 2015. “Narratives of Progress: Cooking and Gender Equality among Swedish Men.” Journal of Gender Studies 9236(January):1–13. Neumark, David. 1988. “Employers’ Discriminatory Behavior and the Estimation of Wage 198  Discrimination.” The Journal of Human Resources 23(3):279–95. Oaxaca, Ronald. 1973. “Male-Female Wage Differentials in Urban Labor Markets.” International Economic Review 14(3):693–709. Oaxaca, Ronald L., and Michael R. Ransom. 1999. “Identification in Detailed Wage Decompositions.” The Review of Economics and Statistics 81(1):154–57. Oerton, Sarah. 1997. “‘Queer Housewives?’: Some Problems in Theorising the Division of Domestic Labour in Lesbian and Gay Households.” Women’s Studies International Forum 20(3):421–30. Oerton, Sarah. 1998. “Reclaiming the ‘Housewife’?” Journal of Lesbian Studies 2(4):69–83. Parsons, Talcott. 1942. “Age and Sex in the Social Structure of the United States.” American Sociological Review 7(5):604–16. Patterson, Charlotte J. 2000. “Family Relationships of Lesbians and Gay Men.” Journal of Marriage and Family 62(4):1052–69. Perales, Francisco, Janeen Baxter, and Tsui-o Tai. 2015. “Gender , Justice and Work : A Distributive Approach to Perceptions of Housework Fairness.” Social Science Research 51:51–63. Piper, Nicola, and Mina Roces. 2003. “Introduction: Marriage and Migration in an Age of Globalization.” Pp. 1–22 in Wife Or Worker?: Asian Women and Migration, edited by Nicola Piper and Mina Roces. Lanham, Maryland: Rowman & Littlefield Publishers, Inc. Polachek, Alicia J., and Jean E. Wallace. 2015. “Unfair to Me or Unfair to My Spouse: Men’s and Women’s Perceptions of Domestic Equity and How They Relate to Mental and Physical Health.” Marriage & Family Review 51(3):205–28. Presser, Harriet B. 1994. “Employment Schedules Among Dual-Earner Spouses and the Division 199  of Household Labor by Gender.” American Sociological Review 59(3):348–64. Prince Edward Island Advisory Council on the Status of Women. 2003. Policy Guide: Women and Unpaid Work. Charlottetown, PEI. Pronovost, Gilles. 2012. “Transformations Des Significations Du Loisir Au Québec.” Recherches sociographiques 53(3):621–43. Puhani, Patrick A. 2000. “The Heckman Correction for Sample Selection and Its Critique.” Journal of Economic Surveys 14(1):53–68. Reimers, Cordelia W. 1983. “Labor Market Discrimiantion Against Hispanic and Black Men.” The Review of Economics and Statistics 65(4):570–79. Reitz, Jeffrey G. 2013. “Closing the Gaps Between Skilled Immigration and Canadian Labor Markets: Emerging Policy Issues and Priorities.” Pp. 147–63 in Wanted and Welcome? Policies for Highly Skilled Immigrants in Comparative Perspective, edited by Triadafilos Triadafilopoulos. Toronto, ON: Springer. Reskin, Barbara F., and Heidi I. Hartmann, eds. 1986. Women’s Work, Men’s Work: Sex Segregation on the Job. Washington, D.C.: National Academy Press. Risman, Barbara J. 2009. “From Doing to Undoing: Gender as We Know It.” Gender and Society 23(1):81–84. Robinson, John P. 1985. “The Validity and Reliability of Diaries versus Alternative Time Use Measures.” Pp. 33–62 in Time, goods, and well-being, edited by F. Thomas Juster and Frank P. Stafford. Ann Arbor, MI: University of Michigan Press. Robinson, John P., and Ann Bostrom. 1994. “The Overestimated Workweek? What Time Diary Measures Suggest.” Monthly Labor Review August:11–23. Robinson, John P., and Geoffrey Godbey. 1997. Time for Life: The Surprising Ways Americans 200  Use Their Time. Penn State Press. Rodman, Hyman. 1967. “Marital Power in France, Greece, Yugoslavia, and the United States: A Cross-National Discussion.” Journal of Marriage and the Family 29(2):320–24. Rodman, Hyman. 1972. “Marital Power and the Theory of Resources in Cultural Context.” Journal of Comparative Family Studies 3(1):50–69. Roth, Jean. 2016. “Reading Current Population Survey (CPS) Data with SAS, SPSS, or Stata.” National Bureau of Economic Research. Retrieved August 10, 2016 (http://www.nber.org/data/cps_progs.html). Satzewich, Vic, and Nikolaos Liodakis. 2007. “Race” & Ethnicity in Canada: A Critical Introduction. Toronto, ON: Oxford University Press. Sayer, Liana C. 2005. “Gender, Time, and Inequality: Trends in Women’s and Men’s Paid Work, Unpaid Work, and Free Time.” Social Forces 84(1):285–303. Sayer, Liana C. 2010. “Trends in Housework.” Pp. 19–38 in Dividing the Domestic: Men, Women, and Household Work in Cross-National Perspective, edited by Judith Treas and Sonja Drobnic. Stanford, CA: Stanford University Press. Sayer, Liana C., and Leigh Fine. 2011. “Racial-Ethnic Differences in U.S. Married Women’s and Men’s Housework.” Social Indicators Research 101(2):259–65. Scanzoni, John. 1972. Sexual Bargaining: Power Politics in the American Marriage. Englewood Cliffs, NJ. Scanzoni, John. 1978. Sex Roles, Women’s Work and Marital Conflict - A Study of Family Change. Lexington, MA: Lexington Books. Scanzoni, John. 1980. Family Decision-Making: A Developmental Sex-Role Model. Beverly Hills, CA: Sage Publications. 201  Scanzoni, John, and Karen Polonko. 1980. “A Conceptual Approach to Explicit Marital Negotiation.” Journal of Marriage and Family 42(1):31–44. Shelton, Beth Anne. 1992. Women, Men and Time: Gender Differences in Paid Work, Housework and Leisure. New York, NY: Greenwood Press. Shelton, Beth Anne, and Daphne John. 1993. “Does Marital Status Make a Difference? Housework Among Married and Cohabiting Men and Women.” Journal of Family Issues 14(3):401–20. Shelton, Beth Anne, and Daphne John. 1996. “The Division of Household Labor.” Annual Review of Sociology 22(1):299–322. Silver, Hilary, and Frances Goldscheider. 2013. “Flexible Work and Housework: Work and Family Constraints on Women’s Domestic Labor.” 72(4):1103–19. Smale, Bryan. 2010. “Leisure and Culture - a Report of the Canadian Index of Wellbeing.” Cahiers d’odonto-stomatologie 6(June):97–98. Smith, Sandra S. 2000. “Mobilizing Social Resources: Race, Ethnic, and Gender Differences in Social Capital and Persisting Wage Inequalities.” The Sociological Quarterly 41(4):509–37. Social and Aboriginal Statistics Division. 2009. General Social Survey, 2010 Cycle 24 – Time-Stress and Well-Being. Main Survey - Questionnaire Package. Ottawa, ON. Sorensen, Annemette, and Sara McLanahan. 1987. “Married Women’s Economic Dependency, 1940-1980.” American Journal of Sociology 93(3):659–87. Starrels, Marjorie E. 1994. “Gender Differences in Parent-Child Relations.” Journal of Family Issues 15(1):148–65. Statistics Canada. 2011. General Social Survey. Cycle 24: Time-Stress and Well-Being Public Use Microdata File Documentation and User’s Guide. Ottawa, ON. 202  Statistics Canada. 2015. Marriages and Crude Marriage Rates, Canada, Provinces and Territories, 1981 to 2008. Statistics Canada. 2016a. Annual Average Unemployment Rate, Canada and Provinces, 1976-2015. Statistics Canada. 2016b. “Ethnic Origin.” 2011 National Household Survey: Data tables. Retrieved August 13, 2016 (http://www12.statcan.gc.ca/nhs-enm/2011/dp-pd/dt-td/Rp-eng.cfm?TABID=2&LANG=E&APATH=3&DETAIL=0&DIM=0&FL=A&FREE=0&GC=0&GK=0&GRP=1&PID=105396&PRID=0&PTYPE=105277&S=0&SHOWALL=0&SUB=0&Temporal=2013&THEME=95&VID=0&VNAMEE=&VNAMEF=). Stewart, Jay. 2013. “Tobit or Not Tobit?” Journal of Economic and Social Measurement 38(3). Stone, Leroy O., and Catherine Pelletier. 1998. Statistics Canada Total Work Accounts System: Technical Guide to the 1998 Edition. Ottawa, ON. Sullivan, Oriel. 2006. Changing Gender Relations, Changing Families: Tracing the Pace of Change over Time. Oxford: Rowman & Littlefield Publishers, Inc. Sullivan, Oriel, and Jonathan Gershuny. 2013. “Domestic Outsourcing and Multitasking: How Much Do They Really Contribute?” Social Science Research 42(5):1311–24. Sutphin, Suzanne Taylor. 2010. “Social Exchange Theory and the Division of Household Labor in Same-Sex Couples.” Marriage & Family Review 46(3):191–206. Sykes, Gresham M., and David Matza. 1957. “Techniques of Neutralization: A Theory of Delinquency.” American Sociological Review 22(6):664–70. Szabo, Michelle. 2013. “Foodwork or Foodplay? Men’s Domestic Cooking, Privilege and Leisure.” Sociology 47(September 2012):623–38. Szabo, Michelle. 2014. “Men Nurturing through Food: Challenging Gender Dichotomies around 203  Domestic Cooking.” Journal of Gender Studies 23(1):18–31. Thompson, Linda, and Alexis J. Walker. 1989. “Gender in Families: Women and Men in Marriage, Work, and Parenthood.” Journal of Marriage and Family 51(4):845–71. Ting, Shun, Francisco Perales, and Janeen Baxter. 2015. “Gender, Ethnicity and the Division of Household Labour within Heterosexual Couples in Australia.” Journal of Sociology 1–18. Treas, Judith. 2010. “Why Study Housework?” Pp. 3–18 in Dividing the Domestic: Men, Women, and Household Work in Cross-National Perspective, edited by Judith Treas and Sonja Drobnic. Stanford, CA: Stanford University Press. Treas, Judith, and Tsuio Tai. 2016. “Gender Inequality in Housework Across 20 European Nations: Lessons from Gender Stratification Theories.” Sex Roles 74:495–511. Twiggs, J. E., J. McQuillan, and M. M. Ferree. 1999. “Meaning and Measurement: Reconceptualizing Measures of the Division of Household Labor.” Journal of Marriage and the Family 61(34950):712–24. Vanek, Joann. 1978. “Household Technology and Social Status: Rising Living Standards and Status and Residence Differences in Housework.” Technology and Culture 19(3):361–75. West, Candace, and Sarah Fenstermaker. 1995. “Doing Difference.” Gender & Society 9(1):8–37. West, Candace, and Don H. Zimmerman. 1987. “Doing Gender.” Gender & Society 1(2):125–51. West, Candace, and Don H. Zimmerman. 2009. “Accounting for Doing Gender.” Gender & Society 23:112–22. Wight, Vanessa R., Suzanne M. Bianchi, and Bijou R. Hunt. 2013. “Explaining Racial/Ethnic Variation in Partnered Women’s and Men’s Housework Does One Size Fit All?” Journal of 204  Family Issues 34(3):394–427. Williams, Fiona. 1995. “Race/ethnicity, Gender, and Class in Welfare States: A Framework for Comparative Analysis.” Social Politics: International Studies in Gender, State & Society 2(2):127–59. Windebank, Jan. 2007. “Outsourcing Women’s Domestic Labour: The Cheque Emploi-Service Universel in France.” Journal of European Social Policy 17(3):257–70. Young, Marisa, Jean E. Wallace, and Alicia J. Polachek. 2015. “Gender Differences in Perceived Domestic Task Equity: A Study of Professionals.” Journal of Family Issues 36(13):1751–81.   205  Appendices Appendix A  Variables Coding Supplement  Table A1. Variable Coding for the Canadian GSS  Variables Description 1986 1992 1998 2005 2010        Cooking DVCOOK durac_10, durac_11 dur101, dur102, dur110 dur101, dur102, dur110 dur101, dur102, dur110 DUR1010, DUR1020, DUR1100 Cleaning DVCLEAN durac_12, durac_13, durac_14, durac_15 dur120, dur130, dur140, dur151, dur152 dur120, dur130, dur140, dur151, dur152 dur120, dur130, dur140, dur151, dur152 DUR1200, DUR1300, DUR1400, DUR1510, DUR1520 Maintenance DVMAIN durac_16 dur161, dur162, dur163, dur164 dur161, dur162, dur163, dur164 dur161, dur162, dur163, dur164 DUR1610, DUR1620, DUR1630, DUR1640 Shopping DVSHOP dvshop dvshop dvshop dvshop DVSHOP Age  dvagegr dvage AGE dge age Female  dvsex dvsex SEX sex sex BornInCanada Born in Canada untry_a1 dvbornrc BRTHCAN brthcan brthcan EduInYears Education in Years dvedr dveduc EDU10, EDUYR edu10, eduyr EDU10, eduyr PaidWork Paid work durac_01 dur011, dur012 dur011, dur012 dur011, dur012 DUR0110, DUR0120 Professional engagement ProfessionalEngagement durac_60 dur600 dur600 dur600 DUR6000 Political engagement PoliticalEngagement durac_61 dur610 dur610 dur610 DUR6100 Family Engagement FamilyEngagement durac_62 dur620 dur620 dur620 DUR6200 ReligiousEngagement ReligiousEngagement durac_63, durac_64 dur630, dur640 dur630, dur640 dur630, dur640 DUR6300, DUR6400 Volunteer Engagement VolunteerEngagement durac_66 dur660, dur671, dur672, dur673, dur674, dur675, dur676, dur677, dur678 Dur660, dur661, dur671, dur672, dur673, dur674, dur675, dur676, dur677, dur678 dur660, dur661, dur671, dur672, dur673, dur674, dur675, dur676, dur677, dur678 DUR6601-DUR6605, DUR6609, DUR6610, DUR6711,DUR6712, DUR6720, DUR6732, DUR6733-DUR6735, DUR6739, DUR6740, DUR6751-DUR6754, DUR6759, DUR6760, DUR6770, DUR 6780  206  Variables Description 1986 1992 1998 2005 2010 Fraternity/Sorority Engagement Fraternity/Sorority Engagement durac_65 dur651, dur652 dur651, dur652 dur651, dur652 DUR6510, DUR6520 Other Engagement OtherEngagement durac_66 dur680 dur680 dur680 DUR6801, DUR6802 Leisure Study LeisureStudy durac_56 dur560 dur560 dur560 DUR5601, DUR5602 AttendSportsEvents AttendSportsEvents durac_70 dur701, dur702 dur701, dur702 dur701, dur702 DUR7010, DUR7020 Attend concerts AttendConcerts durac_71 dur711, dur712 dur711, dur712 dur711, dur712 DUR7110, DUR7120 Attend cinema AttendCinema durac_72 dur720 dur720 dur720 DUR7200 Attend classics AttendClassics durac_73 dur730 dur730 dur730 DUR7300 Attend museums AttendMuseums durac_74 dur741, dur742, dur743 dur741, dur742, dur743 dur741, dur742, dur743 DUR7410, DUR7420, DUR7430 Attend clubs AttendClubs durac_76 dur760 dur760, dur770 dur760, dur770 DUR7600, DUR7700 Attend other AttendOther durac_78 dur780 dur780 dur780 DUR7801, DUR7802 Leisure sports LeisureSports durac_80 dur800, dur801, dur802, dur803, dur804, dur805, dur806, dur807, dur808, dur809, dur810 dur800, dur801, dur802, dur803, dur804, dur805, dur806, dur807, dur808, dur809, dur810 dur800, dur801, dur802, dur803, dur804, dur805, dur806, dur807, dur808, dur809, dur810 UR8000, DUR8011, DUR8012, DUR8013, DUR8014, DUR8015, DUR8016, DUR8017, DUR8021, DUR8022, DUR8031, DUR8032, DUR8041, DUR8042, DUR8051, DUR8052, DUR8053, DUR8061, DUR8062, DUR8071, DUR8072, DUR8073, DUR8074, DUR8080, DUR8090,DUR8101, DUR8109 Leisure camping LeisureCamp durac_81 dur811, dur812, dur813, dur814, dur815, dur816 dur811, dur812, dur813, dur814, dur815, dur816 dur811, dur812, dur813, dur814, dur815, dudr816 DUR8110, DUR8120, DUR8130, DUR8140,DUR8150,DUR8160 207  Variables Description 1986 1992 1998 2005 2010 Leisure friends LeisureFriends durac_75 dur751, dur752, dur753 dur751, dur752, dur753 dur751, dur752, dur753 DUR7510, DUR7520, DUR7530 Leisure hiking LeisureHike Durac_82 Dur821, dur822 Dur821, dur822 Dur821, dur822 DUR8211, DUR8212, DUR8213,DUR8220 Leisrure crafts LeisureCrafts Durac_83, durac_84 Dur831, dur832, dur841, dur842 Dur831, dur832, dur841, dur842 Dur831, dur832, dur841, dur842 DUR8310, DUR8320, DUR8410, DUR8420 Leisure performance LeisurePerformance Durac_85 Dur850 Dur850 Dur850 DUR8501, DUR8502 LeisureGames LeisureGames Durac_86 Dur861, dur862 Dur861, dur862 Dur861, dur862 DUR8610, DUR8621, DUR8622 LeisureSights LeisureSights Durac_87 Dur871, dur872, dur873 Dur871, dur872, dur873 Dur871, dur872, dur873 DUR8710, DUR8720, DUR8730 OtherLeisure OtherLeisure Durac_88 Dur880 Dur880 Dur880 DUR8800 Leisure radio LeisureRadio Durac_90 Dur900 Dur900 Dur900 DUR9001, DUR9002 Leisure TV LeisureTV Durac_91 Dur911, dur912, dur913, dur914 Dur911, dur912, dur913, dur914 Dur911, dur912, dur913, dur914 DUR9110, DUR9120, DUR9130, DUR9141, DUR9149 LeisureCD Leisure CD Durac_92 Dur920 Dur920 Dur920 DUR9200 Leisure reading LeisureReading Durac_93, durac_94 Dur931. Dur932, dur940 Dur931, dur932, dur940 Dur931, dur932, dur940 DUR9310, DUR9321, DUR9322, DUR9401, DUR9402 PersInc PersInc Dvtotinc Dvperinc INCM Inr_q012 INR_Q032 Household Income HhldInc Houseinc3 Dvhinc INCMHSD Incmhsd INR_Q110 FullTime, PartTime, Other, Student  Ptft, act7days Dvspern, act7days LFSGSS Lfsgss lfsgss Own home OwnHome Ownrent K2 DWELLOWN Dwellown dwellown Married Married Dvmarst Dvms MARSTAT Marstat marstat SameSex    PRTYPE Prtype prtypec Children  Age_e1 Dvchild CHR0014, CHR1524 Chrunhsdc chrinhsdc Under5  Age_e1, age_e2, age_e3, age_e4, age_e5, age_e6 Dvagryc AGECHRY Agechryc AgeChild 208  Variables Description 1986 1992 1998 2005 2010 Household size HhldSize Dvhhldsz Dvhhscap HSDSIZE Hsdsizec hsdsizec Weekday  Dvtday Dvtday DVTDAY Dvtday dvtday ID  Seqnum Seqnum RECID Recid recid Year  1986 1992 1998 2005 2010 Weight  Fwgt_ms Fwght WGHTFIN Wght_per WGHT_PER province  Prov Dvprov PRV Prv prv English  Ethnic1 K13ac02 L17_C01 Languagech_eng LANGUAGECH_ENG French  Ethnic2 K13ac01 L17_C02 Bpr_q30, bpm_q30, bpf_q30, languagech_fre, bpr_q20 LANGUAGECH_ENG, province Chinese  Trysp_a1 K13ac07, k25a, k26a L17_C04, BRTHCD Bpr_q30, bpm_q30, bpf_q30, languagech_chi LANGUAGECH_CHI, vmpchin South Asian  Trysp_a1 K25a, k26a L17_C12, BRTHCD Bpr_q30, bpm_q30, bpf_q30, languagech_pun Vmpsasia, LANGUAGECH_PUN Filipino  Trysp_a1 K25a, k26a L17_C14, BRTHCD Bpr_q30, bpm_q30, bpf_q30, languagech_tag Vmpfilip, LANGUAGECH_TAG         Table A2. Variable Coding for the ATUS38  Variables Description 2003-2015    Cooking DVCOOK t020201, t020202, t020203, t020299 Cleaning DVCLEAN t020101, t020102, t020103, t020104, t020199, t020301, t020401 Maintenance DVMAIN t020302, t020303, t020399, t020402, t020499, t020701, t020799, t020801, t020899                                                     38 The coding procedures to obtain the ATUS sample for Stata are available on GitHub: https://github.com/Kolpashnikova/random/blob/master/codingATUS.txt 209  Variables Description 2003-2015 Shopping DVSHOP t070101, t070102, t070103, t070104, t070105, t070199, t070201, t070299, t070301, t070399, t079999, t080201, t080202, t080203, t080299, t080301, t080302, t080399, t080401, t080403, t080499, t080501, t080502, t080599, t080601, t080602, t080699, t080701, t080702, t080799, t080801, t080899, t089999, t100101, t100102, t100103, t100199, t100201, t100299, t100381, t100383, t100399, t100401, t100499, t109999, t180101, t180199, t180701, t180782, t180801, t180802, t180803, t180804, t180805, t180806, t180807, t180899, t181081, t181002, t181099 Age  teage Female  tesex BornInUSA Born in the US prcitshp EduInYears Education in Years peeduca PaidWork Paid work t050101, t050102, t050103, t050189, t050201, t050202, t050203, t050204, t050289 PersInc PersInc trernwa Household Income HhldInc Hefaminc, hufaminc FullTime, PartTime, Other  trdpftpt Own home OwnHome hetenure Married Married trsppres Children  trchildnum Under5  tryhhchild Household size HhldSize hrnumhou Weekday  tudiaryday ID  tucaseid Year  tuyear Weight  tufnwgtp State   gestfips White, Black, Asian, Native American  ptdtrace Hispanic  pehspnon    210  Appendix B  Summary of the Decomposition of the Gender Gap in the US  Appendix Figure B 1 Percent Explained of the Gender Gap in Cooking by Different Frameworks, ATUS data     Appendix Figure B 2 Percent Explained of the Gender Gap in Cleaning by Different Frameworks, ATUS data  9.9 10.35.06.312.8 12.912.8 13.46.711.216.515.17.1 7.43.910.1 10.88.3-0.1 0.0 -0.90.21.5-1.2-5.00.05.010.015.020.0All White Black Native Asian HispanicPercent Explained of the Gender Gap in Time Spent on Cooking, USRelative resources Time Availability Autonomy Gendered Processes11.0 11.05.9 6.717.5 17.219.7 20.310.415.727.0 26.97.6 7.64.02.113.99.8-0.1 0.0 -0.90.0 1.9-0.7-5.00.05.010.015.020.025.030.0All White Black Native Asian HispanicPercent Explained of the Gender Gap in Time Spent on Cleaning, USRelative resources Time Availability Autonomy Gendered Processes211   Appendix Figure B 3 Percent Explained of the Gender Gap in Shopping by Different Frameworks, ATUS data  Appendix Figure B 4 Percent Explained of the Gender Gap in Maintenance by Different Frameworks, ATUS data 10.5 11.7-1.514.7 23.9 15.244.4 45.722.8228.079.1 90.26.0 7.2-3.2-52.46.4 7.20.1 0.0 4.6 2.1-0.81.2-100.0-50.00.050.0100.0150.0200.0250.0All White Black Native Asian HispanicPercent Explained of the Gender Gap in Time Spent on Shopping, USRelative resources Time Availability Autonomy Gendered Processes-7.0 -7.71.8-12.1-15.8-10.3-16.8 -17.9-5.6-23.7-17.2-25.1-6.4 -6.9-0.2-4.6-10.0-5.50.1 0.0 0.8 0.6-1.0 -0.2-30.0-25.0-20.0-15.0-10.0-5.00.05.0All White Black Native Asian HispanicsPercent Explained of the Gender Gap in Time Spent on Maintenance, USRelative resources Time Availability Autonomy Gendered Processes212  Appendix Table B1 Explained and Unexplained Gender Gap in Domestic Tasks by Pooled Decomposition among Americans, in %   All White Black Native Asian Hispanic Cook explained 14*** (0.194) 15*** (0.210) 9*** (0.584) 16** (2.153) 19*** (1.547) 17*** (0.823) unexplained 86*** (0.489) 85*** (0.523) 91*** (1.783) 84*** (4.453) 81*** (2.870) 83*** (1.575) Clean explained 20*** (0.335) 21*** (0.374) 14*** (0.764) 11 (4.248) 24*** (1.671) 28*** (1.288) unexplained 80*** (0.813) 79*** (0.892) 86*** (2.341) 89*** (8.357) 76*** (3.457) 72*** (2.366) Shop explained 38*** (0.301) 39*** (0.328) 24** (0.910) -47 (4.535) 80*** (2.041) 59*** (1.072) unexplained 62*** (0.747) 61*** (0.807) 76*** (2.588) 147 (10.564) 20 (3.394) 41** (2.090) Main explained -14*** (0.160) -16*** (0.185) -1 (0.336) -3 (1.900) 0 (0.217) -20*** (0.392) unexplained 114*** (0.510) 116*** (0.581) 101*** (1.125) 103** (4.491) 100* (0.740) 120*** (1.006) All explained 25*** (0.688) 26*** (0.759) 15*** (1.909) 14 (8.358) 27*** (3.892) 29*** (2.315) unexplained 75*** (1.251) 74*** (1.366) 85*** (4.030) 86*** (14.680) 73*** (5.604) 71*** (3.539) Robust standard errors in parentheses; * p < 0.05, ** p < 0.01, *** p < 0.001   213  Appendix C  Supplement to the Theoretical Framework of Brines (1994) on the Domain of Relative Economic Contribution Some of the important limitations inherent to the analysis of the relative economic contribution that were not addressed by Brines (1994) are discussed in this section. The main factor used in the explanatory models has a limited domain [-1, 1] by definition. Relative nature of the factor implies that it is limited in terms of the maximum and the minimum amount of gap between the compared measures of partners’ economic contribution. The outlined theory, however, assumes that the compared measures would run along the entire real line. Therefore, some adjustments are necessary for accounting for the limited character of the relative income measure. This limitation can be represented in the following manner: ∆𝑖 ∈ [min (𝑦𝑖 − 𝑦𝑖∗𝑦𝑖 + 𝑦𝑖∗) , max (𝑦𝑖 − 𝑦𝑖∗𝑦𝑖 + 𝑦𝑖∗)] , 𝑜𝑟 ∆𝑖 ∈ [−1, 1],                        (1) Where Δi – relative economic contribution, the difference between the proportion of the respondent’s economic contribution over the contribution of the partner in the household i; yi – contribution of the respondent; 𝑦𝑖∗– the base for comparison, or the contribution of the partner. The association of relative income with the time spent on housework is  𝑇𝑖 =  𝑓(∆𝑖), where 𝑇𝑖  – time spent on housework participation of the respondent in the household i; 𝑓(∙) – the model specification, establishing the relationship between the housework participation and relative economic contribution. The specifications of association 𝑓(∙) between the relative economic contribution and time spent on housework can emulate each other. There are a few specific cases:  Case 1. If the best fit model is a quadratic model then we have to test the following. If ∃ ∆∈ [−1,1] | 𝑓′(∆) = 0, then we use the gender display explanation of the association 214  of between the relative economic contribution and time spent on housework. Else, we use the simple linear explanation. Case 2. If the best fit model is a cubic model, then we have to test the following. If ∃ ∆∈[−1,1] | 𝑓′′(∆) = 0, then we use the cumulative disadvantage explanation of the association between the relative economic contribution and time spent on housework. Else, we use the gender display model if ∃ 𝑓′(∆), and the bargaining explanation (simple linear model) if otherwise. Therefore, to establish the relationship between the relative economic contribution and housework participation on a limited domain, it is not sufficient to test the order of the specification but also it is necessary to analyze the behaviour of the function within the domain.  Considering that the consequences of choosing a quadratic function instead of cubic or a linear instead of quadratic result in major shifts in the theoretical explanation, the cases presented in this supplement present an indispensable addition to the theory developed by Brines (1994).     215  Appendix D  OLS Regression Results for the US Sample Appendix Table D1 Year Fixed-Effects Models on Log of Time Spent Cooking for American Women and Men with Heckman Adjustment, 2003-2015  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) -0.027 0.015 -0.025  -0.031 -0.012 -0.039 (0.022) (0.057) (0.022)  (0.027) (0.093) (0.027) Δ2  0.082* 0.084*   0.128** 0.125**   (0.033) (0.033)   (0.044) (0.043) Δ3  -0.051    -0.030    (0.068)    (0.101)  Control Variables             Born in the US -0.275*** -0.275*** -0.275***  -0.467*** -0.462*** -0.462***  (0.040) (0.040) (0.040)  (0.058) (0.059) (0.059)      Education (years) -0.018*** -0.018*** -0.018***  -0.011 -0.011 -0.011  (0.005) (0.005) (0.005)  (0.006) (0.006) (0.006)      Age -0.004** -0.004** -0.004**  0.001 0.000 0.000  (0.001) (0.001) (0.001)  (0.002) (0.002) (0.002)      Children -0.041 -0.041 -0.040  -0.071 -0.076 -0.076  (0.040) (0.040) (0.040)  (0.058) (0.058) (0.058)      Under 5 -0.086** -0.088** -0.088**  -0.007 -0.012 -0.012  (0.031) (0.031) (0.031)  (0.045) (0.045) (0.045)      Household Size 0.021 0.022 0.021  0.012 0.011 0.011  (0.015) (0.015) (0.015)  (0.020) (0.020) (0.020)      Weekday -0.268*** -0.263*** -0.263***  -0.347*** -0.340*** -0.340***  (0.031) (0.031) (0.031)  (0.040) (0.040) (0.040) 216   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Black 0.111* 0.111* 0.110*  0.023 0.031 0.030  (0.055) (0.055) (0.055)  (0.084) (0.084) (0.084) Native American 0.010 0.002 0.006  0.299 0.306 0.305  (0.174) (0.175) (0.174)  (0.241) (0.245) (0.245) Asian 0.111 0.112 0.112  0.051 0.059 0.058  (0.061) (0.061) (0.061)  (0.094) (0.094) (0.094) Other Non-White 0.043 0.042 0.042  -0.168 -0.169 -0.169  (0.119) (0.119) (0.119)  (0.153) (0.154) (0.153)      Year variables Yes Yes Yes  Yes Yes Yes      State level vars. Yes Yes Yes  Yes Yes Yes Constant 5.626*** 5.589*** 5.581***  6.353*** 6.295*** 6.295***  (0.266) (0.266) (0.266)  (0.376) (0.377) (0.377) ρ -.947*** -.947*** -.947***  -.936*** -.938*** -.938*** σρ -1.065 -1.064 -1.064  -1.393 -1.400 -1.399 N(uncensored) 13601(9832) 13601(9832) 13601(9832)  17561(7062) 17561(7062) 17561(7062) Log likelihood -23909.09 -23902.57 -23903.09  -26733.03 -26724.59 -26724.68 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001    217  Appendix Table D2 Year Fixed-Effects Models on Log of Time Spent on Cleaning for American Women and Men with Heckman Adjustment, 2003-2015  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) 0.048 0.124 0.055  0.045 0.306 0.036 (0.033) (0.086) (0.034)  (0.045) (0.157) (0.046) Δ2  0.122* 0.129*   0.208** 0.182*   (0.052) (0.051)   (0.078) (0.075) Δ3  -0.090    -0.313    (0.103)    (0.172)  Control Variables             Born in the US -0.305*** -0.307*** -0.308***  -0.148 -0.148 -0.147  (0.062) (0.062) (0.062)  (0.094) (0.094) (0.094)      Education (years) -0.027*** -0.027*** -0.027***  -0.046*** -0.048*** -0.048***  (0.007) (0.007) (0.007)  (0.011) (0.012) (0.012)      Age -0.007** -0.007** -0.007**  -0.004 -0.005 -0.005  (0.002) (0.002) (0.002)  (0.003) (0.003) (0.003)      Children -0.241*** -0.239*** -0.238***  0.086 0.064 0.070  (0.061) (0.061) (0.061)  (0.101) (0.103) (0.103)      Under 5 0.070 0.065 0.066  -0.155* -0.168* -0.168*  (0.048) (0.048) (0.048)  (0.068) (0.070) (0.070)      Household Size 0.015 0.016 0.015  -0.023 -0.023 -0.023  (0.022) (0.022) (0.022)  (0.033) (0.033) (0.033)      Weekday -0.346*** -0.341*** -0.341***  -0.245*** -0.209** -0.212**  (0.040) (0.040) (0.040)  (0.072) (0.073) (0.074) Black 0.167 0.165 0.163  0.040 0.067 0.059 218   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m  (0.088) (0.088) (0.088)  (0.120) (0.121) (0.122) Native American 0.049 0.033 0.039  0.210 0.255 0.251  (0.264) (0.266) (0.267)  (0.330) (0.341) (0.343) Asian 0.006 0.002 0.002  -0.081 0.149 0.149  (0.112) (0.112) (0.112)  (0.309) (0.142) (0.141) Other Non-White 0.129 0.135 0.133  0.013 -0.106 -0.100  (0.160) (0.157) (0.158)  (0.028) (0.316) (0.312)      Year variables Yes Yes Yes  Yes Yes Yes      State level vars. Yes Yes Yes  Yes Yes Yes Constant 6.050*** 5.986*** 5.973***  6.041*** 6.150*** 6.118***  (0.389) (0.388) (0.388)  (0.774) (0.774) (0.780) ρ -0.898*** -0.895*** -0.895***  -0.772*** -0.798*** -0.796*** σρ -1.298 -.1.288 -1.288  -1.205 -1.282 -1.278 N(uncensored) 13601(6599) 13601(6599) 13601(6599)  17561(3391) 17561(3391) 17561(3391) Log likelihood -21457.97 -21451.17 -21451.88  -15916.88 -15907.16 -15910.72 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001    219  Appendix Table D3 Year Fixed-Effects Models on Log of Time Spent on Shopping for American Women and Men with Heckman Adjustment, 2003-2015  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) 0.003 0.234** 0.003  0.021 0.077 0.023 (0.029) (0.074) (0.029)  (0.026) (0.097) (0.026) Δ2  0.012 0.025   -0.022 -0.027   (0.046) (0.045)   (0.044) (0.043) Δ3  -0.296***    -0.060    (0.090)    (0.104)  Control Variables             Born in the US -0.357*** -0.356*** -0.357***  -0.290*** -0.293*** -0.292***  (0.057) (0.057) (0.057)  (0.055) (0.055) (0.055)      Education (years) -0.010 -0.011 -0.010  -0.030*** -0.030*** -0.030***  (0.007) (0.007) (0.007)  (0.006) (0.006) (0.006)      Age 0.007*** 0.006** 0.007***  0.007*** 0.007*** 0.008***  (0.002) (0.002) (0.002)  (0.002) (0.002) (0.002)      Children -0.151** -0.157** -0.152**  -0.024 -0.025 -0.023  (0.052) (0.052) (0.052)  (0.061) (0.061) (0.061)      Under 5 0.153*** 0.149*** 0.152***  0.019 0.020 0.020  (0.043) (0.043) (0.043)  (0.045) (0.045) (0.045)      Household Size 0.015 0.019 0.015  0.006 0.006 0.006  (0.019) (0.019) (0.019)  (0.020) (0.020) (0.020)      Weekday -0.179*** -0.176*** -0.178***  -0.164*** -0.164*** -0.165***  (0.036) (0.036) (0.036)  (0.037) (0.037) (0.037) Black 0.077 0.081 0.077  -0.035 -0.033 -0.035 220   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m  (0.065) (0.065) (0.065)  (0.071) (0.071) (0.071) Native American 0.067 0.048 0.066  -0.050 -0.052 -0.053  (0.264) (0.264) (0.264)  (0.198) (0.198) (0.198) Asian 0.123 0.120 0.122  0.137 0.137 0.136  (0.092) (0.092) (0.091)  (0.099) (0.099) (0.099) Other Non-White -0.072 -0.067 -0.072  0.041 0.042 0.042  (0.125) (0.124) (0.125)  (0.141) (0.141) (0.141)      Year variables Yes Yes Yes  Yes Yes Yes      State level vars. Yes Yes Yes  Yes Yes Yes Constant 5.403*** 5.425*** 5.388***  6.263*** 6.284*** 6.284***  (0.363) (0.363) (0.364)  (0.379) (0.379) (0.379) ρ -0.941*** -0.941*** -0.941***  -0.907*** -0.907*** -0.907 σρ -1.224 -1.222 -1.223  -1.300 -1.300 -1.300 N(uncensored) 13601(6533) 13601(6533) 13601(6533)  17561(6595) 17561(6595) 17561(6595) Log likelihood -20554.98 -20544.38 -20554.67  -25178.05 -25177.34 -25177.67 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001   221  Appendix Table D4 Year Fixed-Effects Models on Log of Time Spent on Maintenance for American Women and Men with Heckman Adjustment, 2003-2015  Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m Independent Variables Income transfer (Δ) 0.020 0.469 0.021  0.018 -0.004 0.031 (0.459) (0.556) (0.525)  (0.086) (0.291) (0.084) Δ2  -0.094 -0.004   -0.252 -0.249   (0.361) (0.433)   (0.134) (0.131) Δ3  -0.584    0.039    (0.515)    (0.315)  Control Variables             Born in the US -0.582 -0.547 -0.583  -0.328 -0.351 -0.351  (0.580) (0.508) (0.647)  (0.187) (0.190) (0.190)      Education (years) 0.037 0.035 0.037  0.035 0.036 0.036  (0.039) (0.039) (0.039)  (0.019) (0.019) (0.019)      Age 0.000 -0.001 0.000  -0.016* -0.015* -0.015*  (0.010) (0.010) (0.010)  (0.006) (0.006) (0.006)      Children 0.155 0.130 0.155  -0.096 -0.090 -0.090  (0.323) (0.316) (0.323)  (0.164) (0.164) (0.164)      Under 5 0.135 0.109 0.136  -0.020 -0.008 -0.008  (0.375) (0.375) (0.448)  (0.140) (0.140) (0.140)      Household Size -0.069 -0.059 -0.069  -0.015 -0.015 -0.015  (0.129) (0.120) (0.134)  (0.057) (0.057) (0.057)      Weekday -0.442 -0.441 -0.441  0.164 0.158 0.158  (0.325) (0.295) (0.387)  (0.119) (0.118) (0.117) Black -0.173 -0.211 -0.172  -0.008 -0.026 -0.025 222   Women  Men  Bargaining (Linear) Model 1w Bargaining (Cumulative Disadvantage) Model 2w Gender Display Model 3w  Bargaining (Linear) Model 1m Bargaining (Cumulative Disadvantage) Model 2m Gender Display Model 3m  (0.804) (0.702) (0.888)  (0.352) (0.361) (0.361) Native American (omit.)39 (omit.) (omit.)  -0.686 -0.666 -0.665      (0.701) (0.695) (0.695) Asian -0.524 -0.511 -0.524  0.629 0.600 0.602  (0.629) (0.595) (0.643)  (0.351) (0.354) (0.355) Other Non-White -0.190 -0.164 -0.191  0.243 0.207 0.209  (0.773) (0.731) (0.796)  (0.358) (0.355) (0.355)      Year variables Yes Yes Yes  Yes Yes Yes      State vars. No No No  No No No Constant 8.005 8.649 8.647  10.791*** 11.031*** 11.036***  (9.735) (6.191) (8.588)  (1.524) (1.510) (1.509) ρ -0.717*** -0.712*** -0.719***  -0.947*** -0.947*** -0.947*** σρ -1.307 -1.288 -1.315  -2.330 -2.331 -2.332 N(uncensored) 13601(332) 13601(332) 13601(332)  17561(1398) 17561(1398) 17561(1398) Log likelihood -2357.34 -2356.387 -2357.339  -8484.275 -8480.562 -8480.578 Adjusted coefficients; Robust standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001                                                       39 The ‘Native American’ category was omitted automatically by Stata because of collinearity.  223  Appendix E  Trends for Cohabiting Canadians and Bivariate Correlation Tables for Main Continuous Dependent and Independent Variables  Appendix Figure E 1 Time Spent on Housework Among Cohabiting and Married Canadians Overall, there were 4512.62 weighted observations of cohabiting women and 4706.72 of cohabiting men in the GSS sample.  224  Appendix Table E1 Correlations of Time Spent on Housework Tasks with Main Independent Variables, GSS Sample  Variables 1 2 3 4 5 6 7 8 9 1. Cooking          2. Cleaning .245***         3. Shopping  .013*** .003***        4. Maintenance -.068*** -.060*** -.023***       5. Income Transfer -.318*** -.267*** -.075** .070***      6. Paid Work -.280*** -.281*** -.260*** -.100*** .246***     7. Personal Income -.037*** -.029*** -.010 .003 .078*** .040***    8. Education -.063*** -.066*** .015** -.022*** .062*** .141*** .056***   9. Age .018*** .005 .048*** .003 .054*** -.261*** .010 -.106***  *p < .05.  **p < .01.  ***p < .001.    225  Appendix Table E2 Correlations of Time Spent on Housework Tasks with Main Independent Variables, ATUS Sample  Variables 1 2 3 4 5 6 7 8 9 1. Cooking          2. Cleaning .179***         3. Shopping  -.009* .015***        4. Maintenance -.041*** -.034*** -.012***       5. Income Transfer -.184*** -.121*** -.029*** .019**      6. Paid Work -.230*** -.261*** -.228*** -.077*** .362***     7. Personal Income -.152*** -.100*** -.015*** .009** .784*** .302***    8. Education -.057*** -.055*** .028*** -.020*** .134*** .081*** .334***   9. Age .003 -.013*** -.013*** .027*** -.298*** -.167*** -.196*** -.066***  *p < .05.  **p < .01.  ***p < .001. 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0340502/manifest

Comment

Related Items