UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Changing early child communities McPhail, Cory 2013

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

Item Metadata

Download

Media
24-ubc_2013_fall_mcphail_cory.pdf [ 1.88MB ]
Metadata
JSON: 24-1.0074214.json
JSON-LD: 24-1.0074214-ld.json
RDF/XML (Pretty): 24-1.0074214-rdf.xml
RDF/JSON: 24-1.0074214-rdf.json
Turtle: 24-1.0074214-turtle.txt
N-Triples: 24-1.0074214-rdf-ntriples.txt
Original Record: 24-1.0074214-source.json
Full Text
24-1.0074214-fulltext.txt
Citation
24-1.0074214.ris

Full Text

   Changing Early Child Communities by Cory McPhail B.Sc., McGill University, 2001  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF SCIENCE  in  The Faculty of Graduate and Postdoctoral Studies (Population and Public Health) THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver)           August 2013  ? Cory McPhail, 2013  ii  Abstract:  Changes in the proportion of children vulnerable on the Early Development Instrument (EDI) over time can be used to identify communities with an improvement or decline in its ability to foster healthy children. Positive change communities had a significant reduction in the proportion of children in a community deemed vulnerable. Negative change communities had a significant increase in the proportion of children in a community deemed vulnerable. Communities exhibiting positive change fell above the 83rd percentile on a composite of those SES variables found to correlate with EDI vulnerability, while negative change communities all fell below the 83rd percentile. Stable communities were those with no significant change in the proportion of children deemed vulnerable, and meaningful differences were found between stable high and stable low vulnerability communities. This community typology provides a priority setting lens for where early child interventions may be most effective. A methodology for identifying and analyzing a group of Early Child Development (ECD) communities is presented. A heat map tool is created to synthesize all data relevant to community ECD. Community stakeholders have to choose and evaluate best practices for providing a stimulating cognitive and social environment for all children before they reach kindergarten. This includes universally targeted variations of pre-kindergarten programs. New Investments would be required, but there would be a financial return to governments in future health, labor, and crime outcomes.    iii  Preface:  All of the literature review, obtaining data for secondary analysis, and analysis was performed by McPhail, C.  Hertzman, C; Frankish F; McGrail K performed editing and feedback for the whole document.  Ethics approval obtained from the UBC Behavioral Research Ethics Board, #H11-00516.  iv  Table of Contents:       Abstract ..................................................................................................... ii Preface ..................................................................................................... iii Table of Contents ..................................................................................... iv List of Tables ............................................................................................. v List of Figures ........................................................................................... vi Acknowledgements ................................................................................. vii     Dedication .............................................................................................. viii 1: Introduction / Lit Review ...................................................................... 1     2: Methods .............................................................................................. 26     3: Results ................................................................................................. 31     4: Creating a Tool .................................................................................... 50     5: Discussion / Limitations / Strategic Directions ................................... 55    Bibliography ............................................................................................ 70 Appendix ................................................................................................. 74   v   List of Tables:  Table 1: Vulnerability Rate Cut-offs by EDI Scale ................................................. 27 Table 2: EDI Vulnerability Trend Community Typology ....................................... 28 Table 3: Socio-economic Characteristics of Neighborhoods that associate with Vulnerability by EDI Scale, and how they  change as Vulnerability rises................................................................... 29 Table 4: Vulnerability trend of communities with vulnerability change ..................................................................................................... 31 Table 5: Vulnerability trend of communities with vulnerability change 5% or greater .............................................................................. 32 Table 6: Positive Changers in 5% and greater ...................................................... 32 Table 7: Negative Changers in 5% and greater .................................................... 33 Table 8: Changer communities defined by critical differences............................ 34 Table 9: Top 10 Stable Low Vulnerability ............................................................. 35 Table 10: Top 10 Stable High Vulnerability .......................................................... 35 Table 11: Communities of Interest: Summary of vulnerability size     shifts over three waves ....................................................................... 36 Table 12: SES Index Z-Scores for Communities of Interest .................................. 37 Table 13: Communities of Interest by School District ......................................... 40 Table 14: Community vulnerability by school district over waves 2-4 ................ 41 Table 15: EDI Vulnerability Change Profile .......................................................... 44 Table 16: EDI Construct Validity with census variables for Comox ..................... 44 Table 17: Census variables to check from Child Atlas for    communities of interest ....................................................................... 47 Table 18: Changer communities identified with critical differences ................... 74 Table 19: Appendix, Communities Removed from Analysis ................................ 75 Table 20: Vulnerability Profile, Revelstoke .......................................................... 78 Table 21: EDI Construct Validity with census variables for Revelstoke ............... 78         vi   List of Figures:  Figure 1: Conceptual framework for school readiness and health ........................ 4 Figure 2: Common Disciplinary Approaches to Human Development .................. 5 Figure 3: Framework for the social determinants of developmental health ....... 10 Figure 4: HELP SES Index Z-Scores for Changing and Stable Communities .......... 39 Figure 5: Positive Changer Intersectoral Connectedness Chart........................... 48 Figure 6: Heat map for ECD Communities ........................................................... 52 Figure 7: Heat map for stable low vulnerability................................................... 53 Figure 8: Heat map for stable high vulnerability ................................................. 54 Figure 9: Success by 6 Program Logic Model ....................................................... 77 Figure 10: Positive Stable Intersectoral Connectedness Chart ............................ 82  vii  Acknowledgements:       I'd like to thank Clyde Hertzman, Jim Frankish, Kim McGrail and the rest of HELP and SPPH for taking me on as a student and having patience with me. Clyde passed away while I was finishing this paper, and I'll always remember the time I spent learning from him, Jim, and Kim. The ideas presented here would not have been possible without the efforts he, Jim, and Kim made to link together ideas from different fields.    viii  Dedication:    I sincerely appreciate the efforts of my parents Bruce and Mavis, my brothers Jamie and David, and my friends. Their support made finishing my degree possible.     1  Chapter 1: Introduction and Literature Review  The purpose of this project is to support a process that ensures quality experiences for all children so they reach their full potential. A protocol is developed, and evidence is collected and evaluated that will advise early child development (ECD) policy at the community level and beyond. The Early Development Instrument (EDI; Janus & Offord, 2007) is a tool with which kindergarten teachers use to collect information from children in the classroom setting. The EDI gives a snapshot of children's development aggregated at the community level. This paper will use EDI scores to identify communities where their children are doing well or poorly, stable or changing on community EDI vulnerability over time.   The feasibility of using EDI vulnerability differences as a priority setting tool for early child development policy at a community, provincial, and national level is evaluated. The EDI can be used to determine community ECD vulnerability rates that fall on spectra ranging from high to low, and from stable to changing over time. The EDI can be used to find communities of interest by identifying specific trends of vulnerability. By performing an environmental scan of these communities for their defining properties, a better understanding may be gained of necessary and sufficient community structures that play a role in the determination of health for children. This step is also crucial in validation of the EDI - do the observed community properties for ECD correspond to the EDI scores in communities of interest? If so, this understanding of community structures could guide investments of resources to minimize early child vulnerability. The development of an ECD diagnostic and planning tool could aid in that process. This research attempts three approaches to studying ECD within communities. First, the EDI is used to identify communities of interest, according to their patterns of ?vulnerability.? Second, a structure, process, and outcome framework is adopted to contrast properties of high and low, stable and changing communities. Third, a visualization tool is developed to synthesize relevant community data for ECD for practitioners, policy makers, and researchers.  2  If successful, there could be policy opportunities for programs in communities concerning those ECD properties that matter. Inequities that would otherwise grow over the life course may be able to be detected by the EDI and mitigated or removed, through mobilizing community resources around early child development programs or macro level policies. The methodology and tool will fill a gap in the literature. It may aid other EDI researchers to perform an environmental scan of the properties that matter in their regions, and couch them in a ?structures, processes, outcomes? approach to evaluate programs. A literature review explains the seminal work in establishing the EDI, and frames the EDI in a sociological perspective. The Convention on the Rights of the Child is discussed in explaining the rationale for this research. The ECD community is defined in different ways, as is the nature of its stakeholders. The ECD literature on community properties which relate to childhood vulnerability is reviewed. This gives a starting point for what type of information is required to find communities of interest with which to cross-validate EDI vulnerability scores. Community specific census data will play a role in cross-validation with EDI vulnerability rates, provided the time periods and levels of aggregation are specific enough. A conceptual framework for a tool aggregating these community properties into structures (economic and social policies and demographics), processes (ECD programs, education, and health services) or outcomes (EDI, MDI, BC health services data, high school graduation rates, rates of incarceration) is provided. Such a tool provides the next steps for program selection and evaluation for a given community. Research Questions 1. How can critical difference scores be used to identify communities of interest by change or stability of EDI vulnerability rates over time? 2. Which structures, processes, and outcomes of communities of interest are theorized to affect ECD?  3  3. What would a tool look like to display this information to allow easy interpretation of community properties of ECD communities? Literature Review:  The following literature review provides the rationale for this research, as well as the conceptual framework on which this study is based. The structure of the EDI, and definition of early child community is presented. The components of an early child community are identified, and redefined in a structure, process, and outcome framework for the purposes of the subsequent analysis. Rationale  There are several human rights documents which protect and guide ECD. The United Nations Convention on the Rights of the Child was signed in 1990. All signatory countries are bound to these articles, which increases the timeliness of indicators for child development such as the EDI. There are several articles which protect ECD as conceptualized here, but one specifically concerns education. Article 29 states education must ensure ?development of the child's personality, talents and mental and physical abilities to their fullest potential? (article 29a), and ?preparation of the child for responsible life in a free society? (article 29d). The Federal / Provincial / Territorial Early Childhood Development Agreement was passed by provinces and territories in Canada in 2000. This agreement claims government responsibility for  i) ensuring children will be physically and emotionally healthy, safe and secure, ready to learn, and socially engaged and responsible; and ii) helping children reach their potential and helping families support their children within strong communities (Lloyd, 2006). General Comment 7 is an addendum to the UN Convention of the Rights of the Child which details the obligations signatory countries have to children 0-8, and reminds worldwide governments what their responsibilities are (UN, 2005).   Pivik (2012) offers a conceptual framework for school readiness and health (Figure 2). This framework highlights the complexity of interactions between variables at different levels which affect 4  school readiness. Pivik's environment scan should signal that a clear litmus test may prove elusive for ECD evaluation. No single variable is sufficient in explaining child vulnerability, but as this analysis will show, it doesn?t have to be. A useful tool should be comprehensive in the variables it considers, but is also adequately sensitive to those variables. An instrument that could deal with the interrelatedness of key ECD variables is the desired goal. Figure 1: Conceptual framework for school readiness and health (Pivik 2012)  Guhn and Goelman (2011) describe different disciplinary conceptions of child development (Figure 2).  5  Figure 2: Common Disciplinary Approaches to Human Development (Guhn & Goelman, 2011). They conceptualize child development as focusing on a combination of the individual child (Maturational-Individual Differences Approach), the family and classroom (parents and teacher; 6  Psychological Educational-Interventional Approach), and the community (neighborhood; Sociological Approach). Each approach lends itself to the collection of different variables.  The sociological approach theorizes that the socioeconomic status or pattern of economic investment in the community can affect child development. Guhn & Goelman (2010) summarize recent support in the literature for a sociological approach to child development (Figure 2). They note Kohen et al. (1998) linked socioeconomic neighborhood characteristics, school readiness, and early child development outcomes in Canada, and Duncan et al. (2007) conducted a meta-analysis of research which found an association between school readiness and future school achievement. Structure of the EDI The EDI is filled out by kindergarten teachers for each child in her/his classroom, and measures children?s readiness for school as defined by five sub-domains. They are: physical well-being, social competence, emotional maturity, language and cognitive development, and communication skills and general knowledge (Janus & Offord, 2007).  There are also 17 sub-domains. The EDI data is filled out for each individual child in the classroom by the teacher, but then EDI data are aggregated at a neighborhood level. This group data is linked to individual school and health records. This allows us to add individual processes - such as programs and services, and outcomes -such as grades and personal health information, to the community analysis. Such a data set should provide opportunities to evaluate social policies and make new policy recommendations for communities. In this manner, EDI data empowers local decision-makers within communities by providing evidence for policy discussions with external stakeholders, and creating data sources for a coordinated provincial and national response to child development. Use of the sociological approach specifically with the EDI has been supported as well. Guhn & Goelman (2010) provide an informative overview of the relationship between socioeconomic neighborhood characteristics and aggregated EDI ratings of child development (Kershaw et al 2005), pre-7  school intervention relationships with EDI ratings (Pelletier & Corter, 2005), and linking individual EDI ratings with later academic achievement (Lloyd & Hertzman, 2009). The Manitoba Centre for Health Policy released a report containing studies that add to the validity of using the sociological approach with the EDI: Santos et al. found that lower SES correlates with birth status and to higher EDI vulnerability at age 5 (2012). The observed SES gradient among individuals was much larger in cities when compared to rural areas, and this was reflected in community vulnerability rates. Further, Santos observed a positive correlation between low birth weight and increased vulnerability on the EDI. The odds ratio (OR) of being vulnerable was 2.37 for males compared to females, and significantly higher than average for children of single mothers, mothers on income assistance, or as part of the Child and Family Service System. No significant risk in EDI vulnerability related to maternal depression or residential mobility was found. An emerging application of the EDI is as a measurement of community vulnerability. A community?s vulnerability is determined by the percent of its children falling below at least one EDI subscale cut-off score for vulnerability.  For British Columbia, Vancouver cut-offs from Wave 1 are used for each domain.  Vulnerability has been defined as the value which separates the lowest 10% of the EDI subscale score from the rest (Lloyd, 2006). Hertzman has measured and described the variability in child development between communities, ?ranging from 4 to 68% vulnerable...a 17 fold variation?(2011). Uncertainty exists regarding meaningful differences in vulnerability between groups or communities as defined by the EDI. As the number of children in the community shrinks, the ability to measure significant differences in vulnerability between groups or communities over time periods with the EDI is reduced. In order to establish and take advantage of the EDI?s applications for policy, thresholds for what constitutes a significant vulnerability shift over time within a community adjusted for the number of children needs to be determined.  Work by Forer, Guhn, & Zumbo (2011) has provided a sliding scale of 'critical differences' according to the number of EDI children in a community. 8  Critical differences work by providing a minimum threshold of vulnerability change required to deem a change significant between regions or time periods. The smaller the community, the larger the observed vulnerability shift will have to be to be considered significant. The critical differences allow us to identify communities which display a significant positive or negative trend over time (sustained over three time periods) as defined by EDI vulnerability given their sample size. In these positive changer and negative changer communities, EDI differences could indicate a significant latent variable difference. For the purposes of this analysis, negative changers are defined as communities where vulnerability rates increase three EDI waves in a row. Conversely, positive changers are defined as communities with a decrease in vulnerability rates three EDI waves in a row. It helps to think of change from the community?s perspective with these labels ? falling vulnerability is a positive change for the community, whereas rising vulnerability is a negative change. A review of the ECD community and properties that affect ECD is necessary to inform the following analysis of changer communities as defined by vulnerability on the EDI (Chapter 2), and the design of a tool to aid in future community analysis (Chapter 3). Definition of Community  Community is crudely defined as a group of people living in the same place. As a unit of analysis the community is the intersection of aggregated individual factors, social factors, economic factors, and environmental factors. Schroeder et al. have published a toolkit for change in ECD communities (2009), and write, ?[ECD] community means?schools and kindergarten teachers, early child development organizations, early childhood educators, Aboriginal organizations, municipal, provincial and federal governments.? Schroeder notes that community membership extends beyond these traditional partners to parents and grandparents, and local entrepreneurs and labor unions.  The environment in which the community is situated is also crucial for providing the quality of experiences which shape a child?s developmental trajectory. Each community is a collection of people 9  and environments which can differentially affect ECD. ?Environments influence child development independent of and in combination with a child?s biological characteristics. The nurturing qualities of the environments to which children are exposed in their earliest years literally ?sculpt? the developing brain? (Schroeder et al. 2009). Maggi (2010) notes that neighborhood characteristics can influence children?s development ?through stresses (exposure to toxins, and social and psychological conditions such as high crime rates), through social organization (role models, collective efficacy and shared values), through institutions (function of schools, police, neighborhood services, etc.) and through ?epidemic? forces (power of peer influences).?  Risk or protective factors for ECD from the individual and community can have a cumulative effect on ECD vulnerability when they converge in place and time. These factors interact at the community level, exhibiting positive or negative multiplicative relationships. For example, low SES children may get by in communities with strong program support for all children, exhibiting acceptably low vulnerability on the EDI. But in resource poor neighborhoods (without strong programs) those same low SES children may struggle, exhibiting high vulnerability. Siddiqi (2007b) offers a useful framework for integrating these different individual and community factors with the environment conceptually (See Figure 3).  10  Figure 3: Framework for the social determinants of developmental health (From Siddiqi 2007b) The authors describe levels of environment aggregation as the individual child < family < residential area (community) < region < nation < global environment. The environments are not strictly hierarchical, but overlapping and interconnected. Notably, relational communities and ECD services and programs cut across levels of aggregation. This idea of environmental aggregation was first described by Bronfenbrenner (1977). Hertzman (2010b) makes a distinction between ?residential communities (where the children and family live) and relational communities (family social ties to those with a common identity) in which they are embedded,? but notes ?both residential and relational communities offer families multiple forms of support, from tangible goods and services that assist with child rearing to emotional connections with others that are instrumental in the well-being of children and their caregivers.?  Community membership may be defined either locally through collective action or 11  externally through government representation and services. ?At the residential/locality level, both governments and grassroots organizations play a highly influential role. Many resources available to children and families are provided on a community level through local recognition of deficits in resources, problem solving and ingenuity? (Hertzman 2010b). He contextualizes residential communities within the regional and national environments. ?The influence of regional and national environments is fundamental in determining the quality and accessibility of services and resources to families and communities? (Hertzman 2010b). A method is required to determine what properties of these different communities relate to ECD, and hence should inform ECD policy.   Community properties that relate to ECD There are three categories of community properties for the purposes of this analysis ? structures, processes, and outcomes. All three types of community properties can relate to ECD, and give us categories within which to frame the community data. Economic and social properties of the community are included as structures for this analysis. These properties form the substrate for the environment within which the children develop. Economic properties include policies of investment of stakeholders along with existing and future infrastructure, and SES of residents. The social properties include social cohesion, ethnicity, transiency, and the urban/rural mix. There will be some overlap between types - some economic properties have social implications and vice-versa. It is also true that the economic properties of individuals affect the social properties that are available to them. Programs, and the coordination of those programs are the two main processes of this analysis. Health services would fit here too in future analyses as the linkages are made available. Finally, outcomes are the positive short, medium, and long term benefits of the processes, measured for this analysis primarily with the EDI. Schroeder identifies this overlap of social properties, economic properties, and programs, writing ?inequalities in child development emerge according to well-recognized factors: family income, parental education, parenting style, neighborhood safety and 12  cohesion, neighborhood socio-economic differences, and access to quality child care and developmental opportunities? (2009).  Structures -Economic Properties   The economic properties of a community which affect ECD include all policies of investment by the government, individuals, and corporations and the resulting economic consequences. This includes direct dollars invested in education and developmental health by each level of government, and the total and projected value of ECD infrastructure across sectors. Socio-economic status (SES), a composite of education, income, and employment, determines what resources individuals and families have access to within their communities. These factors can result in an economic gradient among families and individuals, presenting health inequities which affect ECD.  Policies of Investment, Existing and Future Infrastructure  Economic policies can be as macro as the annual budget from the federal government, but include policies from provincial governments, and their ministries of health, education, and family. Municipal governments, corporations, and individuals also have a part to play where they invest their money. Policies usually affect existing or future infrastructure through institutional / historical time (Hertzman 2010b). For example, the Constitution Act of Canada defined a financial framework that health and education were the responsibilities of the provinces. Health transfers in Canada from the federal government to the provincial governments are based off of further policy framework in the Canada Health Act. These policies result in health and education being handled by the provinces, and consequently result in provincially dominated health and education systems and infrastructure. Healthcare certainly qualifies as economic policy and infrastructure. It is rigorously accounted, and there is an extensive policy around its formal administration by provincial governments in Canada. My first hypothesis was to look at health care services with respect to access and quality in determining predictors of early childhood vulnerability. Those vulnerable on the EDI may have children or families 13  with health problems not receiving adequate care, including the primary caregiver. However, the data that exists suggests that health care financing in British Columbia is equitable. There are three principles of equity in the healthcare system: provision of services based on need, financing of those services from tax revenues separating contributions from use, and financing hospitals and physician services through progressive taxation (McGrail 2007). Using Gini coefficients, concentration indexes, and Kakwani progressivity indicators, McGrail found that universal publicly-funded health care (Medicare) is accessed unevenly across income groups, but that this distribution is solely the result of the positive correlation between health status and income. ?Lower-income groups use more services because they tend to have greater healthcare needs and the healthcare system generally responds to those needs? (McGrail 2007). Generally, there is broad equity in the health care system. Kids, mothers, and families are generally getting the access to care they need at the hospital and health services level regardless of take home pay. But some communities bear a disproportionate burden of disease absent non-health care buffering agents in the residential and relational community. In turn children in some communities are at greater risk of increased vulnerability during sensitive periods of development. This vulnerability appears mutable. Reducing vulnerability would be better for the individuals who would reduce their risk of poor health, and it would be better for the government that is stuck with the bill for later and more costly health services. Halfon suggests just this, that ?population-based interventions focused on shifting the risk curve for an entire population, have the potential to save more lives and improve health to a greater extent than individually focused biomedical interventions? (2010). This does not mean that health care doesn?t have a proactive role to play in the ECD of its community. Instead, interest is in linkages between healthcare and stakeholders in ECD communities. Hertzman (2010c) writes about how linking ECD programs and services to healthcare systems holds mutual benefit. ?The healthcare system already employs trained professionals, provides facilities and services, and, most importantly, is a 14  primary contact for mothers and children.? He also notes that children have more contact with healthcare than education until ages 6-8, and so these contacts through health care can be done inexpensively. There is an opportunity for greater linkages between healthcare and early child development programs.  Governments? economic policies relate to health and ECD vulnerability at the community level. Siddiqi writes ? i) More wealth and spending on health care does not yield better health outcomes, (ii) public provision and income redistribution have greater effects on population health, and (iii) the gradual development of public provision represents the buildup of social infrastructure that has long-lasting effects on health status? (Siddiqi 2007a). Asset mapping will be used to capture these economic properties of investment policy and infrastructure. Socio-Economic Status (SES) of Residents Socio-economic status can be operationalized in different ways, but usually is some combination of occupation, education, and neighborhood (APA, 2007). It is also important to clarify whether aggregated community SES, or the individual SES of its residents is the focus. Income is a function of one?s occupation. This understanding of access to resources is a materialist approach ?because of its focus on the attainment of goods and services as well as access to information and social resources as a function of quantifiable characteristics such as income? (Adler, 1994). It also notes the importance of the neighborhood environment in providing context to SES. More income relative to others allows you to exert more influence in the competition for those resources. Most importantly to this discussion, community SES has been correlated with EDI vulnerability. ?The socio-economic status of the neighborhood demonstrates the most consistently powerful effects on children?s health? (Maggi 2010). At the individual level the evidence supports a ?gradient effect? (Keating 1999). A gradient in health status develops over time that is persistent across age ranges and cultures ? not just children. Family income in the absence of buffering agents can have a relationship 15  with vulnerability in the form of resources which are able to provide shelter, nutrition, and other resources beneficial for child development. Parental education relates to individual SES in that education allows you to earn more income, and income gives you a better opportunity to obtain education. ?Studies have shown that the education level of the primary caregiver, often the mother, is of particular significance to a child?s readiness for school? (Schroeder et al. 2009). It's not just the income that education gives you - education can be a resource itself. It can allow you to acquire information, grant access to resources requiring knowledge, and can be an indicator of being socially connected. SES intersects with many social properties of a community. Parenting style, neighborhood safety and cohesion, and access to quality child care and developmental opportunities will overlap conceptually with the social resources in a community. However, the choice of neighborhood of residence and family time pressures will be affected by an individual?s SES, which in turn affects the social properties that are available to them. Transiency will be covered later, but it does relate to economic security, home-ownership, and neighborhood stability (Schroeder et al. 2009). Specific to ECD vulnerability, ?children from family backgrounds with multiple developmental risk factors will do better growing up in mixed socioeconomic neighborhoods than in poor ghetto areas? (Hertzman & Bertrand 2007). Lloyd & Hertzman (2007) found that a high income concentration at the extreme (ICE) predicted improvement in urban children?s scores over time, meaning those communities at the high end of SES exhibited improvement between kindergarten and grade 4. Income inequality is the greatest driver of ECD vulnerability. ?Post-tax, post-transfer income inequality ? and thus underlying social policies ? is a key determinant of health, and plays a significant role in education?we found no evidence for a role of educational spending or of GDP [as a determinant of health]? (Siddiqi 2012). This study focuses on what is attainable, which rules out broader resource redistribution given the uncontrollable and unregulated nature of the economy today. Making a business case with evidence 16  here may make possible broader resource redistribution in the future. Instead, a poverty reduction strategy using programs is possible, with the goal of supporting low-SES children to have similar health and education outcomes to high SES individuals as adults. Social policies that break this connection between SES and access to services will be most successful in reducing ECD vulnerability. A strong universal social safety net based on renewable funding, in addition to targeted interventions designed to bring the bottom up will best address proportionate universality. Structures - Social Properties   The social properties of a community that contribute to healthy ECD are social cohesion, neighborhood stability, and population density. These properties are interrelated with individual and community transiency, neighborhood ethnicity composition in terms of immigration and aboriginal populations, and urban and rural differences.  Social Cohesion in the context of each ECD community is how interconnected stakeholders are, and how they share information and resources to improve children?s outcomes. Cohesion speaks to cooperation, and implies shared social norms around community safety. Cohesion can also apply to subgroups in communities such as specific immigrant populations who share and provide support. ?Neighborhood safety, cohesion, and crowding are a few of the factors that may influence family practices, family psychological well-being, and thus children?s development. Research also shows that neighborhood cohesion may act to diminish the effects brought on by safety issues, as social networks may provide supportive enclaves where families and children feel safe? (Maggi 2010). Hertzman and Bertrand (2007) define a cohesive community as one that mobilizes resources formally (creates programs) and informally (treats its children like they belong there). Community belonging, which could be seen as an individual rating of cohesion with the community, showed a positive, dose response association with health behavior change (Hystad 2012). ?Community-belonging may represent greater social integration within one?s community, increased access to social support, as well as material, 17  cultural and psychosocial resources for improving one?s health? (Hystad 2012). This may reflect an individual?s ability or willingness to respond to ECD programs and policies, and is something to explore further in relation to EDI vulnerability changers. Siddiqi (2007) talks about cohesion as collective efficacy, the shared belief of collectivity; ?this includes the extent to which adults and children in communities are linked to one another, whether there is reciprocated exchange (of information, in-kind services, and other forms of support), and whether there is informal social control and mutual support.?  Stocks of social capital and collective efficacy have been shown to be positive for children and families in affluent, poor, urban, and rural communities (Siddiqi 2007). Crowding in the neighborhood or school setting relates to SES in choice of dwelling and school, but it is a social property in that it influences family practices, well-being, and thus children?s development (Maggi 2010). Lloyd (2010) found that increased proportions of children to teachers predicted decreases in urban children?s grade 4 scores. Transiency Cohesion and crowding both relate to the stability of a neighborhood. People who feel more engaged are less likely to leave, and people try to move to better neighborhoods if they don?t have enough resources to compete. The coming and going of children and families is known as individual and community transiency. A community can have a high amount of turnover, even though an individual child has remained in the same school and home environment. Conversely, an individual can have high mobility through several moves, even though their classmates may remain in a stable environment over that same time.  The literature is showing that there is an added risk of ECD vulnerability in both of these conditions. Hertzman (2007) found children who have stable neighborhood environments during their early years tend to have better development outcomes than those children who are constantly changing their place of residence. Lloyd (2010) found that residential instability predicted worsening of scores over time for children in rural neighborhoods, and suggests that ?even when rural children are 18  themselves residentially stable, they are nonetheless impacted negatively by the residential flux going on around them.? Lloyd postulates there may be a lag effect in rural neighborhood environments, or a developmental effect specific to later stages of rural children?s development. There is a relationship with SES as communities with high levels of home ownership have less transiency and lower ECD vulnerability (Schroeder 2009). Urban / Rural There are measured differences in vulnerability between urban and rural communities. The districts with the most vulnerable schools are either urban communities, or rural isolated communities with a high proportion of aboriginal children (Hertzman 2007). The concentration of immigrants is generally higher in rural communities than in urban centers, and interestingly concentrated immigration predicted increases in rural children?s grade 4 achievement scores, and lower vulnerability of urban children?s kindergarten and grade 4 scores of the EDI and FSA (Lloyd 2010). This would suggest a healthy immigrant effect as well as cohesion. With a somewhat different but related finding, Hystad (2012) found that ?the odds of personal health behavior change were significantly lower in rural locations characterized by negative population growth and a high percentage of residents receiving transfer income compared with rapidly growing urban locations with low percentages of government transfer income.? Hystad?s rural communities are shrinking, while the healthy immigrant communities described in Lloyd?s work are growing. It should be known that Hystad is describing adult ratings. Ethnicity The ethnicity of a community influences ECD vulnerability. Homogeneity within communities can act as either a protective or risk factor for vulnerability. In BC, some communities, such as East Indian-Canadian communities with a highly homogenized population, have lower vulnerability compared to the provincial average.  Some rural isolated communities with high EDI vulnerability have a high proportion of Aboriginal people (Hertzman 2007). Lloyd (2010) found that the proportion of Aboriginal 19  children in the community predicted decreases in urban children?s scores, as well as residential instability in grade 4. Conversely, the proportion of Aboriginal children in the community predicted improvement in rural children?s? scores over time (Lloyd 2010). The decrease in scores in urban communities with high transiency and aboriginal proportions could represent the stress of adapting to a new environment for Aboriginal children where they are separated from their traditional support systems in their community of origin. The increase in scores in rural communities with high proportions of Aboriginal children may signal that some positive change could be occurring in programs or collaboration around ECD. There could also be an identification bias going on, where those considered aboriginal are more likely to be labeled vulnerable. Processes - Programs Siddiqi touched on public provision as an economic asset ? in this case provision of ECD programs, as the build-up of social infrastructure. The presence and efficacy of community programs as they relate to EDI vulnerability in changer communities is reviewed. It is also important to think of programs as affecting those mutable things not normally captured by the socio-economic context. Over half the neighborhood variation in vulnerability is not explained by socio-economic context, and this should be more mutable variation because it would include such things as ?day-to-day parenting practices, the quality of local governance and resources for young children; the availability of quality early learning, child care and development programs; and the willingness of families from diverse backgrounds to co-operate in the interests of their children? (Hertzman 2010b). All of these can be captured and affected by programs. Now let us discuss program efficacy ? what ECD programs matter for reducing vulnerability in communities? Efficacious intervention programs seek to reduce physical, social, cognitive, and emotional limitations of children affected by poverty or otherwise. ?Comprehensive early childhood development programs are designed to improve the cognitive and social-emotional functioning of 20  preschool children, which, in turn, influences readiness to learn in the school setting? (Anderson 2003). The Early Child Development Network writes improvements should be sustained, ?while simultaneously reducing the immediate and future burden of disease, especially for those who are most vulnerable and disadvantaged? (2007a). Interventions are most effective when they: provide a learning experience for the children and their caregivers (Engle et al. 2007, Anderson 2003) are high intensity, high quality, of longer duration (Engle et al. 2007), are targeted towards younger and more disadvantaged children (Engle et al. 2007, Hertzman 2010c, Anderson 2003), build on different established child and health survival to make programs accessible (Engle et al. 2007, Anderson 2003), and recognize social determinants rarely act in isolation, necessitating an integrated and comprehensive multi-sectoral response (Halfon 2010, Hertzman 2010c, Anderson 2003).   The evidence supporting the benefits of ECD programs is substantial. However, data which is able to differentiate effectiveness between similar programs remains elusive. In a review of ECD programs, Anderson found that ECD programs can prevent the delay of cognitive development and increase readiness to learn, as evidenced by reductions in grade retention, placement in special education classes, and standardized tests of achievement and school readiness (2003). Despite these positive findings, Anderson found insufficient evidence to determine effectiveness in the areas of children?s behavioral and social outcomes, children?s health screening outcomes, or family outcomes.  This raises the issue of soft skills. Success in life depends on more than just the cognitive. Non-cognitive assets like ?physical and mental health, as well as perseverance, attentiveness, motivation, self-confidence, and other socio-emotional qualities ? are also essential? (Heckman, 2012a). Both cognitive and socio-emotional skills develop in early childhood and depend on the family.   The evidence Heckman uses to explain the nature of cognitive versus socio-emotional skills in developing children and what should be targeted and evaluating is compelling. He compares the later success of those with a graduate equivalency degree (GED) to that of those who completed high school. 21  A GED and high school diploma are equivalent certifications of cognitive ability. However, there are drastic differences in success between GED recipients and high school grads despite this rated similarity on cognitive abilities. High school graduates who obtain a graduate equivalency degree (GED) drop out of military and college at a much higher rate when compared to graduates. Also, their earnings, crime rates, and college graduation rates are lower and more similar to high school drop-outs who never graduated (Heckman 2012a). Heckman posits that the same lack of soft skills that caused GED recipients to drop out of high school will result in lower career success when compared to high school graduates. It is the non-cognitive, non-IQ factors such as motivation and other socio-emotional soft skills at play.  These cognitive and socio-emotional skills have an opportunity to develop at home in the family context to different degrees depending on the amount of stimulation by school age. School from kindergarten forward cannot make up for that missed opportunity during pre-school years ? children?s performance on cognitive measures after this period remains relatively constant. ?Children from disadvantaged environments typically have not received the massive doses of early enrichment that children from middle-class and upper-class families have? (Heckman 2012a). Heckman uses the study by Hart & Risley (1995) to illustrate his point. Hart & Risley found that children growing up in professional families heard an average of 2,153 words per hour, compared with children in working-class families (1,251 words per hour), and children in welfare-recipient families (616 words per hour). At age three, children in the professional families had roughly 1,100-word vocabularies, in contrast with 750 words for children from working-class families, and 500 words for children of welfare recipients.  The Perry Preschool Program, which aims at enriching the lives of three and four year-old low-income children, was first tested on African American children with IQs below 85 at age three. The reason that this program gets so much attention is because their evaluation did something that researchers "don't do" - they used a randomized control trial with children. As a result, the strength of 22  evidence trumps most all other child studies. Further, they followed their cohort to age 40 and found many significant results with important policy implications.  The Perry program is an implementation of theories of Lev Vygotsky, and is based on active participatory learning. "Children and adults are treated as equal partners in the learning process, and children engage with objects, people, events, and ideas. Abilities to plan, execute, and evaluate tasks were fostered, as were social skills, including cooperation with others and resolution of interpersonal conflicts"(Heckman 2012c).  An analysis of the Perry Preschool Program found relationships with three concepts - cognition, externalizing behavior, and academic motivation (Heckman 2012c):  Cognition primarily affects achievement tests and also affects certain labor market outcomes. Externalizing Behavior affects crime outcomes, labor market outcomes, and health behaviors. Academic Motivation boosts educational outcomes and reduces long-term unemployment. Treated individuals scored higher on achievement tests, attained higher levels of education, required less special education, earned higher wages, were more likely to own a home, and were less likely to go on welfare or be incarcerated than controls. (Heckman 2012c)   Heckman predicts a return of investment on Perry of 6-10%, which is conservative because it ignores the economic returns of physical and mental health. "Early child interventions have much greater economic and social impact than conventional interventions which target older populations such as job training, convict rehab, adult literacy programs, and police spending"(Heckman 2012a). The Perry Program has shown the most robust results for early-child interventions and it has tied processes to outcomes.  There is further support for non-cognitive approaches to interventions with children, and for those interventions to be organized at the national level. The WHO found that the mental development of stunted children who were given both food supplementation and psychosocial stimulation was about as good as that of non-stunted children (CSDH, 2008). Hertzman (2010b) found that ECD programs ?lead to better child and adult outcomes in breast-feeding, early identification of developmental delays, child 23  care, early childhood education, nutrition, parenting, community strengthening and institutional capacities such as instructional and training programs.? Mort?s work with the EDI in Canada revealed a reduction in vulnerabilities in neighborhoods between wave 1 and wave 2 where interventions were applied over a three- to five-year period (2008). These findings by Mort are compelling, but the results need to be updated, reanalyzed, and extended given two additional waves of EDI and Successby6 data has been collected, and three years have passed. Anderson (2003) sees an extension of this type of evaluation work as the next steps in differentiating good programs from bad, and in which communities to apply which lessons learned: ?Communities can assess the quality and availability of early childhood development programs in terms of local needs and resources?to advocate for continued or expanded funding of early childhood development programs.? The importance of figuring out the impact on kids cannot be overstated. Halfon writes ?because marginal differences in risk exposure early in life compound to produce large health differences over the lifespan, policies that effectively reduce risks and promote health must target the early years and be sustained across developmental transitions if they are to have greatest impact in the long term? (2010).  There is a wealth of evidence showing benefits of ECD programs, more recently focusing on the analysis of programs as outputs of both cognitive and socio-emotional skills. Evidence from Perry Preschool is high quality because it used an RCT design with a longitudinal follow-up of children. It is fundamentally important in this area of scarce and limited resources that the existing programs in place are evaluated, and determine the best course of investment for each ECD community given their unique needs. Aspects of the Perry Preschool program may be employed in some communities and not others and there is an opportunity to evaluate how they are doing in communities of interest. If quality elements of these programs are not in place there may be an opportunity to make an economic argument for their inclusion or improvement. Pivik (2012) performed an extensive environment scan of programs with strong or promising evidence which affect school readiness. 24  Processes - Coordination of Programs  The two main processes are the programs, and then the coordination that happens among them. Bohan-Baker (2002) writes of transition practices which cover some of the informal types of coordination that may occur within an ECD community. Contact with preschool families and children, home visits and activities, visits to the kindergarten, information meetings for parents and support groups - all are considered processes which could aid in reducing vulnerability at school transition. In terms of formal organization and administration of ECD programs, the US (Head Start), the UK (SureStart), and Canada (Successby6) have employed comprehensive networks with different levels of effectiveness. Approximately 30% of HeadStart programs in the US use some form of the Perry Preschool program. Historically, levels of funding have not been adequate to provide access to quality programs for the number of children who would benefit in the US (Anderson, 2003). The UK SureStart program ?has been shown to benefit social behavior, reduce negative parenting, improve home learning environments, and cut violence? (Hertzman, 2010c).  In Canada, Successby6 (2011) has developed a logic model for coordination and evaluation of programs (Appendix, Figure 7). It quantifies inputs as the financial investment, whereas the outputs are the processes of coordination and of the programs themselves.  Successby6 has also developed output measures to help gauge coordination capacity within a community. The Manager's Survey and Stakeholder's Survey are output measures which collect information on ECD tables in communities around the province.  Conclusion This review of structures, processes, and outcomes of early child communities gives us the theoretical context required to identify, measure, and describe salient features within each child?s environment.  They are theorized to be the underlying factors which affect community vulnerability on the EDI. These structures, processes, and outcomes will be used to create a tool to visually synthesize 25  those factors important to ECD during the early years in one place. Once these factors and how they change over time with the EDI are made explicit, changes in EDI community vulnerability can be attributed to certain variables and not others. As changer and stable communities are identified, the data sources outlined here are used to best describe and understand communities. In cases where community properties are stable over time, EDI vulnerability change over the same time period is more easily attributed to that community?s processes ? programs and their coordination. These community properties must be considered when choosing and evaluating programs. If not, you are leaving to chance that the measured changes in the EDI are not due to differences in program or coordination quality, but due to the effects of community structures. This isn?t the first review that has posited investments in early childhood will result in improved future health and school outcomes. What this project adds to the literature is the unique framing of those community features which impact ECD over time with an outcome measure of both cognitive abilities and soft skills at the population level. The EDI provides this measure of children at school entry. By treating the EDI as an outcome measure, it can be either the outputs from previous early child interventions or investments, or the inputs into the school system and eventually, the economy.   26  Chapter 2: Methods  Research Questions 1. How can critical difference scores be used to identify communities of interest by change or stability of EDI vulnerability rates over time? 2. Which structures, processes, and outcomes of communities of interest are theorized to affect ECD? 3. What would a tool look like to display this information to allow easy interpretation of community properties of ECD communities? Analysis Plan:  The first two research questions are covered in Chapter 3: Results. The third question will be addressed in Chapter 4: Creating a tool. Tensions exist between a structuralist approach to understanding ECD and one of program evaluation. What is valid evidence of effectiveness, measurable proof that a program is working or not? With complexity of related variables ? at the individual, community, provincial, and national level ? it is easy to become overwhelmed in disentangling influences on ECD. Program evaluation theory and temporality can be used to sidestep this issue by simplifying measurable outcomes to both the hard and soft skills in children as they enter school. All the structural understanding of a community and its properties are crucial for determining comparable communities and the nature of vulnerability therein. This understanding can and should guide program intervention policy. However, rather than get bogged down with overly complex models, the analysis should isolate a before and after treatment group and note the change between groups when evaluating interventions. "The program evaluation approach replaces traditional economic policy evaluation with the randomized controlled trial... defin[ing] the parameters of interest as summaries of the outputs of experimental interventions "(Heckman 2010). Heckman notes that this practice was started by Neyman, and later Cox, and was later popularized by Rubin and Holland. The basic rationale 27  to program evaluation theory is that the person subject to a particular policy is the same as the person who is not, except for treatment status and, possibly, the outcome associated with that treatment status.   To answer the first research question, can communities be identified by change or stability as defined by critical differences in EDI vulnerability rates, valid communities must be chosen for analysis. The sample population is all BC children. The sample frame is BC communities who have Early Development Instrument (EDI) evaluations on more than 35 students. Community vulnerability is defined by the percent of its children falling below the EDI cut-off (Table 1) on any one scale. BC cut-offs are used to assign vulnerability status. Each scale is scored from 0-10. Table 1: Vulnerability Rate Cut-offs by EDI Scale (Child Atlas 2006) EDI Scale Cut-off Value Physical Health and Well-Being 7.12 Social Competence 5.58 Emotional Maturity 5.83 Language and Cognitive Development 5.38 Communication and General Knowledge 4.72  Changes and stability in community vulnerability will be calculated between each of waves 2 through 4, with significant shifts up or down noted as defined by Critical Differences (Forer 2011). Communities will be sorted and labeled depending on the vulnerability trend I developed (Table 2).   28  Table 2: EDI Vulnerability Trend Community Typology Vulnerability trend Time period 1 (W2->W3) Time period 2 (W3->W4) 1 Positive change Vulnerability decreases (-) Vulnerability decreases (-) 2 Vulnerability decreases (-) Vulnerability stable (=) 3 Vulnerability decreases (-) Vulnerability increases (+) 4 Vulnerability stable (=) Vulnerability decreases (-) 5 Stable Vulnerability stable (=) Vulnerability stable (=) 6 Vulnerability stable (=) Vulnerability increases (+) 7 Vulnerability increases (+) Vulnerability decreases (-) 8 Vulnerability increases (+) Vulnerability stable (=) 9 Negative change Vulnerability increases (+) Vulnerability increases (+)  The number of communities exhibiting each vulnerability pattern will be noted and ordered according to overall magnitude of vulnerability shift. The differences in community vulnerability patterns will be examined. For the inclusion criteria for all questions, 397 of 459 communities had data and sample sizes > 35 in all waves so as to be included. EDI change in communities is calculated in three ways: 1.  Absolute vulnerability change between consecutive EDI waves where a 1% change in vulnerability constitutes a ?change? label for the period, and  2. Vulnerability change of 5% and above between consecutive EDI waves constitutes a ?change? label for the period, where a 1-4% vulnerability shift obtains a label of vulnerability ?stable.? Under the minimum n= 35, a 5% vulnerability shift would constitute at least one additional child as being vulnerable, which makes intuitive sense. 3. Critical Difference (Forer, Guhn, & Zumbo 2011) analysis is applied for each community, between each wave. This technique was developed to account for varying community size for the EDI, and uses simulated EDI data based on actual data patterns to determine the magnitudes of change required given a community?s size to be significantly different (or not). This is statistically the most robust measure of change. 29  Besides sustained positive and negative critical differences, there is interest in stable high and stable low vulnerability communities with no critical differences over multiple time periods. Once ?changer? and ?stable? communities are isolated, a better understanding of structures for what variables are related to vulnerability based on the individual scales flagging is possible (See Table 3, Kershaw 2006). The table has been reorganized in the same format that will be used to display the validity analysis results. Table 3: Socio-economic Characteristics of Neighborhoods that associate with Vulnerability by  EDI Scale, and how they change as Vulnerability rises (Kershaw, 2006)  Scale Expected Community Properties Physical Low income rate?, % of females in manufacturing positions?, % of males in management positions?, % of males performing no unpaid childcare ?,  Social Median family income?, % of adults performing no unpaid housework?, % lone-parent families?, % males that drive to work ? Emotional Male employment rate with children under 6 ?, % of males in management positions ?, % of males performing no unpaid childcare ?, % lone-parent families? Language Median family income ?, male employment rate with child ?, unemployment rate with children under 6 ?, % of lone-parent families ?, % of non-Christians ? Communication Home ownership rate, gender income disparity?, % of males in management positions, % using a foreign home language? Any Scale Low income rate ?, employment rate with children under 6?, % of males in management positions?, % of males performing no unpaid child care ?, % of first generation Canadians ?, % of non-migrant movers ?  How does the table work? Each variable was found to change in the direction of its arrow as vulnerability rose. The Child Atlas looked at correlations only over the first wave. To answer the second research question, update those variables over the waves of EDI, and see if they track with vulnerability change in the affected communities. Are communities of interest with measured vulnerability trends 30  cross-validated by Table 3, with expected levels of community properties related to EDI vulnerability? The Child Atlas?s ideas of relevant community properties is the starting point, but examining the structures, processes, and outcomes of an ECD community should be refined with categories from the literature review. For the third research question, all relevant data holdings must be determined. A tool is proposed to aid in the visualization, collection, and communication of this information among local, provincial, and national stakeholders. This could also be thought of as asset mapping of formal or informal mobilization of community resources (Hertzman & Bertrand 2007). Asset mapping of ECD is a way of categorizing those resources. Formal asset mapping involves an economic or explicit quantification of visible structures and processes that could contribute to ECD.  Informal asset mapping involves more intangible, harder to measure concepts such as social cohesion, support networks, and opportunities for quality childhood experiences. Schroeder et al. (2009) writes that the process of asset mapping data is valuable in bringing together community members with a common goal and celebrates the strengths of their community. ?Asset data can be used for documenting, monitoring, and analyzing community change,? programs and services,? identifying gaps in services, expanding opportunities for partnerships, and enhancing community support systems. Finally, the information can be used to help communities in organizing and advocating for policy change? (Schroeder2009). Asset mapping is used as one of the community data sources, informed by Schroeder (2009), Mort (2008), Kershaw (2007), Siddiqi (2007), and the previous literature review to frame the analysis of community properties as they may relate to ECD. A heat map visualization tool is proposed of community structure, process, and outcome asset data important to ECD, and show how it can be used to design and inform program evaluation research within communities. Once the tool exists with which to compare the structural properties of communities of interest, a priority setting and program evaluation approach may be used to guide program choice and evaluate the results.  31  Chapter 3: Results  Results I: Identification of Communities of Interest Communities are labeled according to EDI vulnerability shifts using Table 2. These are the breakdowns by absolute vulnerability (Table 4) and >= 5% vulnerability (Table 5). Table 4: Vulnerability trend of communities with vulnerability change Vulnerability trend  # of Communities Percent (%) 1 Positive change (-/-) 45 11.3 2 (-/=) 6 1.5 3 (-/+) 150 37.8 4 (=/-) 2 .5 5 Stable (=/=) 1 .3 6 (=/+) 7 1.8 7 (+/-) 100 25.2 8 (+/=) 8 2.0 9 Negative change (+/+) 78 19.6 Missing: 62 - Total: 459 100  Classifying communities by absolute change isn?t very useful in this case because many communities with stable community properties (insignificant change) are classified as changing. To rectify this, a common practice has been to use 5% change or greater as a significant positive or negative shift, and below 5% as stable (in fact a non-significant difference) as seen in Table 5.   32  Table 5: Vulnerability trend of communities with vulnerability change 5% or greater Vulnerability trend # of Communities Percent (%) 1 Positive change (-/-) 13 3.3 2 (-/=) 37 9.3 3 (-/+) 84 21.2 4 (=/-) 25 6.3 5 Stable (=/=) 64 16.1 6 (=/+) 59 14.9 7 (+/-) 46 11.6 8 (+/=) 48 12.1 9 Negative change (+/+) 21 5.3 Missing: 62 - Total: 459 100  The 5% method gives us a better separation into groups of trend types for analysis. There are also many of each positive, stable, and negative change.   Table 6: Positive Changers in 5% and greater:  Tables 6 and 7 list the communities with the largest positive and negative change respectfully. However, because there is large variability between the sizes of EDI communities, differentiating between the top 10 most and least vulnerable isn?t meaningful this way.    Community Average Vulnerability Decrease Per Period 1 Shannon Lake -13.0 2 North Glenmore/McKinley -12.0 3 Comox -12.0 4 Williams Lake ? Westside -11.5 5 Lynn Valley -10.0 6 Sidney -10.0 7 Ladner Centre -8.5 8 Rosedale/Chilliwack East -8.0 9 Thornhill -7.5 10 Burns Lake -6.5 11 Cranbrook ? North -6.0 12 Parksville -6.0 13 Vernon ? BX -5.0 33  Table 7: Negative Changers in 5% and greater:                The 5% vulnerability shift rule doesn?t take into account the size of the population, and as such doesn?t lend itself to determining significant shifts. The work by Forer, Guhn & Zumbo (2011) tries to account for this by giving us a formula for critical differences. Information on communities as classified by critical difference analysis is included in Table 8 below.  Community Average Vulnerability Increase Per Period 1   Lincoln Park 12.0 2 Aldergrove 11.5 3 Citadel Heights/Kilmer Park 10.5 4 West Point Grey 10.0 5 Burnside 10.0 6 High Quadra 9.0 7 Downtown - Vic West 9.0 8 Diver Lake 9.0 9 Guildford East 8.5 10 Stoney Creek 8.5 11 Fleetwood North 8.0 12 Delbrook 8.0 13 West Boundary 8.0 14 Lazo 8.0 15 Hatzic 8.0 16 Whalley East 7.0 17 North Cowichan 7.0 18 Port McNeill 7.0 19 Brookswood 6.0 20 Dallas/Monte Creek 6.0 21 Guildford 5.0 34  Table 8: Changer communities defined by critical differences Vulnerability trend # of Communities Percent (%) 1 Positive change (-/-) 3 0.8 2 (-/=) 35 8.8 3 (-/+) 84 8.6 4 (=/-) 22 5.5 5 Stable (=/=) 64 49.1 6 (=/+) 61 15.4 7 (+/-) 12 3.0 8 (+/=) 32 8.1 9 Negative change (+/+) 3 0.8 Non-Missing Total: 397 100 Missing: 62 - Total: 459 -  You must have significant change in vulnerability rate between all three waves regardless of direction to be classified a ?changer? community (1, 3, 7, 9). Stable communities are those with no significant change between all waves (5). Communities with change over only one of two opportunities to shift (2, 4, 6, 8) don?t lend themselves as readily for clear policy interpretation. However, specific research ideas present themselves for each type. Type 2 communities stabilize after a positive change. Type 4 communities have a positive change after remaining stable. Type 6 communities have a negative change after remaining stable. Type 8 communities stabilize after a negative change. Type 3 and type 7 communities are special because they exhibit vulnerability change over both waves but in opposite directions ? their EDI trend has actually reversed. There may be a precipitating event or community property which has strongly affected EDI vulnerabilities between waves, or it could be noise. The temporality of programs and services plays an important role in all these shifts. All changer communities, their change type, and their school district are identified in Table 9 in the Appendix.     After identifying changer neighborhoods, all neighborhoods without significant changes over any wave can be designated stable. By tabulating stable communities by EDI vulnerability, a picture of the wide gaps that exist between stable high vulnerability and stable low vulnerability communities is 35  formed, and communities of interest are identified for policy applications of EDI scores. See the stable high and low vulnerability communities in Table 9 and Table 10 respectively below. Table 9: Top 10 Stable Low Vulnerability (Change type = 5) Neighborhood Code   Name   School District    Avg Vulnerability 19.20 Revelstoke                          Revelstoke                   9.67 45.02 Dundarave                           West Vancouver               11.33 23.67 Lakeview Heights/Boucherie       Central Okanagan             11.67 35.03 Murrayville                         Langley                      12.33 23.84 Lower Mission                       Central Okanagan             13.67 79.44 Shawnigan Lake                      Cowichan Valley              14.33 44.17 Lynn Valley West                    North Vancouver              15.33 61.10 Oak Bay                              Greater Victoria             15.33 42.02 Maple Ridge East                    Maple Ridge - Pitt Meadows  15.33 44.18 Canyon Heights                      North Vancouver              15.33   Table 10: Top 10 Stable High Vulnerability (Change type = 5) Neighborhood Code   Name   School District    Avg Vulnerability 33.17 Chilliwack North                    Chilliwack                   56.33 52.21 Prince Rupert Central               Prince Rupert                52.67 82.05 Terrace - Horseshoe                 Coast Mountains              51.00 74.11 Gold Trail West                     Gold Trail                   50.33 36.18 Newton                              Surrey                       49.67 43.27 Imperial Park/Birchland             Coquitlam                    48.67 41.17 Middlegate                          Burnaby                      48.00 39.17 Sunset                               Vancouver                    47.33 73.10 North Kamloops                      Kamloops/Thompson      47.00 91.03 Fort St James                       Nechako Lakes                45.67  With this information collected, a list of communities of interest is compiled where areas are doing well or poorly, changing or not. See Table 11 below for communities of interest.   36   Table 11: Communities of Interest: Summary of vulnerability size shifts over three waves     Communities Vulnerability Average Vulnerability  by Wave Total Vulnerability Shift Average Absolute Vulnerability  Shift per wave Positive Changers  W2 W3 W4   Ladner Centre Comox NGlenmore/McKinly 27.00 36  26 19 -17 9 25.33 37 26  13 -24 12 18.33 30 19 6 -24 12 Negative Changers       Aldergrove West Point Grey Fleetwood North 42.00 32 39 55 +23 12 19.67 10 19 30 +20 10 30.00 22 30 38 +16 8 Stable Low Vulnerability       Revelstoke Dundarave L Heights/Boucherie Murrayville Lower Mission Shawnigan Lake Lynn Valley West Oak Bay Maple Ridge East Canyon Heights 9.67 12 7 10 -2 4 11.33 12 9 13 +1 4 11.67 16 11 8 -8 4 12.33 13 8 16 -3 7 13.67 13 12 16  +3 3 14.33 13 11 19 +6 5 15.33 21 14 11 -10 5 15.33 18 17 11 -7 4 15.33 14  14 18 +4 2 15.33 15 19 12 -3 6 Stable High Vulnerability       Chilliwack North Prince Rupert C Terrace ? Horseshoe Gold Trail West Newton Imperial Park/ B Middlegate Sunset North Kamloops Fort St James            56.33 54 60 55 +1 6 52.67 45 56 57 +13 6 51.00 55 45 53 -2 9 50.33 54 46 51 -3 7 49.67 48 54 47 -1 7 48.67 45 48 53 +8 4 48.00 43 49 52 +9 5 47.33 47 45 50 +3 4 47.00 44 49 48 +4 3 45.67 49 42 46 -3 6 37  Table 12: SES Index Z-Scores for Communities of Interest  Communities SES Z-Score T1 (Census 2001) SES Z-Score T2 (Census 2006) Positive Changers   Ladner Centre Comox North Glenmore/McKinley 1.26 1.14 0.91 1.20 0.38 0.98 Negative Changers   Aldergrove West Point Grey Fleetwood North -0.11 0.20 0.82 0.89 0.37 0.01 Stable Low Vulnerability   Revelstoke Dundarave L Heights/Boucherie Murrayville Lower Mission Shawnigan Lake Lynn Valley West Oak Bay Maple Ridge East Canyon Heights 0.45 0.47 0.52 0.33 0.60 0.66 0.78 1.15 1.15 1.10 0.95 1.12 0.95 0.63 1.37 1.54 1.28 1.11 1.47 1.78 Stable High Vulnerability   Chilliwack North Prince Rupert C Terrace ? Horseshoe Gold Trail West Newton Imperial Park/ B Middlegate Sunset North Kamloops Fort St James            -1.50 -1.25 -2.50 -2.60 -0.81 0.09 -0.27 0.36 -1.67 -2.40 -0.78 -0.56 -1.55 -1.92 -1.61 -1.09 -1.79 -1.42 -0.85 -0.77 _____ >= +1 z-score, _____ <= -1 z-score  The HELP SES Index scores (Forer, 2009) were constructed as a composite of available SES components that were found to correlate with EDI vulnerability. All available Census and Tax filer variables were organized into 25 themes; the variables in each theme were then factor-analyzed to reveal approximately 60 ?components.? These components were regressed on the neighborhood vulnerability rates for all EDI scales (and vulnerability on one or more scales). There were eight components that were identified as important across the EDI domains; these were combined into a 38  composite SES Index score. It makes sense that the HELP SES Index scores are so different between stable high and stable low communities as the HELP SES Index scores were designed with its relationship to EDI vulnerability in mind. What is interesting is the HELP SES Index scores for the positive and negative changers. A z-score is a tool to describe where in a normal distribution a specific value of a subject (or community) lies in relation to the values of the rest of group. A z-score of 0 indicates that the value is equal to the mean of the group, and a z-score of +1 indicates that the value is 1 standard deviation above the mean. Using time period 2 between waves 3 and 4, positive changer communities exhibited a z-score of at least 0.98, which means that positive change communities fall above the 84th percentile in terms of HELP SES scores. In negative change communities there were varied levels of HELP SES scores. They did not seem to be overly low like the HELP SES Index scores of stable high vulnerability. This makes sense ? they are still tipping and should not exhibit signs of entrenched low SES. Instead, there may be opportunities to detect a noxious exposure which is negatively affecting community EDI scores. West Point Grey for instance has a HELP SES score of 0.89, in the 81st percentile, yet has shown a negative change over time in EDI vulnerability. The strength of the HELP SES index is that it can be helpful to contextualize community EDI scores among a cross-section of most likely to be relevant variables. You can use the HELP SES index as a type of check ? are there off-diagonal communities where the level of agreement between HELP SES Index scores and EDI scores is lower than expected, and can the individual components of the HELP SES Index point us to the relevant community census properties? Using the HELP SES Index in this way to look for off-diagonals is one way to address the spiraling complexity of this task, while still trying to account for these inter-related variables. Figure 4 shows the HELP SES Index Z-Scores for changing and stable communities. Of interest is that there appears to be a cutoff at a z-score of 0.94, or the 83rd percentile. All the positive change communities have z-scores above, and all the negative change communities have z-scores below this cutoff. 39     Figure 4: HELP SES Index Z-Scores for Changing and Stable Communities  The following tables look at the school district level of aggregation. Table 13 shows us which communities have qualitative info at the school district level. Table 14 shows the variation between number of communities and EDI vulnerability within school districts. Together they show how focusing on the community level may provide more appropriate and targeted policies to improve developmental opportunities for children.   40  Table 13: Communities of Interest by School District Table 13 lists communities of interest by school district, and also lists school districts for which Mort has qualitative asset mapping and case study data. From this table, it shows communities of interest (*) that there is Mort data for. It makes sense that Mort districts would center on positive changers and stable low vulnerability because she chose districts of interest based on positive change over only wave 1 and 2. With wave 4 data, the data can be cross-validated with updated information. This table highlights how there are no data holdings on negative changers and very little on stable high vulnerability communities. Table 18 shows the variation between school districts on mean EDI vulnerability. Many programs are administered at the school district level, a good way to reach all communities. There are wide differences in vulnerability within school districts. There is also large variation in number of communities in each school district. As a result, school districts may be too broad and blunt a platform to tailor resources to individual communities. This analysis shows there is room for a community-level specificity for policy evaluation.  Communities School District Positive Changers  Ladner Centre Comox N Glenmore / McKinley Delta Comox Valley* C Okanagan* Negative Changers  Aldergrove West Point Grey Fleetwood North Langley Vancouver Surrey Stable Low Vulnerability  Revelstoke Dundarave L Heights/Boucherie Murrayville Lower Mission Shawnigan Lake Lynn Valley West Oak Bay Maple Ridge East Canyon Heights Revelstoke* W Vancouver C Okanagan* Langley C Okanagan* Cowichan Vly* N Vancouver Grtr Victoria Maple Ridge N Vancouver Stable High Vulnerability  Chilliwack North Prince Rupert C Terrace ? Horseshoe Gold Trail West Newton Imperial Park/ B Middlegate Sunset North Kamloops Fort St James            Chilliwack Prince Rupert Coast Mtns Gold Trail Surrey Coquitlam Burnaby* Vancouver Kamloops/ T Nechako Lks Mort School Districts*  SD 19 SD 23 SD 41 SD 43 SD 46 SD 51 SD 71 SD 75 SD 79 SD 85 Revelstoke C Okanagan Burnaby Coquitlam Sshine Coast Boundary Comox Mission Cowichan Vly Vancouver Isl N 41  Table 14: Community vulnerability by school district over waves 2-4 School District (# of communities n>35) # of EDI communities Mean Vulnerability Vulnerability Range Std Deviation (mean vulnerability) Alberni  4 28.33 19.00 8.67 Arrow Lakes 1 17.33 - - Boundary 2 24.00 3.33 2.36 Bulkley Valley 1 33.67  - Burnaby 19 32.68 26.00 7.83 Campbell River 5 29.40 15.33 5.97 Cariboo - Chilcotin 6 33.22 16.00 5.85 Central Okanagan 21 23.73 39.00 9.22 Chilliwack 9 36.41 35.33 11.57 Coast Mountains 5 36.67 37.00 13.89 Comox Valley 8 36.41 28.00 8.31 Coquitlam 21 26.67 33.00 8.68 Cowichan Valley 8 26.50 30.00 9.95 Delta 11 27.42 26.67 9.52 Fort Nelson 8 26.50 - 9.95 Fraser - Cascade 2 34.33 18.00 12.73 Gold Trail 1 50.33 - - Greater Victoria 16 27.17 27.33 6.95 Gulf Islands 1 21.33 - - Haida Gwaii / Queen Charlotte 1 38.33 - - Howe Sound 4 27.83 18.00 8.12 Kamloops / Thompson 11 26.64 28.33 9.97 Kootenay - Columbia 5 19.13 9.00 3.36 Kootenay Lake 4 23.67 9.00 4.19 Langley 12 25.06 32.00 9.90 Maple Ridge - Pitt Meadows 7 25.10 16.33 6.29 Mission 6 34.28 26.33 8.71 Nanaimo - Ladysmith 13 30.49 22.33 8.17 Nechako Lakes 4 36.42 19.33 8.96 New Westminster 6 27.83 20.67 8.54 Nicola - Similkameen 1 39.33 - - North Okanagan - Shuswap 6 30.78 22.00 9.88 North Vancouver 17 22.82 28.33 7.61 Okanagan - Similkameen 2 34.5 7.00 4.95 Okanagan - Skaha 4 30.42 7.33 12.18 Peace River North 5 37.07 11.33 3.72 Powell River 2 31.00 10.67 7.54 Prince George 13 29.36 32.33 8.88 Prince Rupert 3 48.22 16.59 4.07 Qualicum 5 28.93 15.33 6.02 Quesnel 2 31.00 11.33 8.01 42  Table 14: Community vulnerability by school district over waves 2-4 (cont?d) Revelstoke 1 9.67 - - Richmond 11 30.42 15.00 4.80 Rocky Mountain 3 25.67 19.33 9.68 Saanich 5 30.33 22.00 8.60 School District (# of communities n>35) # of EDI communities Mean Vulnerability Vulnerability Range Std Deviation (mean vulnerability) Sooke 7 24.19 5.33 2.08 Southeast Kootenay 6 29.94 16.00 6.01 Sunshine Coast 2 31.50 13.00 9.19 Surrey 47 31.83 35.67 8.14 Vancouver 22 35.06 42.33 10.74 Vancouver Island North 2 27.50 4.33 3.06 Vernon 8 24.87 29.67 9.72 West Vancouver 3 17.89 10.67 5.74 Total (53) 397 29.44 53.67 9.19    Table 19 (Communities Removed from Analysis, Appendix) shows communities with less than 35 children reporting the EDI. As a result, critical differences in EDI vulnerability are not possible to determine. This type of analysis must make special consideration for smaller communities, as risks of vulnerability may be greater in remote or rural communities. These smaller communities are not appropriate for generalized policy findings such to be chosen as a community of interest. However, what broad policies are generalizable to smaller communities, what targeted policies to use, and when to appropriately apply them must be determined.   43  Results II: Information on Communities of Interest  The following is a collection of the types of community data that is available on communities of interest. EDI data, census data and program info all give a different picture of propensity of a community to foster healthy children. All communities have census data, but not all have their data aggregated at the community level alongside EDI and program data. This section will show what is currently possible by highlighting data collected in specific communities. The tool design that will incorporated is a synthesis of the best available data collection techniques revealed here. A sample EDI vulnerability profile for Comox is presented. Next, the construct validity of the EDI is examined by comparing the collected community census info to the critical differences observed between EDI vulnerabilities between waves. Finally, data is included from qualitative interviews on specific community programs, policies, and funding arrangements (Mort 2008).  Table 15 provides an example of what a community change profile would look like for EDI vulnerability. It draws attention to positive and negative changes at the instrument and domain level, taking critical differences into account.    44  Table 15: EDI Vulnerability Change Profile; e.g. Comox   EDI Subscale  Vulnerability  by Wave (total n) Critical Difference for each subscale  Average Vulnerability  Total Vuln shift Average Absolute Vuln shift per wave W2 (59) W3 (57) W4 (56) - - - - 1 or more scale 37 26 13 10 25.33 -24 12 Physical 14 9 5 9 9.33 -9 4.5 Social 17 7 4 6 9.33 -13 6.5 Emotional 11 17 7 6 11.67 -4 2 Language 4 7 4 7 5.00 0 0 Communication 16 6 4 8 8.67 12 6 Legend: _____ = positive change, _____ = negative change, (number of students)  The following Table 18 is a collection the variables of community properties the child atlas (Kershaw et al., 2006) would expect to relate to vulnerability change as measured on the scales of the EDI. The social, emotional, communication, and any scale were flagged as vulnerable so special attention should be given to them. The included scan of census info can be checked. Because this is a positive changer and vulnerability dropped, the values of census variables are expected to be in the opposite direction of the arrows (which is the direction census variables are found to correlate with rising vulnerability). If the properties differ in the direction below or above the provincial average expected, validating structures of ECD, they are colored green. If they differ in an unexpected direction, disconfirming the model they are colored red. Note that this table changes for the look at a negative changer, where vulnerability rises and properties are expected to move in the direction of the arrow. Table 16: EDI Construct Validity with census variables for Comox Scale Expected Community Properties Physical Low income rate?, % of females in manufacturing positions?, % of males in management positions?, % of males performing no unpaid childcare ?,  Social* Median family income?, % of adults performing no unpaid housework?, % lone-parent families?, % males that drive to work ?   45  Table 16: EDI Construct Validity with census variables for Comox (cont?d)  Emotional* Male employment rate with children under 6 ?, % of males in management positions ?, % of males performing no unpaid childcare ?, % lone-parent families? Language Median family income ?, male employment rate with child ?, unemployment rate with children under 6 ? , % of lone-parent families ?, % of non-Christians ? Communication* Home ownership rate?, gender income disparity?, % of males in management positions, , % using a foreign home language? Any Scale Low income rate ?, employment rate with children under 6?, % of males in management positions?, % of males performing no unpaid child care ?, % of first generation Canadians ?, % of non-migrant movers ?  Legend: _____=  expected, supports vulnerability reduction observed white / no color = at about the provincial average  _____= unexpected, may not support vulnerability reduction observed N/A - no census data available *-critical difference on scale over time   In order to validate the child atlas, how these tables would play out for all changer and stable communities should be examined. This information is combined into table 22. If the community census values are lower or higher than the provincial average in the direction expected based on the child atlas, they are colored green. If the community's values are not in the direction hypothesized by the atlas, they are colored red.   Mort covered the school district of Central Okanagan, in which the community North Glenmore / McKinley is located. See Figure 5 for Central Okanagan?s intersectoral connectedness chart (Mort, 2008). The specific structures, processes, and outcomes of communities within the school district of Central Okanagan are described as identified by Mort (2008). The community of interest North Glenmore / McKinley draws on this intersectoral network. While the following data on community properties applies to North Glenmore / McKinley, it reflects the whole school district rather than only 46  the community. For better community specific analysis, specific program and site usage for North Glenmore / McKinley should be obtained.  Table 17: Census Variables to Check from the Child Atlas  47  Table 17: Census variables to check  PC1 PC2 PC3 NC1 NC2 NC3 SL1 SH1 Census variables to check from Child Atlas for communities of interest  As EDI Vuln rises Ladner Centre Comox NGlenmore / McKinley Aldergrove West Point Grey Fleetwood North Revelstoke Chilliwack North Income          Median family income ? ________ ________  ________ ________ ________  ________ Home ownership rate ? ________ ________ ________ ________ ________ ________ ________ ________ Low income rate ? ________ ________ ________ ________ ________ ________ ________ ________ Gender income disparity ? ________ ________ ________ ________ ________ ________ ________ ________ Employment          Male employment rate, w children <6 ? ________ ________ ________  ________ ________  ________ Male employment rate, w children any age ? ________ ________ ________ ________ ________ ________ ________ ________ Unemployment rate, w children < 6 ?  ________ ________ ________ ________ ________   Employment rate, w children < 6 ? ________    ________ ________  ________ Occupation          % of females in manufacturing positions ? ________ ________ ________ ________ ________  ________ ________ % of males in management positions ? ________ ________ ________ ________ ________ ________ ________ ________ Domestic work          % of males performing no unpaid childcare ? ________   ________ ________ ________ ________  % of adults performing no unpaid housework ? ________ ________ ________   ________ ________  Family structure          % of lone-parent families ? ________ ________  ________ ________ ________  ________ % of males who drive to work ? ________ ________ ________ ________ ________ ________  ________ Immigration          % of first generation Canadians ? ________ ________ ________ ________ ________ ________ ________  Residential Stability          % of non-migrant movers* ?         Group membership          % using a foreign home language* ? ________ ________ ________ ________ ________ ________ ________ ________ Legend: _____=  expected, supports vulnerability increase observed _____=  unexpected, may not support vulnerability increase observed white / no color = not different than provincial average 48   Figure 5: Positive Changer Intersectoral Connectedness Chart (Mort 2008)   49  Structures Leadership / Policy Mort surveyed several leaders within the school district for their input into why ECD was flourishing, what they were doing right for children in their community. A senior district official attributed the success to decision making. Citing limited resources and limited time, the official thinks evaluation is critical targeting staffing, resources, special education, learning assistance and early learning (Mort, 2008). When asked what policy advice should be given to other school boards, there were clear policy goals to try to achieve to ensure success. A plan with guidelines, structures, and processes to evaluate programs and create indicators for trustees to use to guide decisions is the starting point. It is also important to have the staff and money resources that you can afford to free staff up to research and administer the programs in question (Mort, 2008). Economic Properties  Funding sources for the district of Central Okanagan include the Ministry of Education, the Early Literacy Grant, and Community LINK. Success by Six provides $240,000 in yearly funding (Mort 2008). Infrastructure: Intersectoral coalition partners include SD 23, MCFD, CATCH (Community Action Towards Children?s Health), Kelowna Child Care Society, the Okanagan Child Development Centre, Interior Health, and Success by Six. SES: The median family income stands at $51,136 (which is below the provincial income average). Only 11.6% of adults have university degrees in this district (which is also below the provincial average). Social Properties: 50  Mort?s survey didn?t capture school district specific data on ethnicity, social cohesion, and transiency. Processes:  Mort?s survey was effective in collecting program info from within the school district. The programs and relevant info is included below (Mort 2008): 1. Nutrition Program. It provides a newsletter for parents and warm lunches and snack packs for some families. (During the September 2007 - April 2008 period 7500 snacks were provided to over 200 children)  2.  Wellness Program. The children were provided with healthy snacks and physical activity. The children are ?actively engaged in programs? rather than being disorganized. ?The result for the children attending the centers was fewer discipline problems and fewer children going home.? It supports over 100 children at 5 daycare centers.  3. Literacy initiatives. Each PALS program ran for 6-10 sessions at five centers. At each session, literacy stations were set up for the participants.  4.  Drop and Play. Eight families use the drop and play sessions held at the YMCA. The sessions are designed for families with children aged 18 months to three years.  5.  Parent Talk. The Narrator spoke of four sites for ?Parent Talk? sessions. Each of the sites targets a different age. Parents determine what happens at each session. Guests are brought in to help. The sessions are help weekly and about 40 families use this service.  6. Attachment. A ?Mothers and Children Bond Group? helped families bond with their children. M?tis families also held a playgroup with 12-13 families involved. Home visits were combined with modeling and visits to a centre. The West Bank First Nations ran a small program called ?Make a Connection? where self-esteem and bonding were focuses of the sessions, and elders played a large role.     51  Chapter 4: Creating a Tool   The information collected on communities of interest frames the design of a heat map tool to quickly visualize the barriers to healthy ECD. The heat map can be used to facilitate discussion among community stakeholders, academics, or upper level decision makers. It provides the context for understanding what critical differences on the EDI could relate to, and can help guide future program selection and evaluation policy for communities. Dr. Ziba Vaghri used heat maps on international reporting of birth registration. By using colors to reflect status levels for a specific variable, many variables can be represented on one map.  Thus, heat maps allow one to see a complex amount of information in a comprehensive and intuitive fashion. They can also be used to plan policy research by visually highlighting gaps or priorities (Figure 6). Figure 7 shows an example heatmap using a stable low vulnerability community, and Figure 8 shows an example heatmap using a stable high vulnerability community.    52  Figure 6: Heat map for ECD Communities  Template - Cohort X, Community Y   Structures  (Economic and Social Properties)                        Processes Coordination of Programs (Outputs) Successby6 Council ECD Manager + Stakeholder Survey (2012->) (Output Measure)                                                 Program (s) (Outputs)       Dose (eg Goosetrax)(Output Measure) Program(s) Evaluation (Output Measure) Logic: CAPC/Children First/ Successby6 _____= Evaluation not performed _____= Evaluation performed with some evidence   _____= Evaluation performed with strong evidence  Legend (unless defined in box): _____= at healthy level, associated with low EDI vulnerability in communities _____= at moderate level _____= vulnerable, associated with high EDI vulnerability in communities         Outcomes  Early Development Instrument (EDI) (Outcome measure) Critical difference* / ( # of shifts) / direction of shift One scale or more *(2) ?    Physical *(1)  ?    Social -    Emotional -    Language *(1) ?    Communication -   Linked Life Course Outcome Measures FSA MDI-4 MDI-7 School grades High School completion rates College + University Acceptance Income tax paid Health Outcomes + Hospital Service Use Incarceration Rates     Intersectoral Partners         Investments (Inputs)         Connectedness Chart SES index  Unemployment rate  Occupation - Management                                            Males driving to work  Males performing no  unpaid child care   Adults performing no  unpaid housework              Median family income Home ownership rate % low income % of income from government  transfers Gender income disparity Population Total / Age  0-4 Population Change (5 year) Lone parent families Mobility Status (1 year/5 year) Ethnicity % using foreign language 53   Figure 7: Heat map for stable low vulnerability Example (Revelstoke, 2001)         Structures    (Economic and Social Properties)                              Processes Coordination of Programs (Outputs) Successby6 Council ECD Manager + Stakeholder Survey (2012->) (Output Measure)                                                 Program (s) (Outputs)       Dose (eg Goosetrax)(Output Measure) Program(s) Evaluation (Output Measure) Logic: CAPC/Children First/ Successby6 _____= Evaluation not performed _____= Evaluation performed with some evidence   _____= Evaluation performed with strong evidence  Legend (unless defined in box): _____= at healthy level, associated with low EDI vulnerability in communities _____= at moderate level _____= vulnerable, associated with high EDI vulnerability in communities                Outcomes  Early Development Instrument (EDI) (Outcome measure) Critical difference* / ( # of shifts) / direction of shift One scale or more -    Physical -    Social -    Emotional -    Language -    Communication -   Linked Life Course Outcome Measures FSA MDI-4 MDI-7 School grades High School completion rates College + University Acceptance Income tax paid Health Outcomes + Hospital Service Use Incarceration Rates         Intersectoral Partners         Investments (Inputs)         Connectedness Chart SES index  Unemployment rate  Occupation - Management                                            Males driving to work  Males performing no  unpaid child care   Adults performing no  unpaid housework              Median family income Home ownership rate % low income % of income from government  transfers Gender income disparity Population Change (5 year) Lone parent families Mobility Status (1 year/5 year) Ethnicity % using foreign language 54   Figure 8: Heat map for stable high vulnerability Example (Chilliwack North, 2001)         Structures    (Economic and Social Properties)                              Processes Coordination of Programs (Outputs) Successby6 Council ECD Manager + Stakeholder Survey (2012->) (Output Measure)                                                 Program (s) (Outputs)       Dose (eg Goosetrax)(Output Measure) Program(s) Evaluation (Output Measure) Logic: CAPC/Children First/ Successby6 _____= Evaluation not performed _____= Evaluation performed with some evidence   _____= Evaluation performed with strong evidence  Legend (unless defined in box): _____= at healthy level, associated with low EDI vulnerability in communities _____= at moderate level _____= vulnerable, associated with high EDI vulnerability in communities                Outcomes  Early Development Instrument (EDI) (Outcome measure) Critical difference* / ( # of shifts) / direction of shift One scale or more -    Physical -    Social -    Emotional -    Language -    Communication -   Linked Life Course Outcome Measures FSA MDI-4 MDI-7 School grades High School completion rates College + University Acceptance Income tax paid Health Outcomes + Hospital Service Use Incarceration Rates     Intersectoral Partners         Investments (Inputs)         Connectedness Chart SES index  Unemployment rate  Occupation - Management                                            Males driving to work  Males performing no  unpaid child care   Adults performing no  unpaid housework              Median family income Home ownership rate % low income % of income from government  transfers Gender income disparity Population Change (5 year) Lone parent families Mobility Status (1 year/5 year) Ethnicity % using foreign language 55  Chapter 5: Discussion, Limitations, and Strategic Directions  Discussion   It is the purpose of this project to use sustained change to community EDI vulnerability to identify ECD communities of interest, collect and analyze existing community data, and to conceptually synthesize information on communities in a heatmap tool. Let us first discuss the research questions in order, and the implications of the findings. Next strategic recommendations for policy, practice, and research, as well as the limitations of this analysis are discussed.  Question 1: How can critical difference scores be used to identify communities of interest by change or stability of EDI vulnerability rates over time? Of the three methods of categorizing change in vulnerability within each community summarized in Results I, the critical difference method (Forer, Guhn, & Zumbo 2011) proved to be the most fruitful in conceptualizing change, and most meaningful in categorizing communities. Of 397 valid communities, three communities had sustained positive, and three had sustained negative change to community vulnerability over three waves. This gave us the first of six communities of interest. The second use for critical differences was to obtain those communities with stable vulnerability (no significant change over three waves). Stable communities can be looked at over a spectrum of vulnerability, and the communities with stable high and stable low vulnerability can be highlighted. The requisite community properties to promote healthy ECD in communities undergoing population change may be detected, and policies that promote those properties in comparable communities can be developed.  Question 2: Which structures, processes, and outcomes of the communities of interest affect ECD? This analysis integrated a literature search into the properties which affect ECD. Mort provided detailed info on community properties through a survey to child ECD leaders, and a sample of the qualitative evidence was provided in this analysis. However, the level of depth achieved for Mort communities does not extend to all communities of interest ? most notably the negative changer and 56  stable high vulnerability communities. To better determine community properties that relate to child vulnerability, a deeper focus of negative changer communities and stable high vulnerability communities of interest is needed. The attempt to validate the child atlas (table 22) holds promise conceptually, but falls short with current data sources. The determinants the Child Atlas suggests and the direction the determinants should move or be observed relative to some cutoff (in this analysis the provincial average) would benefit from more communities, with better agreement between EDI and census levels of aggregation. Without the variance of census variables, determining significant differences between each community and the provincial average was not possible. Further investigation would benefit from more regular and precise census info, at the same time point and level of aggregation as the EDI community data. The HELP SES Index is one promising way to reduce redundancy of SES variables related to the EDI.  Question 3: What would a tool look like to display this information and more easily interpret community properties of ECD communities?  There is a real problem with information overload when performing an environment scan of salient features of an ECD community. The heat map proposed here quickly draws attention to gaps or deficiencies in the system, and provides a starting point for meetings with stakeholders. The heat map can also serve as a tool for relating discrete interventions to community change over time.  Recommendations Practice   Community ECD tables which bring together ECD stakeholders including government planners and decision makers, should use a piloted version of the heat map designed here to aid in discussion, planning, and selection and evaluation of programs and services. The heat map should be computerized, linked with other health and education records, and made available to everyone. Dissemination to 57  professionals from education, health, and government should be followed by consultations on how the heat map could inform practice in each field. Policy Community stakeholders including local, provincial, and federal governments from health, education, and family sectors have to consider choice and evaluation of best practices for providing a stimulating cognitive and social environment for all children before they reach kindergarten. This includes targeted and/or universal variations of the Perry pre-school program. New Investments would be required, but there would be a financial return to governments in future health, labor, and crime outcomes.  Research  There are next steps for analyzing change in community EDI vulnerability over time. Changes in community EDI vulnerability over time can be used to: 1. Identify and describe individual communities for program selection and evaluation.   To begin to understand what is going on in any community with respect to child development, the data for structures, processes, and outcomes outlined in the heatmap must be collected. The snapshot of the community that the heatmap provides may be used to affect policy by being shared among ECD community stakeholders. The success of any intervention will be related to the underlying intersection of relevant community properties. The relevance of programs used will be validated by ensuring they address weaknesses or gaps to the ECD system found in the heatmap. Future program choice can be selected based on the EDI domain or sub-domain that is flagging as vulnerable. Also, if changes in EDI vulnerabilities are observed, the first place to check for shifting properties within the community that may have been related to the observed change would be to compare heatmaps over waves of EDI data. This discrete intervention model fits with the ?everything in the community matters? model implicit in the heatmap by using the heatmap as a validity check. By ensuring the variables in the heatmap remain relatively stable over the waves in which an intervention was applied and critical 58  differences in EDI vulnerability were observed, it is much easier to conclude the observed changes were the result of the intervention rather than a change in the community?s composition. 2. Examine and evaluate the presence, breadth and quality of coordination in EDI communities.   There are two main categories of processes in the analysis - programs, and the coordination of those programs. While the programs and their quality is the unknown and can be affected through investment and policy, those programs do not operate in a vacuum. The quality of programs depends on the presence and ability of a community's coordinator and their early child table. Their ability to organize meetings of ECD stakeholders, to secure sustainable resources, and create, link, and sustain a complement of community appropriate programs across sectors could have as much affect on the success of programs in reducing the proportion of children on EDI as the programs themselves. Only by examining measures of community ECD capacity in the context of their program blend - perhaps controlling for it - can it be determined how necessary and sufficient both coordination and programs each are for reducing ECD vulnerability. Objective output measures for community coordination are found on the ECD Manager's Survey and Stakeholder survey (Successby6, 2011). The number of meetings, hours of coordination time, and years of experience for the child coordinator are continuous variables from the survey which may be obtainable at the community level. Linear regression could determine how much effect on community EDI vulnerability levels was observed with increases to number of meetings, hours of coordination, and years of experience of coordinator in both single and multiple linear regression models. These results could be integrated into the heatmap in the appropriate section and color codes developed for tipping points. Values for community of interest are compared to determine if there was a significant difference in coordination between positive and negative changers, stable high vulnerability and stable low vulnerability. If coordination of ECD programs presents a significant effect on community vulnerability rates, positive changers would be expected to have significantly more coordination time than the average community, and even more coordinating time 59  than negative changers. There are other more specified questions that may prove useful in composing a "coordination index," but the same idea remains of finding output measures to quantify the processes of coordination around children's programs. 3. Gain an understanding of structure of necessary and sufficient community properties and   tipping points which relate to mutability of ECD community vulnerability as defined by    the EDI.   The EDI Critical Differences are used to isolate communities with changing or stable, high or low vulnerability. By using the Heatmap for ECD Communities (Figure 6) positive changers and stable low vulnerability can be compared to those of negative changers and stable high vulnerability. Among BC communities, there are three positive changer and three negative changer communities.  The census and program information on the other BC changer communities could prove invaluable in isolating differences between successful and unsuccessful interventions. Without a fully fleshed out comparison group ? census and outcome data on negative changer and stable high vulnerability communities, it cannot be said what community properties are working and which are not. Qualitative input from local Successby6 coordinators if present could be captured from an interview using a heat map. The structural analysis would be much stronger incorporating more changer communities from other Canadian provinces and countries like Australia where three waves or more of data has been collected. Obtaining access to community EDI data from all provinces would be ideal. As BC was a relatively early-adopter of EDI data, there will be an opportunity to see the changer communities over a fourth wave, and to apply that knowledge as three waves become available across the rest of Canada and at some international EDI sites. It would be helpful to have a dynamic computerized version of the heat map on every community, not just changer or stable communities of interest. Community stakeholders could use and update their specific community information as well as contextualize it with other communities. Publicly available census information could be integrated to the level of aggregation required automatically, and new 60  neighborhood groupings have already been harmonized between EDI and census data collection going forward in BC.  "Standard analytic methods such as linear regression cannot do justice to the complex relationships among these variables. Their impact on the SES-health gradient may therefore be best described by statistical methods such as regression trees and GOM that can disentangle the effects of variables that co-occur and interact" (Adler, 1994). There is a comprehensive understanding of what properties tend to be high or low depending on community vulnerability on the EDI. The challenge lies in finding a statistical technique more robust than simple correlations to disentangle what community properties relate to EDI vulnerability. Some method of risk profiling may prove useful. "Critical properties" or tipping points may be defined where dipping into certain levels is defined with an odds ratio for EDI vulnerability on one or more scale. This may provide the basis for the color coding of community properties on the heat map. The heatmap and software could link the timeframe (e.g. age 3-4 years), process (e.g. Perry Preschool) and outcome (e.g. EDI). As you click on different outcome measures, the relevant developmental window, structure, and process info could be displayed with appropriate heat map codes. 4. Construct systematic interventions and evaluations targeted at community typologies.  To credibly demonstrate the effectiveness of a program - you must obtain evidence that it causes the measured effects. A program evaluation approach tests a program by using a randomized controlled trial, and assigning one community or group of communities to a treatment and one without. Because it is unethical to withhold effective interventions from children, a time-lagged RCT should be employed. Communities are randomly assigned to treatment or control of an intervention designed to reduce the proportion of children in a community vulnerable on the EDI. The control community will receive the intervention in a future cohort, when it becomes another treatment group. 61  By using the communities of interest as the sample for community selection, hypotheses for program evaluation can be explored. Communities of the same type as defined by critical differences on the EDI can be grouped together and compared against other types. What will also be valuable is evidence of a particular program causing positive changes in specific community types ? e.g. stable high vulnerability or negative changer communities. Such a program may work in other communities of the same type provided they have a similar heatmap. Positive Changer  Communities have displayed a sustained reduction in EDI vulnerability, and so community properties and programs introduced and sustained leading up to that period will be of interest. The programs were likely efficacious, but without random assignment it cannot be said for sure. For instance - there is also evidence where similar programs were introduced in stable high vulnerability communities. The fact that its vulnerability is mutable may prove important - it may have enough assets or sufficiently passed important tipping points of community properties to shift. The programs used here during sustained vulnerability reductions may prove to be ideal candidates to be tested in negative changer and stable high vulnerability communities. The point is to ?extract? from positive changer communities the patterns of change at the structure and process levels that are associated in time with EDI vulnerability reduction. Negative Changer  Communities have displayed a sustained increase in EDI vulnerability. This should flag a community for analysis and intervention, as stable high vulnerability may not have set in yet. Community properties may be at or near tipping points where intervention has greatest chance of impact. The heatmap will make that visually clear based on how its color coded, and should prescribe a specific intervention based on the community type, heat map, and specific EDI vulnerability flagged.  Stable High Vulnerability 62   Communities have displayed a stable high vulnerability over time. Despite measured deficits over multiple waves, there hasn't been a significant reduction in EDI vulnerability. This may provide an opportunity in finding gaps in program coverage, or poor efficacy or fit of existing community programs. This group would be the most in need in terms of having the largest proportion of children vulnerable at a stable level over three waves. Table 13 shows how this group generally has the lowest SES. These communities may have multiple ECD barriers to necessary and sufficient ECD health which could limit measured gains from initial or non-renewable interventions not sensitive to the community type. Specialized programs may exist which ?unstick? the factors holding EDI vulnerability stable between waves explicitly for stable high vulnerability communities. Stable Low Vulnerability  Communities that display a stable low vulnerability provide an opportunity to see the properties which relate to healthy ECD - healthy tipping points. Policies and programs employed may be best quality if you simply use the low community vulnerability as evidence. However, SES has been shown to provide a protective effect on community vulnerability, and so RCT evidence of program efficacy would still be superior. This would be the community type to analyze to confirm necessary and sufficient community assets for healthy child development. It is possible that high SES pre-empts the need for other things. Due to families having relatively high SES, they are better able to fill in gaps in community resources than those of low SES. This may have prevented children from dipping into vulnerability, and from communities offering the prescribed programs and services. So stable low vulnerability communities may not have the assets required to help inform policy in changing or high vulnerability communities. Further investment in programs or services may be unwarranted when choosing between stable low vulnerability communities and the more vulnerable community types if observing proportionate universality. Of great interest however, is how much SES with how few programs or services a community needs to exist in a stable low vulnerability state. 63   The data for the Perry Preschool program show that low SES individuals who took the intervention benefit from a diverse range of improved outcomes throughout the life-course. Using critical difference analysis, a subset of negative changer and stable high vulnerability communities may be selected, where a treatment and control community are randomly assigned for each community type. A time-lagged RCT would work by administering a modernized version of the program in the treatment community of both a negative changer and stable high vulnerability community, and then administering it in both the treatment and control communities in delayed fashion in the following wave. The program's effects are measured by vulnerability shift on the scales of the EDI before and after the program was introduced. With the quantity and quality of data available, it may be possible to find data like this already present around changer communities on programs currently being administered. The heatmap in each community should be tracked over time, in order to understand the full range of direct, indirect, and unintended influence on the outcomes of interest. These findings will be confirmed and validated over time (or not) through linkages to population health and economic data. In the BC case, it may be helpful to conceptualize the package of Successby6 programs as the intervention. In the Perry program "at the oldest ages studied, treated individuals scored higher on achievement tests, attained higher levels of education, required less special education, earned higher wages, were more likely to own a home, and were less likely to go on welfare or be incarcerated than controls"(Heckman 2012a). In addition to those variables, there are also linked health records. It is important to remember that those from lower SES bear an inequitable burden of disease in BC (McGrail, 2007). In addition to Heckman's findings for Perry, the burden of disease could be distributed more equally across the SES spectrum. With the BC Linked Health Database, it can be determined if the gains in soft skills from successful interventions as measured by the EDI leads to a reduction in the inequity of distribution of burden of disease across the SES spectrum over the life course.   64  Limitations  There are some limitations with the current analysis. The EDI has the potential for scoring bias and has some validity and reliability issues. The problem of providing proportionate universality is complex, and there are some statistical issues with the census and neighborhood boundaries. There is the potential for scoring bias. It is the teacher that collects the data for the child, not the child them self. The people who collect the EDI data from children, their school teachers, are personally invested in the well-being of both their students and their community. The validity of EDI research for evaluating readiness for school could be affected by teacher knowledge of how the EDI results may be used to affect investment policy at the provincial or federal level. Knowing that low readiness for school scores over time may trigger greater monetary support from certain ministries could bias EDI scoring at the classroom level. Compounding the difficulty in keeping teacher raters unbiased and blind to the policy implications are perceptions that the EDI may be used to evaluate the performance of individual teachers. The best solution to scoring bias is to use the limited training time with all the teachers efficiently and effectively. This can be accomplished by explaining the necessary amount of information needed to accurately score the EDI, while maintaining a cooperative child development focus which supports the teachers? needs. Still, Hymel et al. notes ?correlations between EDI scores and comparison measures varied widely across teachers, suggesting considerable individual differences in teacher?s ability to evaluate school readiness relative to direct, child-based assessments? (2011). Another criticism of Hymel et al.?s was that the discriminant validity of EDI domain scores, or the ability to use individual domain scores as indicators, was not supported. However, Janus found correlations between the Language and Cognitive Development domain and direct tests of receptive language, providing some evidence of discriminant validity of EDI domain scores (2010). In light of this inconsistency, the validity of the ?vulnerable community? designation defined by one domain score below a target cut-off must be rigorously evaluated. Increasing the number of domains required below 65  cutoff to be designated vulnerable may address this concern. Regardless, the work of Hymel et al did ?support the convergent validity of overall EDI scores?indicating the EDI is more appropriate for deriving inferences at higher aggregated levels such as the community or region?(2011). This is reassuring as it is at these higher aggregated levels that policies are developed for. There is also much work to be done in linking the EDI to other population health records such as the BC Linked Health Database, the FSA, the MDI, high school completion, university rates, and taxes paid. It should be in concert with these other indicators that community policy on EDI results is formed. Another potential limitation lies in the nature of proportionate universality. A magic bullet type of universal policy recommendation to reduce vulnerability for every community on child development likely does not exist; flying in the face of the complex and unique nature of each community where the EDI is used. Realistic expectations for this analysis would include being provided with an explicit sensitivity to the types of conditions which may foster improved development for children across the socioeconomic spectrum, to further develop and evaluate community-level indicators alongside the EDI, and to propose next steps for evaluations of investments, program choice, and program coordination at the community level.  There are theoretical limitations involving power of statistics on communities when working with a population. Indeed here there are only three positive and three negative changer communities from which to choose, and there are differences among them. Are there enough shifting communities of each type to generalize results or make policy recommendations? Would Canadian and International comparisons give us the needed boost in power with enough EDI communities? Given that the EDI is being administered across Canada and the world, that may be possible. There is the issue of sensitivity with this methodology of community identification, and regression to the mean must be considered. How well does this method identify communities with real change? A traditional sensitivity analysis looks at a binary outcome ? i.e. disease and the absence of 66  disease. The change outcome has three options over each change period ? positive change, no change, or negative change. This design also incorporates multiple time points, so that change outcome is assessed over two change periods. This left us with 9 combinations of change in communities over two change periods. This complexity is far from the binary outcome common in sensitivity research. Our identification of changing communities is also affected by regression to the mean. There is some question as to whether critical differences should be used to assess change over more than two waves in their current form. In this case change was assessed separately between Waves 2 and 3, and then between waves 3 and 4. The concept of regression to the mean is simply that a value above the mean has a higher probability on the next measurement of moving towards the mean than away from it. Those communities with vulnerability rates below the mean therefore are biased towards negative change (increased vulnerability), while those communities with vulnerability rates above the mean are biased towards positive change (decreased vulnerability). As a community changes its vulnerability rate across the mean, the probability that it regresses back on the next change period is increased. It is possible that community types 3 and 7 are affected by this, that their trends are noise from random variation. The starting point of a community?s vulnerability rate also impacts the probability that a community is labeled change or stable. Those with very low vulnerability rates suffer from floor effects ? there isn?t much lower to go given the way community vulnerability is derived from the EDI. Those with very high vulnerability rates suffer from ceiling effects. This analysis could be improved if the starting position of a community?s vulnerability rate was taken into account when assessing change. If community types 1(-/-) are divided by type 3(-/+) (and type 9(+/+) divided by type 7 (+/-)) the occurrence of regressing to the mean vs. the probabilities of trending away cod be assessed. In this way regression to the mean could be controlled for by measuring it and accounting for it before selecting change communities. 67   Has Successby6 or its component programs been active long enough, widespread enough to evaluate their impact with the EDI given the timeline of vulnerability shift? While reliable information exists on the presence of Successby6 and their funding in stable low vulnerability communities, vulnerability evidence from different communities with and without SuccessBy6 is required for comparison. The proposed heat map would capture that information where program data is available, or detail the gaps in coverage. It may be the case that there is a stark contrast in dollars invested by SuccessBy6 between positive and negative, changing and stable communities.  It is important to consider how widely data is collected when for whom the results of EDI analysis are generalized. EDI data collection approaches population data in its comprehensiveness. Although there is data on ?only? 80-90% of the population of children at school entry in BC, having 90% coverage is as good as it ever gets. When results are generalized to the population, remember the EDI scores of non-participants likely differ in their distribution from those of participants.  The frequency of census data collection and the differences between EDI and census neighborhoods are limitations of this analysis. With quicker and more frequent updates to community activity and resources, vulnerability can more reliably be tied to real phenomena in the community. For instance, the Canadian 2001 Census was used in the 2006 Child Atlas to describe vulnerability patterns observed in the EDI. First, 2006 Census data should link with 2006 EDI data. Second, community info is needed more often than every five years, divided up beyond the school district level. Ideally, having census and EDI community boundaries the same, with annual updates that would allow us to monitor communities for ECD more effectively. There have been great efforts made to harmonize EDI neighborhood boundaries with census neighborhoods in BC, but those changes are only possible from wave 5 onward. This will fix this limitation in future analyses, but also introduces the one-time potential incongruency of neighborhood boundaries between waves 4 and 5. This must be considered when prospectively following change communities into wave 5 and beyond. 68   There is previous literature which defines positive and negative change on the EDI as a combination of factors (Lloyd 2006) whereas this analysis primarily relies on reducing the proportions of vulnerable children on one or more scale. Other aspects of EDI score change warrant inclusion in policy discussions to achieve proportionate universality, and could be incorporated into this type of scan and analysis. Lloyd writes:  Change in the distribution of individual scores across the entire district, increasing average scores across neighbourhoods, decreasing inequality in average scores across neighbourhoods, decreasing proportions of vulnerable children across neighbourhoods, decreasing inequality in the proportion of vulnerable children across neighbourhoods are evidence of positive change. Conversely, the reverse will signal clear evidence of negative change. (2006)    The use of average scores and equality of vulnerability provides additional insight, but this analysis was limited to change in the proportion of vulnerable children within a community. Conclusion:   The objective was to determine whether or not communities can be meaningfully differentiated based on change in EDI vulnerability, and if that categorization is useful. Communities presented themselves as exhibiting positive change, negative change, or stable EDI vulnerability. This conceptualization will be useful for evaluating community properties which are thought to affect ECD, and to identify priorities for early child development policy, practice, and research.   This work will also benefit from the further cross validation of the EDI with other linked data sets. EDI vulnerability is but one indicator of community vulnerability ? a critically important one ? but not as meaningful in absence of a greater context and validation. The EDI should be cross-validated with the BC Linked Health Database, the Foundation Skill Assessment (FSA), the middle-years development instrument (MDI), high school graduation rates, university acceptance and completion rates, taxes paid, incarceration rates, homelessness rates, and type and amount of health and social services used. With these indicators, and a better fleshed-out quantitative and qualitative understanding of communities of interest and interventions, answers to important policy questions can be formed. Does the EDI find the 69  right communities in a timely fashion to address preventable childhood vulnerability? Are these the appropriate factors related to ECD, in different communities and populations? Are the programs addressing what they were designed to, and are these findings generalizable to other community types? Answering these questions will reveal the appropriate policy levers to exert a positive influence on children?s lives over the life course.  70  Bibliography  Adler NE, Boyce T, Chesney MA. (1994). Socioeconomic Status and Health: The Challenge of the Gradient. American Psychologist. Vol. 49, No. 1, 15-24. American Psychological Association, Task Force on Socioeconomic Status (2007). Report of the APA Task Force on Socioeconomic Status. Washington, DC: American Psychological Association. Anderson LM, Shinn C, Fullilove et al (2003). The Effectiveness of Early Childhood Development Programs, A Systematic Review. Am J Prev Med; 24(3S).  Bohan-Baker M, Little PMD (2002). The transition to kindergarten: A review of current research and promising practices to involve families. Cambridge, MA: Harvard Family Research Project. www.gse.harvard.edu/hfrp/projects/fine/resources/research/bohan.html. Accessed Aug 2012.  Bronfenbrenner U (1977). Toward an experimental ecology of human development. American Psychologist, 32, 513?531.  CSDH (2008). Closing the gap in a generation: health equity through action on the social determinants of health. Final Report of the Commission on Social Determinants of Health. Geneva, World Health Organization.  CRC. Convention on the Rights of the Child. United Nations. http://www2.ohchr.org/english/law/crc.htm. Accessed Aug 2012.  Duncan GJ, Dowsett CJ, Claessens A et al (2007). School readiness and later achievement. Developmental Psychology, 43, 1428?1446.  Dunn JR (2012). Socio-economic inequalities in healthy child development: the evidence grows. Sense of community-belonging and health-behaviour change in Canada. J Epidemiol Community Health; 66.  ECDKN. (2007). Early child development: a powerful equalizer. Final report of the Early Child Development Knowledge Network of the Commission on Social Determinants of Health. Geneva, World Health Organization.  Forer B (2009). Everything You Wanted To Know About Component Scores and How They Are Used to Create an Overall HELP SES Index. Draft Document.  Forer B, Guhn M, & Zumbo B (2011). Using a Generalizability Theory Framework to Investigate the Reliability of Aggregated Early Development Instrument Change Scores. Presentation at Society for Research in Child Development Biennial Meeting, April 2, 2011. Guhn M, Goelman H (2010). Bioecological Theory, Early Child Development and the Validation of the Population-Level Early Development Instrument. Social Indices Research 103:193-217.  Halfon N, Larson K, Russ S (2010). Why Social Determinants? Healthcare Quarterly Vol 14 Oct. 71   Hart B, Risley RT (1995). Meaningful differences in the everyday experience of young American children. Baltimore: Paul H. Brookes. Heckman JJ (2010). Building Bridges Between Structural and Program Evaluation Approaches to Evaluating Policy. NBER Working Paper No. 16110. June 2010. JEL No. C21, D6, H43. Heckman JJ (2012a). Boston Review Sept / Oct Lead Essay. http://www.bostonreview.net/BR37.5/ndf_james_heckman_social_mobility.php. Accessed Mar 2013. Heckman JJ, Kautz TD (2012b). Hard Evidence on Soft Skills. NBER Working Paper No. 18121 June 2012. JEL No. D01, I20.  Heckman JJ, Pinto R, Savelyev P (2012c). Understanding the mechanisms through which an influential early childhood program boosted adult outcomes. American Economic Review. Forthcoming.  Hertzman C, Bertrand J (2007). Children in poverty and the use of early development instrument mapping to improve their worlds. Paediatr Child Health; 12(8).  Hertzman C (2010a). How neighborhoods matter for rural and urban children?s language and cognitive development at kindergarten and grade 4. Journal of Community Psychology, Vol. 38, No.3.  Hertzman C (2010b). Social Geography of Developmental Health in the Early Years. Healthcare Quarterly Vol 14 Oct.  Hertzman C (2010c). Tackling inequality: get them while they?re young. BMJ February Vol 340.  Hertzman C (2011). Speech: ?Are we the people we need to be?? RSC Governor General Lecture Series. Mar 29, 2011. U of S, Saskatoon.  Hymel S, LeMare L, McKee W (2011). The Early Development Instrument: An Examination of Convergent and Discriminant Validity. Social Indices Research 103:267-282.  Hystad P, Carpiano RM (2012). Sense of community-belonging and health-behaviour change in Canada. J Epidemiol Community Health; 66.  Imbens GW, Wooldridge JM (2009). Recent Developments in the Econometrics of Program Evaluation Journal of Economic Literature, 47:1, 5?86.  Irwin L, Siddiqi A, Hertzman C. Early child development: a powerful equalizer. Final Report for the WHO Commission on the Social Determinants of Health. Geneva, World Health Organization.  Janus M, Offord D (2007). Development and psychometric properties of the early development instrument (EDI): A measure of children?s school readiness. Canadian Journal of Behavioural Science, 39, 1?22.  72  Janus M, Brinkman S, Duku E (2010). Validity and Psychometric Properties of the Early Development Instrument in Canada, Australia, United States, and Jamaica. Social Indices Research 103:283-297.  Keating DP, Hertzman C (1999). Developmental Health and the Wealth of Nations. New York: Guilford Press.  Kershaw P, Irwin L, Trafford K, & Hertzman C (2005). The British Columbia Atlas of child development (1st ed., Vol. 40). Vancouver, BC: Human Early Learning Partnership, Western Geographical Press.  Kohen DE, Hertzman C, & Brooks-Gunn J. (1998). Neighborhood influences on children?s school readiness. Hull, QB: Human Resources Development Canada.  Lloyd JEV, Hertzman C (2006). Bringing it Together: Proceedings from the Early Childhood Learning Knowledge Centre?s Monitoring Committee Workshop. Lloyd JEV, Hertzman C (2009). From Kindergarten readiness to fourth-grade assessment: Longitudinal analysis with linked population data. Social Science and Medicine, 68, 111?123.  Lloyd JEV, Hertzman C (2010). How neighborhoods matter for rural and urban children?s language and cognitive development at kindergarten and grade 4. Journal of Community Psychology, Vol 38, No 3.  Mort JN, Hughes N, Dockendorf M, Quigg P, Hertzman C (2008). The Early Development Instrument (EDI) in British Columbia: Documenting impact and action in schools, community, and early child development. Vancouver, BC: Human Early Learning Partnership.  Maggi S, Irwin LJ, Siddiqi A, Hertzman C (2010). The social determinants of early child development: An overview. Journal of Paediatrics and Child Health 46.  McGrail K (2007). Medicare Financing and Redistribution in British Columbia, 1992 and 2002. Healthcare Policy Vol. 2 No 4.  Pelletier J, Corter C (2005). Design, implementation, and outcomes of a school readiness program for diverse families. The School Community Journal, 15, 89?116.  Pivik J (2012). Environmental Scan of School Readiness for Health: Definitions, Determinants, Indicators and Interventions. Vancouver, BC, Canada: Human Early Learning Partnership with the National Collaborating Centre for Determinants of Health.  Santos R, Brownell M, Ekuma O, Mayer T, Soodeen R. The Early Development Instrument (EDI) in Manitoba: Linking Socioeconomic Adversity and Biological Vulnerability at Birth to Children?s Outcomes at Age 5. Winnipeg, MB: Manitoba Centre for Health Policy, May 2012.  Schroeder J, Harvey  J, Razaz-Rahmati N, Corless G, Negreiros J, Ford L, Kershaw P, Anderson L, Wiens M, Vaghri Z, Stefanowicz A, Irwin LG, Hertzman C (2009). Creating communities for young children. A toolkit for change. Vancouver, BC: Human Early Learning Partnership.  73  Siddiqi A, Hertzman C (2007a). Towards an epidemiological understanding of the effects of long-term institutional changes on population health: A case study of Canada versus the USA. Social Science & Medicine 64.  Siddiqi A, Irwin LG, Hertzman C (2007b). Total Environment Assessment Model for Early Child Development.  Evidence Report for the World Health Organization?s Commission on the Social Determinants of Health.  Siddiqi A, Kawachi I, Berkman L et al. (2012). Education determines a nation?s health, but what determines educational outcomes? A cross-national comparative analysis. Journal of Public Health Policy Vol. 33, 1.  Successby6 (2012). Success By 6 & Children First Community Capacity: Building Initiatives in BC, Shared Measurement Report May 2012. http://www.successby6bc.ca/sites/default/files/Draft%20CC%20report%20FINAL%20JUNE%2025.pdf. Accessed Aug 2012. Successby6 (2011). ECD Evaluation Project User Guide Revised January 2011. http://www.successby6bc.ca/sites/default/files/ECD%20Evaluation%20Guide%202011%20%20Jan%2011.pdf. Accessed Aug 2012. United Nations (2005). Convention on the Rights of the Child. General Comment 7. www2.ohchr.org/english/bodies/crc/.../GeneralComment7Rev1.pdf. Accessed Aug 2012.  74  Appendix:  Table 18: Changer Communities Identified with Critical Differences Neighborhood Code  Name  Changetype School District 37.44 Ladner Centre                       1 Delta                       71.06 Comox                                1 Comox Valley                23.71 North Glenmore/McKinley           1 Central Okanagan            84.01 Island West                         3 Vancouver Island West       54.01 Houston                             3 Bulkley Valley              36.28 Kennedy Trail                       3 Surrey                      36.10 Fraser Heights                      3 Surrey                      71.02 South Valley/Islands                3 Comox Valley                47.11 Powell River Centre                 3 Powell River                82.07 Terrace - Southside                 3 Coast Mountains             23.85 Black Mountain                      3 Central Okanagan            92.01 Nisga'a                              3 Nisga'a                     68.05 Newcastle - Townsite                3 Nanaimo - Ladysmith         57.15 Upper Fraser                        3 Prince George               36.26 Fleetwood West                      3 Surrey                      68.06 Northfield                          3 Nanaimo - Ladysmith         71.05 Comox West                          3 Comox Valley                68.14 Nanaimo - Downtown                  3 Nanaimo - Ladysmith         61.09 Saxe Point                          3 Greater Victoria            83.04 Salmon Arm East                     3 North Okanagan - Shuswap    68.11 Ladysmith                           3 Nanaimo - Ladysmith         68.12 Nan-Westwood                        3 Nanaimo - Ladysmith         37.49 Nordel                               3 Delta                       82.04 Terrace - North                     3 Coast Mountains             39.09 Kensington - Cedar Cottage         3 Vancouver                   53.05 Osoyoos                             3 Okanagan - Similkameen      36.23 Strawberry Hill                     3 Surrey                      53.01 Oliver                               3 Okanagan - Similkameen      79.53 Cowichan Bay/Glenora                3 Cowichan Valley             36.05 Ocean Park                          3 Surrey                      46.03 Pender Harbour                      3 Sunshine Coast              37.48 Burns View                          3 Delta                       39.12 Shaughnessy                         3 Vancouver                   71.04 South Courtenay                     3 Comox Valley                5.01 Cranbrook - East                    3 Southeast Kootenay          83.03 Salmon Arm West                     3 North Okanagan - Shuswap    57.01 College Heights                     3 Prince George               70.11 Tofino - Ucluelet                    3 Alberni                     41.03 Burnaby Mountain                    3 Burnaby                     36.29 Kirkbride                            3 Surrey                      62.01 Metchosin                           3 Sooke                       70.10 Regional Districts                  3 Alberni                     75  40.02 Connaught Heights                   3 New Westminster             40.06 Queensborough                       3 New Westminster             62.09 Highlands                           3 Sooke                       22.04 Vernon - North                      3 Vernon                      50.01 Haida Gwaii                         3 Haida Gwaii/Queen Charlotte 23.66 West Kelowna Estates                3 Central Okanagan            63.11 Central Saanich - Keating           3 Saanich                     36.37 Gateway                             7 Surrey                      35.01 Milner - Hopington                  7 Langley                     46.01 Gibsons/Langdale                    7 Sunshine Coast              41.13 Deer Lake                           7 Burnaby                     42.04 Haney                                7 Maple Ridge - Pitt Meadows  41.14 Metrotown                           7 Burnaby                     37.47 Jarvis                               7 Delta                       40.01 Uptown                              7 New Westminster             85.02 Port Hardy                          7 Vancouver Island North      75.03 Mission - West Heights              7 Mission                     71.09 Black Creek/ Merville                7 Comox Valley                48.04 Pemberton                           7 Howe Sound                  60.03 Fort St John - Southeast            7 Peace River North           8.05 Salmo                                7 Kootenay Lake               74.12 Gold Trail East                     7 Gold Trail                  35.02 Aldergrove                          9 Langley                     39.02 West Point Grey                     9 Vancouver                   36.44 Fleetwood North                     9 Surrey  Table 19: Communities Removed from Analysis Neighborhood Code  Name   School District W2 Missing 23.83 KLO/Casorso                         Central Okanagan            37.52 Delta Rural                         Delta                       44.15 Dollarton                            North Vancouver             70.09 Bamfield - Alberni Canal            Alberni                     82.10 Snow Country                        Coast Mountains             41.09 Government Street                   Burnaby                     23.81 South/East Kelowna                  Central Okanagan            36.04 Crescent Beach                      Surrey                      61.19 Mount Tolmie                        Greater Victoria            73.04 Logan Lake/Savona                   Kamloops/Thompson           33.14 Sardis East                          Chilliwack                  33.10 Greendale/Chilliwack Mountain Chilliwack                  28.03 Quesnel North                       Quesnel                     23.61 Peachland                           Central Okanagan            57.13 Blackburn                           Prince George               27.01 Chilcotin                            Cariboo - Chilcotin         54.04 Smithers - Walnut Park              Bulkley Valley              76  58.02 Princeton                            Nicola - Similkameen        74.12 Gold Trail East                     Gold Trail                  39.23 West End                            Vancouver                     W3 Missing  61.17 Strawberry Vale                     Greater Victoria            47.11 Powell River Centre                 Powell River                54.01 Houston                              Bulkley Valley              28.01 Nazko/Blackwater                    Quesnel                     70.11 Tofino - Ucuelet                    Alberni                     53.02 Tuc-el-Nuit                         Okanagan - Similkameen      67.06 Bench - Naramata                    Okanagan - Skaha            71.02 South Valley/Islands                Comox Valley                75.05 Upper West Heights                  Mission                     64.02 Outer Gulf Islands                  Gulf Islands                46.03 Pender Harbour                      Sunshine Coast              71.05 Comox West                          Comox Valley                92.01 Nisga'a                              Nisga'a                     84.01 Island West                         Vancouver Island West       53.03 Okanagan Falls                      Okanagan - Similkameen      57.15 Upper Fraser                        Prince George               53.01 Oliver                               Okanagan - Similkameen      82.07 Terrace - Southside                 Coast Mountains             87.01 Stikine                              Stikine                     68.15 Gabriola                             Nanaimo - Ladysmith         23.85 Black Mountain                      Central Okanagan            49.01 Central Coast                       Central Coast               52.05 North Coastal Communities         Prince Rupert  W4 Missing  34.04 Sumas Prairie                       Abbotsford                  41.09 Government Street                   Burnaby                     34.07 Matsqui - Mt. Lehman                Abbotsford                  34.05 Kilgaard                             Abbotsford                  34.06 Sandy Hill                           Abbotsford                  34.09 McMillan                            Abbotsford                  34.08 Clayburn                             Abbotsford                  34.02 Airport - Aberdeen                  Abbotsford                  8.05 Salmo                                Kootenay Lake               34.13 Townline East                       Abbotsford                  34.18 Townline West                       Abbotsford                  34.14 West Clearbrook                     Abbotsford                  34.03 South Poplar                        Abbotsford                  34.11 Clearbrook                          Abbotsford                  60.03 Fort St John - Southeast            Peace River North           34.12 North Clearbrook                    Abbotsford                  77  34.17 Mill Lake                            Abbotsford                  34.15 South Clearbrook                    Abbotsford                  34.10 Abbotsford                          Abbotsford                  34.16 Babich                               Abbotsford                   Figure 9: Success by 6 Program Logic Model (Successby6, 2011)     78  Table 20: Vulnerability Profile, Revelstoke   EDI Subscale  Vulnerability  by Wave (total n) Critical Difference for each subscale (n=150)  Average Vuln  Total Vuln shift Average Absolute Vuln shift per wave W2 (150) W3 (150) W4 (138)     1 or more scale 12 7 10 6 9.67 -2 4 Physical 4 3 2 6 3 -2 1 Social 5 3 4 4 4 -1 1 Emotional 4 3 3 5 3.67 -1 0.5 Language 4 3 7 4 4.67 +3 2.5 Communication 4 3 2 5 3 -2 1 Legend: _____ = positive change, _____ = negative change  Table 21: EDI Construct Validity with census variables for Revelstoke  Scale Expected Community Properties Physical Low income rate?, % of females in manufacturing positions?, % of males in management positions?, % of males performing no unpaid childcare ?,  Social* Median family income?, % of adults performing no unpaid housework?, % lone-parent families?, % males that drive to work ? Emotional* Male employment rate with children under 6 ?, % of males in management positions ?, % of males performing no unpaid childcare ?, % lone-parent families? Language Median family income ?, male employment rate with child ?, unemployment rate with children under 6 ? , % of lone-parent families ?, % of non-Christians ? Communication* Home ownership rate?, gender income disparity?, % of males in management positions, % using a foreign home language? Any Scale Low income rate ?, employment rate with children under 6?, % of males in management positions?, % of males performing no unpaid child care ?, % of first generation Canadians ?, % of non-migrant movers ? Legend: _____=  expected, supports vulnerability increase observed white / no color = at about the provincial average _____= unexpected, may not support vulnerability increase observed *-critical difference on scale over time  79   There is also extensive qualitative data from Mort which pertains directly to the community, as Revelstoke represents the school district level as well as the community. See Figure 10 for Revelstoke?s intersectoral connectedness chart (Mort, 2008). Structures - Economic Properties: Policies:  Revelstoke has a very collective, inclusive attitude about intersectoral collaboration and it comes through in their policies. There is a commitment to bring the bottom up. This could be a contributing factor to their track record of stable low vulnerability. They also opened a dedicated StrongStart Center with acceptance from the community (Mort 2008).  The superintendent notes in Mort?s interviews that the school district allied itself with SuccessBy6 in 2004 and called itself the Early Childhood Development Committee:  ?We worked with the School District and the Intersectoral Coalition to relocate service providers to a closed Elementary School. The co-location of services was pivotal in the beginning of a hub in Revelstoke. In the same building, a family could see the Speech Language Pathologist or the Literacy Coordinator and then walk across the hall to access other services? (Mort 2008).   There were also concerted attempts at reducing barriers to children and families. The coalition contributed $3,000.00 to an ?Access Fund? which gave snack money to a disadvantaged family, and also worked with the city to issue free pool passes to any family who is below $30,000 yearly income. ?We collectively eliminated the bureaucracy,? the superintendent notes. ?These kids are now swimming, socializing with other kids, involved in programs such as Water Books. It is a community philosophy for all service providers to offer in-kind services to vulnerable families and their children.? Funding sources include MCFD, Ministry of Education, Ministry of Health, Success by Six, Revelstoke Credit Union, Early Childhood Development Committee, the Columbia Basin Alliance for Literacy, and the City of Revelstoke. The Ministry of Education?s Early Childhood Grant of $106,000 has also proved invaluable in the implementation of new programs. The Revelstoke Child Care Resource and 80  Referral organization is funded by MCFD. Various community partnerships provide funds for early-learning initiatives, and numerous in-kind donations assist in maintaining program sustainability. The Columbia Basin Trust provides funds for early childhood initiatives. School District 19, MCFD, Ministry of Health, and MCFD/Child Protection were the funding partners behind the establishment of Revelstoke?s StrongStart Centre in 2006. Infrastructure:  Intersectoral coalition partners include: School District 19, MCFD, the Ministry of Health, MCFD/Child Protection, Success by Six, Revelstoke Credit Union, Children First, the Revelstoke Early Childhood Development Committee, Revelstoke Parks and Recreation, Okanagan College, the Child Care Society, the RCMP, Okanagan Regional Library, the Columbia Basin Alliance for Literacy, Literacy Now, Revelstoke Literacy Action Committee, Infant Development Program, Community Connections Society and the City of Revelstoke (Mort 2008). SES:  From the Census, median income is above the provincial average, low income % is below the provincial average, and home ownership % is above average. Structures - Social Properties:  Social Cohesion  The community leaders specifically noted the existence and awareness of social distance and social hierarchy in their community. Treating social distance and hierarchy as barriers, the community mobilized by picking up isolated kids with cars, and getting isolated families into programs with no cost registration. Transiency  The school board noted a 37% drop in registration in students over 12 years, which presented sustainability issues. Since school districts are funded per child, this makes sense. 81  Urban / Rural The community is urban, representing the town of Revelstoke. Processes - Programs:  Mort?s collection of program information comes from a kindergarten teacher. The extensive programs and sites include Leap Land Indoor Playground, Family Night Out, Stepping Stones Child Care, Jumping Jacks Preschool, Toddler Time and Cool Kids Preschool Programs, FARWELL early learning hub,  Early Learning Resource Library at the Revelstoke Child Care Society, PALS, POPS, Ready, Set, Learn, StrongStart Centre, Baby Talk, Baby Steps, Toddler Talk, Expectations Pre-Natal program, PACT (Parents and Community Together), Revelstoke Family Literacy, Regional Library Programs: Tiny Tickle, Screen Smart, Story time, Tales for Tots, and Family Literacy Day.  There is also an understanding of the positive impact programs can have. The superintendent explains ?we know that children who attend high-quality early learning programs are more likely to be successful in the primary grades and have a higher completion rate in Secondary Schools? (Mort 2008). 82  Figure 10: (Mort 2008)  

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.24.1-0074214/manifest

Comment

Related Items