Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cardiovascular response to agreement and disagreement: towards explaining the beneficial effect of social… Lenz, Joseph William 1995

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

Item Metadata

Download

Media
831-ubc_1995-059987.pdf [ 2.72MB ]
Metadata
JSON: 831-1.0088355.json
JSON-LD: 831-1.0088355-ld.json
RDF/XML (Pretty): 831-1.0088355-rdf.xml
RDF/JSON: 831-1.0088355-rdf.json
Turtle: 831-1.0088355-turtle.txt
N-Triples: 831-1.0088355-rdf-ntriples.txt
Original Record: 831-1.0088355-source.json
Full Text
831-1.0088355-fulltext.txt
Citation
831-1.0088355.ris

Full Text

CARDIOVASCULAR RESPONSE TO AGREEMENT AN]) DISAGREEMENT:  TOWARDS EXPLAINING THE BENEFICIAL EFFECT OF SOCIAL SUPPORT BY JOSEPH WILLIAM LENZ  B.A., THE UNWERSITY OF ILLINOIS AT CHICAGO, 1970 M. S., WESTERN WASHINGTON UMVERSITY, 1990  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY IN  THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF PSYCHOLOGY  WE ACCEPT THIS THESIS AS CONFORMING  TO THE REQUIRJED STANDARD  Signature(s) removed to protect privacy  THz UNIVERSITY OF BRrnsH COLUMBIA AUGUST, 1995 ©  JoSEPH WILLIAM LENz,  1995  Signature(s) removed to protect privacy  for an advanced In presenting this thesis in partial fulfilment of the requirements Library shall make it degree at the University of British Columbia, I agree that the sion for extensive freely available for reference and study. I further agree that permis the head of my copying of this thesis for scholarly purposes may be granted by that copying or department or by his or her representatives. It is understood t my written publication of this thesis for financial gain shall not be allowed withou permission.  Signature(s) removed to protect privacy  Signature(s) removed to protect privacy  Department of  0’ O\  The University of British Columbia Vancouver, Canada  Date  DE-6 (2188)  /  Ao%  Idjt.S_  Lenz  Response to Audience Agreement  ABSTRACT  Social support has been associated with reduced mortality and morbidity from a number of causes. To assess possible mechanisms of action relating to cardiovascular (CV) responsiveness, 90 male and female university students delivered a five-minute speech on a controversial topic to a same-sex laboratory confederate. Subjects were randomly assigned to one of three conditions in which the confederate either (a) agreed with the subject, (b) remained impassive (neutral), or (c) disagreed with the subject. Blood pressure (SBP and DBP) and heart rate (HR.) were monitored throughout the experiment. Self-report measures of state self-esteem and affective state were taken pre- and post-task, and reactions to the task were assessed with post-task self-report measures. Subjects reported strong differences in supportiveness of the confederate in the three conditions. Self-report data indicated increase in arousal during the speech (a finding synchronous with CV data), and they reported the Disagree condition to be less pleasant than the Agree condition. CV data were analyzed as a 2 x 3 (sex by experimental condition) repeated measures ANOVA assessing changes from baseline to speech task. Sex differences on CV measures matched patterns generally reported: Men had higher SBP and lower HR than women. All CV measures increased significantly and substantially during the speech task. HR was higher in the Disagree and Neutral conditions than in the Agree condition. SBP and DBP did not differ by condition. There were no sex by condition interactions; however, there was a trend towards men’s HR increasing more in the neutral condition and women’ more in the disagree condition.  —  11  —  Lenz  Response to Audience Agreement  These data partially support earlier findings in similar experiments while suggesting that subtleties of context, task selection, and content of supportive interaction may have significant impact on the degree to which social support attenuates CV response to social stressors. Unanswered questions for future research are delineated, and implications for designing and implementing interventions that enhance social support are discussed.  —  111  —  Lenz  Response to Audience Agreement TABLE OF CONTENTS  ABSTRACT TABLE OF CONTENTS  iv  LIST OF TABLES  vi  LIST OF FIGURES  LIST OF APPENDICES  viii  ACKNOWLEDGMENT  x  Dedication  x  LITERATURE REVIEW  1  Introduction  1  Levels of Analysis  3  Measurement  4  The Epidemiology of Social Support and Mortality  6  Interpreting Epidemiological Findings  10  Prospective Studies of Social Support and Cardiovascular Measures  11  The Experimental Investigation of Social Support  13  Laboratory Studies of Cardiovascular Response to Social Support  15  Goals and Import of the Present Study  25 31  METHOD Overview  31  Subjects  31  —  iv  —  Lenz  Response to Audience Agreement  Self-report Measures  33  Procedure  40  Analytic Strategy  51  RESULTS  54  Manipulation Check  54  Other Self-report of Response to the Speaking Task  56  Cardiovascular Reactivity to the Task  58  Pre- and Post-task Subjective Measures  67  Correlates of Self—reported Social Support  73  DISCUSSION  76  Summary of Findings  76  Internal Validity  76  Evaluation of Specific Hypotheses  78  Cardiovascular Response to Support  86  Explanations of Differences Between Findings  88  Directions for Future Research  94  Conclusions  95  I1.EFER.ENCES  98  APPENDIX A APPENDIX B  —  —  QUESTIONNAIRES AND SCRIPTS  109  STATISTICAL TABLES  132  —v—  Lenz  Response to Audience Agreement  LIST OF TABLES  TABLE  1. MEANS AND (STANDARD DEVIATIONS) OF RATINGS OF CONFEDERATE SUPPORTWENESS BY EXPERIMENTAL CONDITION  TABLE  2. CORRELATIONS AMONG CARDIovAscuLAR MEASURES  55 59  TABLE 3. MEANS AND (STANDARD DEVIATIONS) OF SYSTOLIC BLOOD PRESSURE BY EXPERIMENTAL CONDITION AND SEX  62  TABLE 4. MEANS AND (STANDARD DEVIATIONS) OF DIAsToLIc BLOOD PRESSURE BY EXPERIMENTAL CoNDITIoN AND SEX TABLE  63  5. MEANS AND (STANDARD DEVIATIONS) OF HEART RATE BY EXPERIMENTAL CONDITION AND SEX  65  TABLE 6. CORRELATIONS AMONG SELF-REPORT MEASURES AT PRE-TASK  68  TABLE 7. CORRELATIONS AMONG SELF-REPORT MEASURES AT POST-TASK  68  TABLE 8. MEANS AND (STANDARD DEVIATIoNs) OF PLEASANTNESS RATING ON THE AFFECT GRID BY EXPE1UMENTAL CONDITION  70  TABLE 9. MEANs AND (STANDARD DEVIATIONS) OF PERFORMANCE RATING ON THE  SSES BY EXPERIMENTAL CONDITION  73  TABLE 10. CORRELATIONS BETWEEN ISEL Ar’ CHANGE IN CARDIOVASCULAR AND SELF-REPORT MEASURES BY EXPERIMENTAL CONDITION  75  TABLE 11. ESTIMATES OF TASK- AND SUPPORT-RELATED EFFECT SIZES FOR CARDIOVASCULAR MEASURES IN STUDIES USING SOCIAL INTERACTION 87  TASKS  —  vi  —  Lenz  Response to Audience Agreement  LIST OF FIGTJRES  FIGuRE 1. Systolic Blood Pressure at all Measurement Points by Sex  60  FIGuRE 2. Diastolic Blood Pressure at Baseline and Speech by Sex  64  FIGuRE 3. Heart Rate at Baseline and Speech by Experimental Condition  66  FIGuRE 4. Change in Heart Rate by Sex and Experimental Condition  81  —  vii—  Lenz  Response to Audience Agreement  LIST OF APPENDICES APPENIMX A  —  QUESTIONNAIRES AND SCRIPTS  109  1. Recruitment Poster  109  2. Telephone Screening Guide  110  3 Consent Form  112  4. Collection Form for Demographic Data  114  5. State Self-Esteem Scale (Current Thoughts)  116  6. Interpersonal Support Evaluation List (ISEL)  117  7. Affect Grid  120  8. Important Topics Questionnaire  121  9. Reactions to Speaking Questionnaire  122  10. Speech Helper Form  125  11. Debriefing Form (Given to all participants)  127  12. Debriefing Script  129  APPENDIX B  —  STATISTICAL TABLES  132  1. Oneway ANOVA Table for Rating of Confederat&s Supportiveness by Experimental Condition  132  2. Item Descriptions and Correlations among Items in the Distress Scale  133  3. Tables for Multivariate and Univariate Oneway ANOVAs of Distress and Performance Ratings  134  4. ANOVA Tables for Systolic Blood Pressure  135  5. ANOVA Tables for Diastolic Blood Pressure  136  6. ANOVA Tables fo Heart Rate  137  7. ANOVA Tables for the Arousal Subscale of the Affect Grid  138  —  viii—  Lenz  Response to Audience Agreement  8. ANOVA Tables for the Pleasantness Subscale of the Affect Grid  139  9. ANOVA Tables for the Appearance Esteem Subscale of the SSES  140  10. ANOVA Tables for the Social Esteem Subscale of the SSES  141  11. ANOVA Tables for the Performance Esteem Subscale of the SSES  142  12. Means and (Standard Deviations) of Change in Cardiovascular and Self-report Measures by Experimental Condit4on  —  ix  —  143  Lenz  Response to Audience Agreement  ACKNOWLEDGMENT  I want to express my deep appreciation to the team at the Linden laboratory for what must have seemed like never-ending help with this research project. First, I want express my deepest appreciation to Jeff Maurice who did everything from scheduling to acting as confederate, to running subjects, and to Carmen Stossel who always made herself available to solve scheduling problems—and any other problem that happened to come up. Thank you to those who acted as confederates: Susan Bauld, Tammy Brumwell, Mark Hignell, Sean Richardson, and Liz Rocha. In their performances as confederates, everyone had td learn how others perceived their responses, and the learning process challenged everyone. Being non-responsive and disagreeing with subjects did not come easy for the confederates at first, and I can only hope that breaking the habit of being habitually “nice” has broadened everyone’s interpersonal repertoires. Finally, I want to thank Professor Linden himself for the inspiration, the resources, the guidance, and the prodding he provided. Most importantly, I want to thank him for the support he gave me during this project—support that was generally (but not invariably) expressed in terms of agreement.  Dedication: This project is dedicated to the memory of myfather, Russell Lenz. It was his dream from my birth that I would get a university degree, but he never expected—or probably even wished-that it would go thisfar. Still, without the drive and confidence he worked to instill in me, I may never have had the faith to follow my curiosity into research. —x—  Lenz  Response to Audience Agreement  LITERATURE REVIEW  Introduction During the past two decades, there has been a proliferation of research on the relationships between social support and health. Although researchers in health psychology have demonstrated much interest in the links between health and social support, research and theoretical work have been by no means limited to health psychology or even to psychology. The breadth of interest in social support is illustrated by the variety of disciplines that have published extensively on the topic. Major contributions to knowledge and application of social support principles have occurred in journals primarily concerned with epidemiology, cardiology, gerontology, oncology, social work, nursing, rehabilitation, community medicine, and community psychology. This broad and energetic interest has been spurred largely by epidemiological findings that social support is robustly associated with reduced all-cause mortality, is consistently associated with reduced morbidity from a number of specific causes—including coronary disease and hypertension—and with reduced risk of psychopathology. To date, little is known about the mechanisms of action, but the associations between social support and health are stable and the direction of causation is reasonably well established and generally accepted: It appears that strong social networks actually protect health and are not simply the result of good health (Berkman & Syme, 1979; Blazer, 1982; Cohen, S. & Wills, 1985; Kaplan et al., 1988; Welin et al., 1985). In spite of a lack of understanding of  mechanisms, interventions purporting to increase well being by increasing social support  —1—  Lenz  Response to Audience Agreement  have been proposed for a variety of health-related problems. Thus, application currently runs ahead of understanding of underlying processes and, therefore, the risk is high that inefficient or ineffective interventions may be implemented. Given the variety of fields researching phenomena related to health and social support, it is hardly surprising that conceptual and methodological inconsistencies complicate our ability to draw reliable conclusions. Social support itself has been defined in various—and at times conflicting—ways. Berkman and Syme (1979) defined the concept globally, constructing a composite index of support from self-report of a wide variety of social interactions and memberships. Numerous epidemiological studies have followed this lead with variations in the types of social connections sampled for inclusiOn in their indices. Actual statements of definitions of social support are rare in the literature, probably due to both the vagueness of theoretical underpinnings and the generally accepted every-day meaning of the term “social support.” Cassel’s (1976) original formulation defined social support as including elements of social organization, acculturation, and availability of psychosocial resources. Others have simplified and thereby broadened the definition to include social ties and relationships (e.g., Berkman & Syme, 1979; House, Robbins, & Metzner, 1982). More recently, the tendency has been away from global, general definitions and towards situation-specific definitions. Coyne and Bolger (1990) recommended abandoning the term and the concept of general social support in favor of more operationally definable components. They prefer to designate and to research specific behavioral and interactional patterns and to address functional aspects of the relationship between health and social  —2—  Lenz  Response to AudEence Agreement  support rather than to describe structurally supportive networks. This recommendation emphasies the utility of specifiing types of support and components of supportiveness that can be targeted for research separately or in combination. The body of research most closely related to the present study, generally proceeds by examining the impact of specific supportive behaviors on their recipients. In keeping with Coyne and Bolger’s recommendation, reference should be made to actual types of behaviors and exchanges under investigation rather than to the general concept of “social support”. This convention is followed when referring to operationalized components of the present study. However—because the impetus for investigation of these component elements has arisen from theory and data with a wider conceptual base—the term “social support” will be retained when referring to conceptually related issues.  Levels ofAnalysis The differing views on how social support should be defined make it apparent that there are several levels at which social support can be analyzed and researched. First, social support can be discussed at the global or general level. The global construct is linguistically convenient but admits of considerable imprecision in that it refers to a broad set of poorly defined and loosely related constructs. This approach to social support has been apparent in epidemiological studies in which composite measures tap various aspects of the general construct. At a slightly more detailed level of analysis, it is possible to discriminate different types of support. For example, Wills’s (1985) typology includes esteem support, social companionship, instrumental support, informational support, and  —3—  Lenz  Response to Audience Agreement  several others. It would be possible to examine the independent contributions of each of these broad types of support, but to date that type of investigation has not been undertaken in any systematic manner. Instead, it appears that the specification of types of support has had largely a heuristic value. The utility of typing support is in leading to a still more detailed level of analysis at which specific types of supportive behaviors or supportive transactions are examined for the purpose of identifying functional aspects of support.. Research focusing on this last and most detailed level of analysis targets specific supportive interactions as the unit of analysis. At this level, it is important to examine cognitive, emotional, and physiological concomitants of supportive interactions, and the support giver’s behaviors must be specified. This last level of analysis is essential in experimental laboratory procedures, and it is at this last level that the present study operates. The present investigation is also intended to respond to calls by numerous writers for research that is directed towards examination of potential processes or towards identification of mechanisms of action of social support on health (e.g., Cohen,. S. & Wills, 1985; Waliston, Alagna, DeVellis, & DeVellis, 1983).  Measurement There are numerous approaches to measuring social support. Obviously, the definitional issues previously explained have major impact on measurement strategies. Practical problems associated with operationalizing definitions for application lead to additional variety and inconsistency in measurement. Additional issues include the following: (a) emphasis on actual versus perceived support, (b) emphasis on quantity  —4—  Lenz  Response to Audience Agreement  versus quality of support, (c) emphasis on utilization versus availability of support, and (d) emphasis on social connections or “membership’ versus types and quality of social interchanges. Prospective epidemiological studies of social support have generally utilized indices of supportive connections composed of a number of discrete self-report measures of social connections. The earliest study (Berkman & Syme, 1979) used an index composed of measures of marriage, contacts with friends and extended family, church membership, and other formal and informal affiliations. The derived measure weighted the first two of these kinds of contact—the intimate connections—more heavily than the other kinds. Replications of the Berkman and Syme study and later epidemiological work used a variety of different combinations of measures. House et al. (1982) included in their composite index measures of formal organizational involvement outside work and measures of leisure activities involving social contact. Blazer (1982) added measures of perceived adequacy of social support and frequency of social interactions to counts of available supportive connections. In response to positive findings with composite measures, numerous investigators have developed scales for measurement of network size and satisfaction with social support. Psychometric properties of 23 such scales were reviewed by Heitzmann and Kaplan (1988). Scales vary widely in approach, in length, and especially in availability of data on reliability and validity, but several among those reviewed were found to have satisfactory psychometric properties. Among the most psychometrically sound and most commonly used is the Interpersonal Support Evaluation List (ISEL) (Cohen, S.,  —5—  Lenz  Response to Audience Agreement  Mermeistein, Kamarck, & Hoberman, 1985). This instrument is of particular interest for the present study for a number of reasons. It measures perceptions of available support, an aspect of support most appropriate for evaluation through self-report and an aspect of support frequently considered to be most closely associated with health (Cohen. S. & Wills, 1985; Helgeson, 1993). In addition, the ISEL has subscales measuring appraisal of support and esteem support—elements of support that are particularly relevant to aspects of the present investigation.  The Epidemiology of Social Support and Mortality Syme, Hyman, and Enterline (1964) reported evidence that patterns of social interaction may be in part responsible for differences in prevalence of a variety of diseases across a wide variety of cultural settings. These and other early findings prompted a number of theoretical papers suggesting empirical examination of links between health and social activity (Cassel, 1976; Cobb, 1976; Syme, 1974). Syme’s formulation has proved productive over time. He recommended investigation along the lines of what he called a “social epidemiological approach.” In general, Syme recommended epidemiological methodology that enables researchers to uncover existing relationships between two very broadly conceptualized areas: social support (or availability of social resources) and health. It was inherent in this early understanding that the approach was largely exploratory precisely because health and social support were both multifaceted concepts. A number of prospective epidemiological studies have been conducted using mortality as an outcome measure. Berkman and Syme (1979) reported on the first large-scale  —6—  Lenz  Response to Audience Agreement  prospective study indicating an association between social integration and mortality. They followed a 4725-subject probability sample of male and female residents of Alameda County, California, over a 9-year period. Four types of social involvement were tracked (marriage, contact with extended family and friends, church membership, and other group affiliations), and each contributed independently to prediction of mortality. A composite index of the extent of social ties was a significant predictor of mortality after controlling for the following known predictors: physical health status, socioeconomic status, cigarette smoking, alcohol consumption, level of physical activity, obesity, race, life satisfaction, and use of preventive health services. The strength of their findings and the indication of a graded response to social involvement (i.e., increasing social support was associated with decreasing risk of mortality) encouraged numerous replications in North America and later elsewhere. The Alameda County study was replicated and extended with 2754 men and women in Tecumseh County, Michigan, by House et al. (1982). In this study, composite measures of social activity and social relationships somewhat different from those used by Berkman and Syme (1979) were inversely related to mortality over a 10- to 12-year follow-up period independent of a variety of biomedical risk factors (e.g., blood pressure, cholesterol, respiratory function, and electrocardiograms). In Tecumseh County, unlike Alameda County, the effects were a good deal stronger for men than for women. In Durham County, North Carolina, a sample of 331 elderly subjects (age 65 and over) were followed for a 30-month period (Blazer, 1982). Based on a broadly defined measure of social support much like Berkman and Syme’s, associations between mortality and social  —7—  Lenz  Response to Audience Agreement  involvement were evident after controlling for ten variables normally associated with mortality. In Evans County, Georgia, 2059 older adults (mean age  =  54) were followed  over a 11- to 13-year period (Schoenbach, Kaplan, Fredman, & Kleinbaum, 1986). Using a social activity index similar to Berkman and Syme (1979), Schoenbach et al. found social activity predictive of mortality in a proportional hazards model independent of age and major biomedical and self-reported risk factors. Although the trend was the same across all races and genders and ages, the effect was most evident in older white men, and there was no indication of a gradient of risk. Replications elsewhere have contributed to a more complex view of the linkage between social support and health. A 9-year study of men in GOttenburg, Sweden, indicated a strongly graded effect of social contact in two different cohorts (Welin et al., 1985). These results linked broadly defined social support with all-cause mortality and, therefore, replicated the body of North American findings for men. In Hawaii, Reed, McGee, and Yano (1984) followed 4251 men of Japanese ancestry for seven years. They reported no association between social support and stroke, cancer, all diseases combined, or mortality (Reed et al., 1984). In spite of these negative findings that contrasted strongly with numerous other studies, strong associations were found between social support and coronary heart disease (Reed, McGee, Yano, & Feinleib, 1983). One possible explanation of differences between the Reed, McGee, and Yano (1984) findings and other studies is that in Hawaii a highly homogenous group was sampled—a group that may in fact differ from earlier populations in both social behavior and susceptibility to disease (Orth-Gomer & Johnson, 1987; Reed et al., 1984). In an attempt to reduce error due to sampling  —8—  Lenz  Response to Audience Agreement  homogenous groups, Orth-Gomer and Johnson (1987) constructed a huge (17,433) national sample of the Swedish population and followed them for six years, tracking mortality due to cardiovascular disease in addition to all-cause mortality. They controlled for health risks such as age, smoking, exercise, and chronic illness and found that below a critical level, reduced social support was associated with increased risk of mortality for both men and women. This increase in risk was similar for all-causes pooled and for death due to cardiovascular disease. Another large study in North Karelia, Finland, also examined mortality due to cardiovascular disease separately from all-cause mortality (Kaplan et a!,, 1988). In this study, as in Tecumseh, Michigan (House et al., 1982), and in Evans County, Georgia (Schoenbach et al., 1986), it was men who showed the strongest influence of social contact on mortality. For men only, Kaplan, et a!. found strongly graded associations between social activity and death due to all causes, cardiovascular disease, and ischemic heart disease. Unlike other epidemiological studies, the North Karelia study included numerous measures of health status at the outset of the study. Subjects were questioned concerning history of myocardial infarction, angina, emphysema, chronic bronchitis, asthma, diabetes, cardiac insufficiency, and hypertension. Of these variables, only hypertension proved to be a significant moderator of the effects of social support on mortality. In men with higher blood pressure at the outset. of the study, the association between social support and mortality was stronger than in normotensives.  9  Lenz  Response to Audience Agreement  Interpreting Epidemiological Findings Direction of causality. It is always problematic to infer causality from findings in epidemiological and correlational investigations. In social support theory, in fact, there are reasons to suspect bi-directional causality. That is, it appears likely that people who are not well or who perceive themselves as unwell may fail to make social contacts. In this case, low social support may be a result of and not a cause of poor health. This commonsense alternative explanation has led a number of investigators to address the issue in their analyses. Several different strategies have been employed to tease out causal direction. One of the most direct was the utilization of cross-lagged panel and structural equation modeling in a longitudinal, 12-month follow-up of post-myocardial infarction patients (Fontana, Kerns, Rosenberg, & Colonese, 1989). These data supported the general hypotheses that social support ameliorated cardiac symptomatology during the first half year following myocardial infarction while alternative causal hypotheses were supported only at the level of chance. Other investigators have generally addressed issues of causality in a less direct manner, but findings have consistently indicated that the major causal direction is for social support to protect health (Blazer, 1982; Kaplan et al., 1988; Welin et a!., 1985). General findings. The overall picture from the prospective studies indicates a consequential relationship between social support and physical health, but there are inconsistencies concerning strength of effects and sex differences. Differences in findings may be accounted for in part by differences between the populations studied. In addition, measurement of social support has been inconsistent across studies, and these variations  —10—  Lenz  Response to Audience Agreement  may partially explain differences in magnitude of risk across studies (Orth-Gomer & Johnson, 1987). Reviews of the epidemiological research on social support and health by Cohen and Wills (1985) and later by House, Umberson, and Landis (1988) led to the following conclusions: (a) The findings are more consistent and the effects larger for allcause mortality than for various measures of morbidity; (b) since the outcome measures showing the strongest effects are mortality from all causes, the effects of social support on health are apparently non-specific. Directions  f future research. Since mortality is clearly not without cause, and since  mortality can logically be seen as an indication of advanced morbidity, stronger findings for mortality than for morbidity is an unsatisfactory state of affairs. Such findings appear to indicate incomplete identification and measurement of morbidity, and these failures probably center around the difficulties in precisely identif,ring etiologic pathways for many diseases. Based largely on this line of reasoning, Cohen and Wills (1985) recommended implementation of studies designed to test hypotheses concerning the effects of social support on specific organ systems and the linkages between social support and specific diseases.  Prospective Studies of Social Support and Cardiovascular Measures The trend in epidemiological studies has been away from simple all-cause mortality as an outcome measure and towards a number of specific measures of morbidity. Numerous writers have indicated a need to identi,r the various mid-range health measures—that is, indications of health state somewhere between weilness and mortality—that are responsive  —11—  Lenz  Response to Audience Agreement  to social support and that are likely to be eventual precursors of mortality (Berkman, 1985; Berkman, Vaccarino, & Seeman, 1993; Cohen, S. & Wills, 1985; Germ, Pieper, Levy, & Pickering, 1992; Kaplan et al., 1988; Lepore, Allen, & Evans, 1993). To date, investigations of cardiovascular disease have been productive in filling in this gap in understanding the linkages between health and social support. The data from Finland (Kaplan et al., 1988), Sweden (Orth-Gomer & Johnson, 1987), and Hawaii (Reed et al., 1983) all suggest that cardiovascular diseases may be especially influenced by social integration. The relation of social support to cardiovascular measures other than mortality has also been studied at length. The most programmatic investigations of the effects of social support on cardiovascular health have been conducted in Sweden in response to the strong implication of cardiovascular disease in the Gottenburg epidemiological study (Orth Gomer & Johnson, 1987). One extensive Swedish study of hemodynamics and social support (Knox, Theorell, Svensson, & Wailer, 1985) was conducted as a 10-year followup of hypertensives, normotensives, and hypotensives. Repeated measures of a number of physiological parameters related to cardiovascular function were analyzed for their relation to social support. Path analysis and linear regression techniques revealed that a substantial proportion of the variance in systolic blood pressure was accounted for by plasma adrenaline level which was in turn related to levels of attachment and to the number of social contacts. Diastolic blood pressure was positively correlated with heart rate, and heart rate was in turn negatively correlated with number of social contacts. These data provide strong indication of the role of social support (defined as integration in a  —12—  Lenz  Response to Audience Agreement  social network) in both immediate cardiovascular measures and in the development of signs iefated to the onset of cardiovascular disease. Ambulatory monitoring of blood pressure has also been utilized to examine relationships between social support and health. Unden, Orth-Gomer, and Elofsson (1991) monitored electrocardiograms, blood pressure, and heart rate over a 24-hour period for 148 working men and women. It was found that low reported social support in the work environment was associated with increased heart rate and increased systolic blood pressure over the entire 24-hour monitoring period. These associations were strengthened when statistical controls for other known cardiovascular risk factors were included in the data analysis. A similar ambulatory monitoring study in Canada produced similar—but slightly weaker—results (Linden, Chambers, Maurice, & Lenz, 1993). Ambulatory blood pressure and heart rate monitoring of a sample of 129 university students over an 8-hour period during their school day, showed social support to be an independent predictor of ambulatory blood pressure for women but not for men.  The Experimental Investigation ofSocial Support The weight of evidence in prospective studies leaves little doubt that social support (variously conceptualized) has a positive impact on cardiovascular health. Experimental investigations need not address this issue. Instead, there are two sets of questions that have not been—and probably caimot be—addressed adequately in prospective or cross sectional studies that need to be addressed experimentally. The first has to do with mechanisms of action, and the second refines the concept of social support. First, what are  —  13  —  Lenz  Response to Audience Agreement  the physiological pathways through which social support exerts its effects on cardiovascular health? Secondly, which of the broad assortment of behaviors and perceptions associated with social support or social integration are in fact essential and ’ ingredients? The present study proposes to address the second of these questions t “active directly and the first indirectly through the identification of arousal patterns. Rationales for conducting research on phenomena related to human stress have been proposed and debated repeatedly over the past decade or more (e.g., Cohen, S. & Matthews, 1987; Krantz & Manuck, 1984; Pickering & Germ, 1990). Conflict in this field appears to revolve around the difficulty of identif,ring specific etiologic links between stress reactivity observed in the laboratory and the development of disease (cf. Manuck, 1994; Pickering & Germ, 1990). In spite of the difficulty in establishing specific links between cardiovascular reactivity in the laboratory and disease processes, it has been well established that numerous physical and psychological stressors elicit short-term cardiovascular and neuroendocrine reactions in human subjects during laboratory stress procedures (Schneiderman & McCabe, 1989). Prospective studies have associated high levels of cardiovascular response to stress with a variety of cardiovascular diseases (Manuck, Kasprowicz, & Muldoon, 1990). It has also been shown that individuals at risk for development of hypertension exhibit heightened cardiovascular response to laboratory stressors prior to development of clinically elevated blood pressure (Manuck, Kasprowicz, & Muldoon, 1990). Since hypertension and other forms of cardiovascular disease are associated with heightened cardiovascular responsiveness to a variety of stressors  —14—  Lenz  Response to Audience Agreement  (Fredrikson & Matthews, 1990; Pickering & Germ, 1990), it is possible that social support may protect cardiovascular health by reducing cardiovascular responsiveness to stressors. Numerous laboratory procedures have utilized the high levels of control possible in laboratory research to successftully assess specific deleterious aspects of stress reactivity. Probably the best rationale for this work has been articulated by Taylor (1991). She proposed that positive and negative emotions exert unequal (asymmetrical) effects on stress reactivity and on health in general. Negative emotions require more intense shortterm mobilization of the organism than do positive emotions, and—once mobilized—the organism attempts to find ways to effectively minimize stress-related arousal. Negative social events have, in fact, been found more predictive of health outcome than pOsitive events (Helgeson, 1993). The self-regulatory mechanisms posited by Taylor lead directly to three conclusions relating to laboratory stress procedures. First, the impact of a negative event may extend beyond the duration of the event itself. Secondly, the effect of emotional and stress-related events depends on patterns of response, not on simple arousal levels. And finally, given the arousal levels often observed in laboratory stress procedures, there is little reason to assume that patterns of reactivity and self-regulation would not be observable in the laboratory as well as in a subject’s natural environment.  Laboratory Studies of Cardiovascular Response to Social Support A small body of experimental studies suggests that social support attenuates cardiovascular responsiveness to stress. It is useflul for more detailed examination, to group experimental approaches to the relationship between cardiovascular reactivity and  —  15  —  Lenz  Response to Audience Agreement  social support by the type of laboratory stressor and by the type of social support variable utilized. Most simply, social suppOrt can be either manipulated as a part of the experimental protocol, or subjects can be selected on the basis of level of self-reported social support or social network size. Laboratory stress tasks can also be of two general types: (a) cognitive challenges or (b) tasks involving social interaction. Designating study type, or sorting studies by approach is useful because each type of study addresses somewhat different issues and the data from each type of study contribute slightly different pieces of the puzzle of how social support may protect cardiovascular health. Social support  selection variable. Studies of this type generally ask how self-  reported levels of social support or integration into social networks is related to cardiovascular responses to laboratory stressors. Only three studies to date have addressed this issue, and results are mixed. Married couples (aged 25-44) and caregivers for Alzheimer Disease victims (ages 30-84) showed attenuated systolic blood pressure in response to laboratory stressors when they reported high levels of social support (Kamarck, Peterman, & Marco, 1992; Uchino, Kiecolt-Glaser, & Cacioppo, 1992). A third study (Boyce & Chesterman, 1990) found no association between levels of social support and cardiovascular reactivity to. either social or cognitive tasks in a sample of adolescent boys. Differences in findings between studies may be a function of populations under investigation or of methods of assessing social support. It is noteworthy that the studies by Kamarck et a!. and by Uchino et al. used instruments to assess social support that were developed according to accepted psychometric principles; in contrast, the Boyce and Chesterman study measured social support on the basis of an interview in which  —16—  Lenz  Response to Audience Agreement  subjects reported number of best friends, frequency of contact with friends and family, membership on teams and so forth. The interview-derived data may have been less effective in identifying valid levels of social support than the Social Network Index and the Social Support Index used in the other studies. Alternatively, adolescent boys may react to social stressors and to social support in different ways than adults. Findings from the two studies with positive outcome (Kamarck et al., 1992; Uchino et al., 1992) are far from conclusive, especially when it is noted that only systolic blood pressure and not diastolic blood pressure or heart rate were associated with levels of social support. On the other hand, it must be noted that the very design of such experiments would lead one to predict less than dramatic effects because they are examining protective carry-over effects of social support and not the protective effects of support present during the stressor. Manipulation Qf social support. A number of other studies have brought support itself into the laboratory. Generally, findings in studies where social support is manipulated are stronger than those in which it is utilized as a selection variable. By the nature of the experimental designs, studies with laboratory stress tasks and manipulated social support examine the stress buffering aspects of social support, That is, this type of experiment hypothesizes that the presence of social support (however operationalized) attenuates cardiovascular response to stress. As described at length previously, many writers have found the concept of social support to be poorly defined and vague (e.g., Cohen, S. & Wills, 1985; Coyne & Bolger, 1990). In fact, this very lack of conceptual clarity poses the experimentalist a difficult challenge. How can a broadly-defined concept such as social  —17—  Lenz  Response to Audience Agreement  support—one that by its very nature comprises a broad array of loosely related social events, perceptions, and behaviors—be operationally defined for controlled study in a laboratory? As will be described shortly, experimentalists have met this challenge in a variety of ways. Given the array of behaviors and perceptions surveyed by epidemiologists (e.g., Berkman & Syme, 1979; Blazer, 1982) in their examinations of the relationships of social support with mortality and disease, it is clear that each laboratory paradigm will involve itself with only a small fragment of what is conceptualized as “social support”. Operationalization of constructs always constitutes the crux of experimental approaches, and nowhere is this more apparent than in social support research where the importance of operationalization can hardly be overemphasized. The challenge for the experimentalist is to extract elements from the rather fhzzy concept of social support that can be independently investigated and thereby differentiate active elements from inactive ones. The end result of successfiul approaches will be to simultaneously clarifj the nature of social support and to begin to define mechanisms by which it exerts effects on health. Operationalizing strategies that fail to add to our understanding of the functions of social support may appear reductionist; whereas strategies that differentiate active versus inactive elements of support—however simple—may make small but important contributions to our collective understanding. Because of the breadth of the concept and the necessity for experiments to address operationalized “chips” of the larger issues, this particular area of research relies heavily on accumulations of data from related studies. In designing research on support, investigators need to remain cognizant of an important caveat explained by Cohen and Wills (1985) and by Wills (1985). The buffering  —  18—  Lenz  Response to Audience Agreement  effect of social support will only be detected if the measure of social support assesses resources that are relevant to the needs elicited by stressful events in the individuals’ lives. They were writing about measurement issues, but the point is every bit as valid when applied to the experimental manipulation of support. The application of Cohen and Wills’s principle to laboratory studies suggests that social support will be found to attenuate taskrelated stress reactions if and only if the experime’ntally provided support provides resources that are relevant to the needs elicited by the laboratory stress task. A simple example may serve to indicate the consequences of ignoring this principle during research design. It has been found that being touched while performing mental arithmetic has no impact on perception of support or on performance (Edens, Larkin, & Abel, 1992). This finding is less than surprising when it is considered how little physical touch relates to performance of this task. Similarly, in another study (Lepore et al., 1993), subjects were offered a glass of water by a laboratory confederate intending to be supportive. Not one subject accepted the offer, and this uniformity of refusal may indicate that the offer was irrelevant to the needs evoked by the speech task. In keeping with Cohen and Wills’ principle, it is considered vital that support provided or offered in experimental manipulations ought to fulfill a need related to the task at hand. Presence/absence studies. Some studies have examined the impact of the presence of a person or a pet during a stress task (Allen, Blascovich, Tomaka, & Kelsey, 1991; Edens et al., 1992; Germ, Mimer, Chawla, & Pickering, 1995; Kamarck, Manuck, & Jennings, 1990; Sheffield & Carroll, 1994). Although the mere presence or absence of someone (or of a pet) is inadequate as a general operationalization of social support (Sheffield &  —19—  Lenz  Response to Audience Agreement  Carroll, 1995), these studies examine the impact of a specific type of support best called companionship support (Wills, 1985). Judging from the mixed findings in these studies, companionship appears to have a variable impact on cardiovascular response. Response to a stress task in the presence of a friend has been contrasted with a number of other conditions, including response to the following: (a) performing alone (Allen et al., 1991, Edens et al., 1992; Germ et al., 1995; Kamarck et al., 1990; Lepore et al., 1993; Sheffield & Carroll, 1994); (b) performing in the presence of a stranger (Edens et al., 1992; Sheffield & Carroll, 1994); (c) and performing in the presence of the subject’s dog (Allen et a!., 1991). Allen et al. compared response in the presence of a beloved dog to response alone or with a friend. Compared to response while alone, the dog reduced cardiovascular reactivity to the stressor while the presence of a friend increased it. In contrast, Edens, Abel, and Larkin compared the presence of a friend with that of a stranger and found the stranger’s presence unrelated to cardiovascular reactivity while the friend’s presence attenuated cardiovascular response to the stressor. Preliminary data suggest that differences between these findings may be related to incompletely controlled perceptions of competition with the subjects’ friends (Kors, Linden, & Germ, 1995). At this time, however, the only clear conclusion is that the impact of the presence of someone during a stress task can be expected to vary. This variability may be due to the task, the person performing it, who the “supportive” other is, and to the perceptions of the performer. Studies examining the impact of specific supportive actions. Several specific supportive behaviors have been examined for impact on cardiovascular response. Being touched by a companion appears largely unrelated to cardiovascular responsivity; but if  —20—  Lenz  Response to Audience Agreement  the person offering supportive touch is a stranger to the subject, cardiovascular responsivity is increased (Edens et al., 1992). Kiecolt-Glaser and Greenberg (1984) operationalized high and low supportiveness as personally warm versus neutral interviewer behavior during a pre-experimental interview. They found a supportive interview associated with less responsiveness to the succeeding cognitive tasks event though the interviewer was no longer present during the stress task. Two studies most closely related to the present study deserve description in more detail. Germ et al. (1992) operationalized support as a stranger (lab confederate) either agreeing (support) with the subject on a topic of discussion or saying nothing (no support). Further, the agreeing (supportive) confederate continued to smile, nod, and make eye contact during the subjects speech, and the non-supportive confederate exhibited little facial expression and few non-verbal responses to the subject’s speech. Data analysis indicated that subjects in the non-supportive condition exhibited more cardiovascular response to the speech task than those in the supported condition. Large differences between conditions were observed on all cardiovascular indices (heart rate, systolic and diastolic blood pressure). In a closely related study, Lepore et al. (1993) operationalized support during a speech task as an observer’s carefhlly rehearsed statements of encouragement. The statements included the following comments at prescribed times during the speech: (a) “Remember, it will be over in a few minutes,” (b) “That’s good!”, and (c) “You did fine.” In contrast, the observer in the non-supportive condition did not initiate any verbal behavior and remained inattentive and reserved throughout the speech. If asked a direct  —21—  Lenz  Response to Audience Agreement  question by the subject, the non-supportive observer responded minimally by saying “I don’t know,” or “Do your best.” The observer’s posture was described as open in the support condition and reserved in the non-support condition. Findings in this study were very similar to those of Genn et a!. (1 992)—that is, cardiovascular reactivity was greater in the condition without support than in the condition including supportive observer behaviors. Like Germ and colleagues, Lepore et al. interpreted findings as evidence that social support may protect cardiovascular health by reducing reactivity to stressful social situations. Alternative interpretations. The results of these two studies have been taken as strong evidence that supportive interpersonal behaviors attenuate cardiovascular response to speech-related stressors (Germ  Pieper, Levy, et al., 1992; Lepore et al., 1993; Sheffield &  Carroll, 1994). However, an alternative interpretation is possible and needs to be ruled out before these findings can be taken at face value. This alternative view centers around the fact that both studies used a non-responsive observer in the non-supportive condition. It may be that when the observer is neither supportive nor antagonistic but merely silent and non-communicative, a situation of uncertainty or unpredictability is created for the subject. The subject is in a socially ambiguous situation and does not know the silent confederate’s position, nor can the subject tell for sure whether that silent member will suddenly become a strident adversary (or ally, for that matter). The ambiguity of the observer’s reactions leads to unpredictability and perceived lack of control, and these two factors are known stressors. By contrast, subjects in the “support” condition of both studies knew the observer to be largely supportive and helpful. In short, the experimental intent in both  —  22  —  Lenz  Response to Audience Agreement  studies was to construct groups contrasting on only one dimension, the supportiveness of the observer. This goal was not achieved because experimental conditions also varied along another dimension related to stress: the clarity and predictability of the observer’s behavior. Not knowing the opinion or response of a non-communicative audience could accentuate cardiovascular responsiveness to the speech task through three different mechanisms—none of which is directly tied to social support. First, unsupported subjects in the Germ et al. study (1992) may have worried that the silent member would eventually join in the discussion in concert with the subject’s attackers. This would be, in general, a fear or worry response and, as such, would likely increase cardiovascular responsiveness. A second possibility—one applicable to both studies in question—is that the subjects could increase intensity of their arguments in order to win over the “undeclared” or non committal observer. This would be an expectancy response and could increase cardiovascular responsiveness through effort or involvement. A third possibility—and probably the most likely of the three—is that the undeclared position of the silent participant may create an ambiguous situation for the subject. It is unclear to the subject whether to defend against, win over, or ignore the silent listener. Not knowing the silent participant’s position or intent may create social ambiguity in the situation and thereby decrease the subjects’ sense of control over the situation and decrease the subjects’ appraisal of their ability to predict situational outcome. Since ambiguity, low confidence in predictability, and lack of control are all known to be potent stressors (Burger, 1992; Germ, Pieper, Marchese, & Pickering, 1992; Katz & Wykes, 1985; Weiss, 1968), it is  —23—  Lenz  Response to Audience Agreement  essential that these factors be controlled during the stress induction procedure. Without controls on factors related to social ambiguity, the two groups cannot be regarded as equivalent in all respects except presence or absence of social support. It is the major goal of the present study to gather data to evaluate the possible validity of this alternative interpretation of the Germ et al. (1992) and the Lepore et al. (1993) findings. Differences between laboratory stress tasks. Experimental examination of the impact of supportive behavior on cardiovascular reactivity requires a stressor capable of evoking a substantial cardiovascular response. Stress tasks used to date have been either cognitive or psychosocial. Cognitive tasks include performance of mental arithmetic (Allen et al., 1991), playing a video game without interference from others (Germ et al., 1995), a verbal anagram task (Uchino et al., 1992), and the Stroop color-naming task (Kamarck et al., 1992). Stress tasks with psychosocial components have included giving a speech on a controversial topic (Germ, Pieper, Levy, et al., 1992; Lepore et al., (1993); Allen, & Evans, 1993), playing a video game in a competitive setting (Germ et al., 1995), and identifying the artist of paintings while an art expert watched and commented (Sheffield & Carroll, 1995). Of these approaches, the strongest cardiovascular response to the stress tasks has been reported in those requiring subjects to prepare a speech on a strongly-held personal opinion in front of two laboratory confederates expressing opposing views (Germ  Pieper, Levy, et al., 1992) and in front of a video camera and one laboratory  confederate (Lepore et a!., 1993). Other investigators have determined speech tasks of this general type to be the most potent of stress tasks currently in use in stress research (Blumenthal et al., 1992; Fredrikson & Matthews, 1990). For this reason, and because the  —24  —  Lenz  Response to Audience Agreement  present study built on previous research using similar tasks, subjects in the present study were asked to give a speech on a controversial topic.  Goals and Import of the Present Study The line of inquiry in this study was intended to encourage development of social interventions effective in the reduction of cardiovascular—and to some extent other— stress-related diseases. The general approach was to close conceptual gaps left by previous research and to replicate and extend earlier findings in a manner that could lead to more reliable interpretation. In summary form, the three major goals of the present study were as follows: (a) to clarif,’ an interpretive difficulty regarding the role of ambiguous audience response in heightening stress levels, (b) to extend the generalizability of the experimental reactivity work on social support by including male subjects and by making direct gender comparisons in reactivity, and (c) to examine whether differences in cardiovascular reactivity are associated with differences in levels of self-esteem. These three goals and three additional minor goals are explained in detail below. Clarify the effect of ambiguous audience response. Two of the previous studies (Germ, Pieper, Levy, et al., 1992; Lepore et al., 1993) involving speech tasks as stressors and audience agreement as a socially supportive behavior both found the following: (a) The speech tasks themselves evoke substantial cardiovascular response, and (b) this response is significantly lessened when the audience expresses agreement relative to conditions in which the audience is non-committal and largely non-responsive. The  —25—  Lenz  Response to Audience Agreement  findings have been taken to indicate the buffering or protective effects of one element of social support (supportive agreement) on cardiovascular responsiveness. In short, the results have been interpreted to mean that social support buffers the reactivity to the laboratory stress procedures. Although these findings are intriguing and potentially important, a thorough review of the literature suggests that they are far from conclusive. The literature on active coping and informational control offers an alternative interpretation of these results. The social awkwardness or ambiguity that results from speaking to a non-responsive audience may heighten reactivity. To the extent that this is so, a non-responsive audience may be an inappropriate control or comparison condition for identifying the reduced cardiovascular response due to social support. It is apparent that if a comparison group increases reactivity, then the apparent attenuation due to social support would be exaggerated. In order to address this issue of interpretation, the present study includes conditions in which the audience (a) clearly disagrees with the subject, (b) clearly agrees with the subject, or (c) remains non-committal and non-responsive to the subject’s expression. Comparison of cardiovascular response in these three conditions allows conclusions to be drawn concerning the independent contributions of (a) clarity of the audience’s reaction to the subject’s speech and (b) the favorableness of the reaction (i.e., whether the audience agrees or disagrees with the subject). Only when these two contributions are separated can conclusions be drawn relating to the impact of social support on cardiovascular reactivity. Examine sex differences. In spite of sex differences in cardiovascular response and in utilization of social resources (Berkman et al., 1993), most previous experiments on  —  26  —  Lenz  Response to Audience Agreement  cardiovascular response to social support have used only female subjects. No reason for this selective approach has been given by researchers for this selective approach, but sampling convenience may have played a role in some studies. For example, at least two studies using female subjects were conducted at all-women colleges using students as subjects (Germ et al., 1995; Germ  Pieper, Levy, et al., 1992). The exceptions are Lepore  et al. (1993) who found sex differences in reactivity to the stress task but not in response to social support and Sheffield and Carroll (1993, 1995) who reported generally negative findings in relation to social support but who found men more reactive than women. Evidence thus exists that men and women may respond differently to stressors and may respond differently to social support. To date, however, experiments involving aspects of social support and cardiovascular reactivity have seldom been designed explicitly to assess sex differences. The present study proposes to fill this gap in the literature. Examine the relationship between agreement and self-esteem. Three types of social support identified by Wills (1985) that are salient to the present study are esteem support (also referred to as emotional support), social companionship, and appraisal support. Companionship was provided to subjects in all conditions. Of all types of support delineated by Wills, it appears most appropriate to consider agreement or disagreement with a speaker’s opinion as related to a combination of appraisal and esteem support. One likely hypothesis concerning the action of social support is that the agreement of another person with a strongly held opinion may decrease the stress of self-declaration by increasing self-esteem and the concomitant belief in the value of one’s opinions. The present investigation examined these proposed relationships by measuring state self  —27—  Lenz  Response to Audience Agreement  esteem prior to and following engagement in the speech task. It was predicted that the supp6hve (or agreement) condition would have a less deleterious effect on self-esteem than the other two conditions—or perhaps it may even augment self-esteem. It was also of interest to determine whether a non-committal or non-responsive audience led to larger decrements in self-esteem than open disagreement. Such an outcome would be in keeping with basic tenets specified in Pragmatics of Human Communication (Watzlawick, Bavelas, & Jackson, 1967) because acknowledgment of an individual’s expression (regardless of whether it is believed or accepted) is regarded as more satisfactory and less threatening to self-esteem than a failure to respond. Additional minor goals. Previous laboratory examinations of effects of social support on cardiovascular reactivity have failed to find substantive relationships between psychophysiological measures and self-report of stressfullness of the laboratory procedures. This desynchrony is a common finding in laboratory stress research as well as in examination of anxiety-related phenomena both in the clinic and in the laboratory (e.g., Rachman, 1990). Though a common finding, dissociation between self-report and physiological measures frequently presents interpretive difficulties in psychophysiological research. It is not clear to what extent this phenomenon relates to the subject’s lack of awareness of emotional and other physiological states and to what extent it is the result of inadequacy of assessment. Since it is typical for post-experimental questionnaires to be somewhat ad hoc, assessment failures may well predominate. Several steps were taken to address this problem in the present study. Subjects were asked to complete the Affect Grid (Russell, Weiss, & Mendelsohn, .1989), a measure chosen because it is a simple,  —28—  Lenz  Response to Audience Agreement  psychometrically sound way of assessing arousal and positive versus negative emotional tone. Analysis of subjective response to the experiment as measured on the Affect Grid and comparison of these reports with cardiovascular reactivity may contribute to our understanding of the general desynchrony between self-report and physiological measures. The study also included a number of self-report ratings of the subjects’ experiences during the experimental procedures. Questions incorporated features of self-report measures that have been found associated with either experimental condition or with cardiovascular reactivity in previous research. They can be categorized in the following ways: (a) reports of the subjects’ perceptions of their own emotional states during the speech task, (b) reports of the subjects’ perception of their own performance in the speech task, (c) reports of the subjects’ perception of the behavior and intentions of others. These self-report data were used to check the impact of experimental manipulations and to examine cognitions related to appraisal of the experimental procedures. Finally, the study explored the relationship between (a) subjects’ reports of the extent and adequacy of their existing social network and (b) response to experimentally provided support. There is evidence that report of supportiveness outside the experimental setting is associated with reduced cardiovascular response to laboratory stressors (e.g., Kamarck et al., 1992; Kiecolt-Glaser & Greenberg, 1984; Uchino et al., 1992) and with lower ambulatory blood pressure (Linden et al., 1993). There is also evidence that support during the laboratory stress procedures leads to attenuated cardiovascular response (e.g., Edens et al., 1992; Germ, Pieper, Levy, et al., 1992; Lepore et al., 1993). It has not been investigated whether report of a satisfactorily supportive social network outside of the  —  29  —  Lenz  Response to Audience Agreement  laboratory is associated with response to provision of support in the laboratory. Any such relationships are likely to be complex, and it was beyond the scope of the present study to provide definitive data on the question. The intent was to explore possible relationships and suggest directions for future research. To this end, subjects were asked to complete the Interpersonal Support Evaluation List (ISEL) (Cohen, S. et al., 1985) at the beginning of the study, and correlations between the ISEL and self-report and cardiovascular responses were examined. Summary of goals. The present study undertook the following three major goals: (a) to clarify the impact of using a non-responsive audience as a control group, (b) to examine sex differences in response to experimentally provided social support, (c) to assess the impact of experimentally provided support on self-esteem. In addition, two minor goals were addressed. Self-report measures of affective state and reaction to the experimental procedures were refined to overcome frequently reported difficulties in interpreting desynchrony between cardiovascular and subjective responses. Finally, a measure of existing environmental social support was included to enable exploration of the relationships between subjects’ existing social support and their response to experimentally provided support.  —30—  Lenz  Response to Audience Agreement  METHOD  Overview Male and female university students were asked to deliver a five-minute speech declaring and defending their position on a controversial topic. The speech was made in front of a same-sex lab confederate introduced as another subject and in front of a video camera. Confederates acted in one of the three following manners: 1) The confederate expressed agreement with the subject and a pleasurable response to the speech; 2) the confederate gave no feedback concerning his or her own position or reaction to the subject’s performance; or 3) the confederate expressed disagreement with the subject and displeasure with the subject’s speech. The training of lab confederates emphasized equating the following non-verbal communications across conditions: posture, initiation of eye contact, and proximity to the subject. Blood pressure and heart rate were monitored continuously prior to, during, and after the speech task. Self-report measures of state selfesteem and affective state were taken before and after the speech task. A measure of social network size and satisfaction was taken prior to the speaking task, and a variety of selfreport measures were taken after the speech to assess reactions to the task, to the experimental manipulation, and to assess self-report of stressftilness of the situation.  Subjects The final data analysis included measures from 90 undergraduate university students (45 women; 45 men) recruited for participation from the University of British Columbia,  Department of Psychology subject pooi. Participating subjects received class credit for —31—  Lenz  Response to Audience Agreement  participation in the study. Mean age was 19.18 (SD Maximum ) 87 (t(  =  =  2.13; Minimum  =  17,  34), and there was no age difference between men and women in the sample  —0.79, p  =  .43).  Ethnicity. The ethnic backgrounds of these 90 subjects were as follows: 68 were born in North America (75.6%), 15 in Asia (16.7%), four in Europe (4.4%), and three in other places (3.3%). Forty-three subjects reported speaking only English (47.8%) while 40 reported fluency in two languages (44.4%) and seven reported fluency in three or more languages (7.8%). The favored language was reported as English by 88 (97.8%) of the subjects with one subject reporting preference for Cantonese and another for Japanese. Self-described ethnicity was Canadian for 45 of the subjects (50%), and some variant of Asian-Canadian for 33 (36.7%). Four (4.4%) described themselves as a variety of European-Canadian, and eight (8.9%) indicated national identities that were not easily categorized as ethnic groups. Recruitment. A call for subjects was posted on the subject notice board (Appendix A1), and invitations to participate were delivered in appropriate undergraduate classes. Volunteers indicating interest in the study were telephoned to determine suitability and, if found appropriate, to schedule participation. Potential subjects were excluded from the study if they acknowledged diagnosis of any of the following physical conditions: hypertension (i.e., blood pressure greater than 145/90 mmHg), cardiac disease, liver or kidney disease, or diabetes; two subjects were excluded for medical reasons. Because the added effort of speaking in an unfamiliar language could cause increased blood pressure for subjects who were not fluent in English, it was decided to exclude subjects whose  —32—  Lenz  Response to Audience Agreement  major language of instruction was other than English after age ten. The largc immigrant population in Vancouver resulted in numerous exclusions for this last reason. Fully qualified subjects were asked to refrain from caffeine, alcohol, and vigorous exercise for at least two hours prior to their scheduled participation. (See Appendix A-2 for screening instructions.) Attrition. Although complete data sets were available for the 90 subjects described above, fifteen additional subjects came to the laboratory and participated in portions of the protocol. Two subjects were excluded for failing to meet language fluency requirements. One subject requested that the experiment be stopped when she was selected to be the speaker. The experiment was stopped on two occasions because the confederate was mistakenly chosen to be the speaker. Three subjects were unable to speak for the entire five-minute period; of these, two were in the Disagree condition, one in the Agree condition. Two subjects were dropped after they reported suspecting that the assignment of the role of speaker was rigged. The most common cause of interruption of the protocol was failure of the subject to declare a clear opinion with which the confederate could disagree. Data from five subjects were omitted from analysis for this reason. All data descriptions and analyses include only the 90 subjects completing the entire protocol.  Self-report Measures Self-esteem. Self-esteem was measured twice with the State Self-Esteem Scale (SSES) (Heatherton & Polivy, 1991). The first measure was taken before the subject  —33—  Lenz  Response to Audience Agreement  knew whether he or she would be assigned the role of listener or speaker. The second was after the speech was completed and after the post-task recovery period. The SSES consists of 20 self-statement items. The subject rates the truth of each item at the present moment on a five-point Likert-type scale. Anchors for the scale are as follows. 1  =  not at all, 2  a little bit, 3  =  somewhat, 4  =  very much, 5  =  extremely. The  S SES yields a total score and scores on three factor- analytically derived subscales— performance self-esteem, social self-esteem, and appearance self-esteem. The SSES has been shown to be a reliable and valid measure of clearly defined aspects of self-esteem that are subject to change over time. Because state self-esteem is by definition variable over time, test-retest measures of reliability are inappropriate. Reliability of the instrument is based largely upon a high degree of internal consistency (coefficient a  =  .92) (Heatherton & Polivy, 1991). Construct validity of the scale was  established in a series of studies examining the relationships between SSES total and subscales and a variety of other conceptually related measures. SSES total score was found to be positively correlated with global self-esteem, social desirability, and satisfaction with body shape and to be negatively correlated with hostility, depression, anxiety, and likelihood of being currently on a diet (Heatherton & Polivy, 1991). Subscales were found to exhibit appropriate associations with these other constructs. For example, the performance self-esteem subscale was most highly correlated with depression and anxiety (both negative correlations), the social self-esteem subscale was most highly correlated with social self-esteem as measured by the Janis-Field Feelings of Inadequacy Scale (Janis & Field, 1959); and the appearance self-esteem subscale was most highly  —34—  Lenz  Response to Audience Agreement  correlated with satisfaction with current figure, dieting behavior, and body size estimation (Heatherton & Polivy, 1991). Discriminant validity of the SSES subscales is supported by experimental evidence (Heatherton & Polivy, 1991). Students told they would be facing a very difficult test registered decreases in performance self-esteem but no changes in appearance or social self-esteem. (See Appendix A-5 for the complete SSES.) Social support. Levels of social support in subjects’ everyday life were assessed with the Interpersonal Support Evaluation List (ISEL) (Cohen, Set al., 1985). The ISEL is a 40-item questionnaire in true/false format intended as a measure of perception of availability of social support. Subjects are asked to indicate whether each statement is “probably true” or “probably false” about them. The instrument produces a total score and scores on each of the following four subscales: appraisal, belonging, tangible, and selfesteem. The appraisal subscale measures the availability of others whose opinions and perceptions are trusted. The belonging scale evaluates the extent to which subjects feel that they are integrated into a social network. The tangible subscale assesses availability of physical help. Finally, the self-esteem scale gauges availability of individuals who help subjects feel good about themselves. Each of the subscales contains 10 items. The general population version of the ISEL used in the present study was standardized on four samples. Test-retest reliability is considered adequate at r = 87 over a two-day period and r  =  .70 over a six-week period. Six month test-retest data indicate  general stability of the measure (r = .74). Internal consistency is strong (coefficient  cc .88 to .90). Examinations of consistency of subscale scores have yielded cc ranging =  from .62 to .82 depending on the scale and the sample in question. In brief, test-retest data  —35—  Lenz  Response to Audience Agreement  and internal consistency measures of the ISEL indicate adequate reliability (Cohen, Set a!., 1985). The validity of the ISEL has been investigated in a number of ways. It is uncorrelated with the Marlowe-Crowne Social Desirability Scale, an important indication that response bias does not interfere with subjects t scores on the ISEL. As with many measures of social support, face validity and construct validity have been emphasized. In addition, ISEL scores have been compared to a number of other measures of social support with resulting moderate correlations. Subscales have been compared extensively with other measures providing strong indications of discriminant and convergent validity. The reliability and validity of the ISEL have been supported by comparison of psychometric properties of the ISEL with other measures of social support (Heitzman & Kaplan, 1988). The authors ofthe ISEL have published correlations between subscales for four different samples. In the largest of these samples (154 men and 62 women) the correlations range from r = .64 (Belonging / Self-esteem) to r = .48 (Appraisal / Selfesteem). Correlations of this magnitude between subscales may limit the validity of conclusions drawn from individual scale scores, and as a result only the total score is used in the present study. (See Appendix A-6 for the complete ISEL.) Affective state. To assess changes in report of the subject’s general affective state during the speech, subjects were asked to complete the Affect Grid pre- and post-speech task. The Affect Grid is a single-item scale assessing affect along the dimensions of pleasure-displeasure and arousal-sleepiness. It is intended for simplicity and validity of repeated administration (Russell et al., 1989). The instrument was developed in  —36—  Lenz  Response to Audience Agreement  accordance with prevailing theories indicating that affect is most parsimoniously represented by the two orthogonal factors measured by the grid (e.g., Russell, 1979; Watson & Tellegen, 1985). The Affect Grid is presented to subjects as a nine-by-nine grid, and subjects are asked to indicate “how you are feeling right now” by placing an “X” in the grid. Polar opposites are labeled “High Arousal” and “Sleepiness” on the vertical dimension and “Unpleasant Feelings” and “Pleasant Feelings” on the horizontal. In addition, corners of the grid are labeled to assist subjects in identifying affective states that are extreme on both dimensions. Because of the single-item format, internal consistency measures are not possible, and because the affective states measured by the Affect Grid are assumed to be temporally unstable, test-retest strategies are inappropriate. Reliability was—for these reasons— assessed indirectly and report of psychometric properties centered on issues of validity. Comparisons have been made between Affect Grid ratings of emotion-related words and facial expressions and ratings of similar materials by means of a more time-consuming semantic-differential method with established high internal consistency (Mehrabian & Russell, 1974). Correlations generated by this technique were universally high (r from 91 to .97) for related scales and non-significant for scales hypothesized to be orthogonal (Russell et a!., 1989). Using a sample of 162 university students, comparisons between self-report of current mood as measured by the Affect Grid and the semantic differential technique produced correlations of .77 for the pleasure component and .80 for arousal. These correlations are taken as evidence of adequate convergent validity. (See Appendix A-7 for the Affect Grid.)  —37--  Lenz  Response to Audience Agreement Speech topics. Topical issues of concern to the subject were identified by means of  the Important Topics questionnaire, an instrument designed for the present study. Subjects were asked to rank the three most important topics from a list of 11 current issues. The following is a list of Issues named on the questionnaire with number of subjects speaking on each topic in parentheses: abortion rights (28), death penalty (11), physician-assisted suicide (10), the Young Offenders’ Act (10), clear-cut logging (8), the use of United Nations “peacekeeping” forces (6), legal handling of pedophiles (4), animal experimentation (3), refhgee status for immigrants (2), first nations land claims (1), and penalties for environmental protesters (0). Response lines labeled “Other topics” were included to allow subjects to name their own topics (7 subjects chose to do this). Instructions indicated that rankings should be based on personal belief about the importance of the topics and on personal knowledge of the issue. Instructions finished as follows: “You should care strongly about and know a good deal about your number one choice.” (See Appendix A-8 for the Important Topics questionnaire.) Subjective response to the experiment. A questionnaire entitled “Reactions to Speaking” was designed for use in the present study. Each question was accompanied by a 10-centimeter line with labeled ends, and subjects were asked to “mark each line to indicate how you think or feel.” Items were arranged into three sets. The first set assessed the general stressfl.ilness of the speaking experience by asking subjects to mark lines at appropriate positions between the following pairs of adjectives: stressed/relaxed, safe/unsafe, uncomfortable/comfortable, anxious/not anxious, calm/excited, worricd/content, and not nervous/nervous. The second set consisted of nine items which  —38—  Lenz  Response to Audience Agreement  were presented in question format. The subjects were asked to mark a line somewhere between “Not at all” and “Very Much” to indicate an answer. Questions in this set assessed two separate issues: (a) the subjects’ perceptions of their performance during the speech task, and (b) appraisals of the behavior of the listening student. In addition, two items in this section rated the general aversiveness of the experience. These items were closely related to the only self-report items correlating with cardiovascular measures in the Germ et al. (1992) study. Questions in the final set were intended as a manipulation check. Subjects were asked to describe the listening student by marking a 10-centimeter line between the following pairs of adjectives: supportive/unsupportive, close/distant, rejecting/accepting, ffiendly/unfriendly, warm/cold, and helpful/unhelpful. Like items in the first set, these were adapted from the manipulation check used successfully in a related study by Lepore et al. (1993). An additional set of twelve items was appended to the “Reactions to Speaking” questionnaire for approximately the last half of data collection. These items asked subjects to rate features of the study, indicating “how upsetting (that is, bothersome or anxietyprovoking)” each feature was for them. As with the other items in the questionnaire, ratings were made by marking a position on a 10-centimeter line. Anchors for these items were “Not at all upsetting” and “Very upsetting”. The items listed experimental features new in the present study and other features shared with the Lepore et al. study (1993). The added questions were exploratory in nature. Observation of subjects’ responses in the laboratory and preliminary data analysis showed that subjects were evidencing greater cardiovascular response to the speaking task than had been anticipated. The  —39—  Lenz  Response to Audience Agreement  additional items were included in an attempt to identi1,r environmental and procedural features accounting for high levels cardiovascular reactivity. A brief analysis of these additional items is presented in the Discussion section to aid in explaining the divergence of responses in the present study from those in the study by Lepore et al. (1993). (Reactions to Speaking, including the added twelve items, is included in Appendix A-9.) Instrumentation. Blood pressure and heart rate were monitored repeatedly at approximately two-minute intervals using a Dinamap 845 sphygmomanometer. This instrument uses an oscillometric method that determines arterial pressure by measuring cuff pressure oscillations as pressure in an automatically inflated arm cuff is reduced by discrete increments. A built-in micro-processor tests oscillation data for artifacts, averages obtained values and then displays heart rate, mean arterial pressure, and systolic and diastolic values in digital format. A single blood pressure determination cycle is completed in 40 to 60 seconds. Blood pressure readings by the Dinamap 845 have been shown to correlate highly with manual Baumanometer readings performed by the standard auscultatory method (systolic: r  .95; diastolic: r = .92; and heart rate: r = .99; (Linden &  Zimmermann, 1984).  Procedure Instructions to subjects. On arriving at the laboratory, subjects were introduced to the same-sex experimental confederate as another subject, and instructions were given to the subject and confederate together. They were told that the study dealt with physiologic changes during speaking and listening. The experimental tasks were described as follows:  —40—  Lenz  Response to Audience Agreement  According to a random draw, you will be assigned to be either the speaker or the listener in this experiment. The one of you that is chosen to speak will be asked to prepare and deliver a brief talk- expressing your opinions to the other on a topic about which you care deeply. If yoi are chosen to be the listener, you will be asked to listen as silently as possible to the opinions of the other. Blood pressure and heart rate will be monitored throughout the experiment for both of you—including a preparation period and a resting period after the speech. The speaking/listening portion of the session will be recorded on videotape so we can analyze it later. Subjects were then asked to complete a consent form (Appendix A-3), the first SSES, a demographics questionnaire (Appendix A-4), the ISEL, the first Affect Grid, and the Important Topics sheet. The order of these questionnaires was varied to avoid systematic impact of order. Confederates completed matched sets of questionnaires. Subject and confederate were asked not to speak to each other except during the portion of the experiment designated for speaking. The subjects were asked to complete the questionnaires, check their work for completeness, and then lay their clipboard down to signal they were finished. They were told that 15 minutes were allotted for questionnaires and following that time period there would be a five-minute rest period to relax and look at cartoon books provided within easy reach. Cardiovascular baseline. After completing the consent form and before the questionnaires, the subject and confederate were fitted with cuffs of the Dinamap 845 on —41—  Lenz  Response to Audience Agreement  their non-dominant arms. The operation of the machine was explained, and they were asked to relax their arms whenever measures were being taken. The experimenter left the room and initiated two blood pressure readings within the first three minutes of form completion. No further measures were taken until the rest period following completion of the questionnaires. An adaptation period of at least 20 minutes was maintained before initiation of the speech preparation. This period allowed subjects to adapt to the room and the experimental situation. Accordingly, a minimum of 15 minutes was allotted for form completion and five minutes for ensuing rest period, during which baseline cardiovascular measures were taken. If subjects finished form completion in less than 15 minutes, the rest period was increased to make a total of 20 minutes of adaptation. Subjects took an average of 13 minutes, 15 seconds to complefe the forms (minimum  8:3 0;  maximum = 19:50). When forms were complete, the experimenter told the subject and confederate over an intercom the length of the rest period, reminded them to relax, and indicated that three more blood pressure measures would be taken during the rest period. Proximity and room arrangement. Throughout the experiment—including form  completion, rest periods, and throughout the speech task—subjects remained seated in the same position. Subject and confederate sat faeing each other at an oblique angle. Measuring from the nearest point, chairs were 75 centimeters apart and positioned at an angle of approximately 135 degrees. The room measured 2.7 by 3.3 meters. Directly across from the subject was a one by one meter one-way mirror, and subjects were informed that the experimenter observed the experiment through the mirror and could hear  —42—  Lenz  Response to Audience Agreement  speech on an intercom. A video camera was placed on a tripod in the far corner of the room aimed at both the subject and confederate. Speech preparation. After the five-minute (minimum) rest period, the experimenter re entered the room and took clipboards holding completed forms from the subject and from the confederate. While removing the Important Topics form from each clipboard, the experimenter said, “Now, I would like each of you to prepare a speech. on the topic you ranked highest on this form.” The experimenter showed the subject and the confederate each the form they had completed, and pointing to the highest ranked topic, said to each, ‘This is the topic you are to prepare to speak on,” The experimenter then gave the following instructions for the speech preparation:  You will be given five minutes to prepare a statement that is to be five minutes long. In this speech, please explain exactly how you feel about the topic you are assigned. Don’t pull any punches in your statement. Please express your opinions and feelings as strongly as you can. Your goal is not to convince your listener but rather to simply do as good a job as you can in expressing your position clearly and strongly. You may use this paper and pen to make notes to focus your talk (laid clipboard andpen in front of the subject). Of course, with only five minutes to prepare, you will not have time to write the whole speech. Instead, concentrate on making notes that will remind you of major points as you speak.  It is important that you begin your talk by stating your position in a clear and straight-forward manner. Right at the beginning, please tell what your topic is  —43  —  Lenz  Response to Audience Agreement  and where you stand on it. After your position is clear, explain why you feel this way or how you arrived at this conclusion. Whenever possible give examples and arguments that support your opinion. In addition, you may choose to attack arguments that are typically leveled against your point of view. The next page on your clipboard is the “Speech Helper. It is intended to help you generate ideas. You may use it however you like and may write notes on it and on the blank sheet. (See Appendix A-b for the complete Speech Helper form.)  Again, you will have five minutes to prepare a five minute speech. Please prepare yourself to speak the entire five minutes with minimal pausing. Your blood pressure and heart rate will be monitored repeatedly as you prepare and also as you speak. If you take notes as you prepare, please be careful not to move your left (or right f left-handed) arm or hand; movement disturbs the recording. If you have any questions before beginning to prepare your speech, I can answer them now. The experimenter offered brief answers to the subject’s questions and then left the room once again. Role assignment. At the end of the five-minute speech preparation period, the experimenter returned to the room carrying two unmarked envelopes. After explaining that the preparation time was over, the experimenter said, “Now we will find out who will speak and who will listen.” Holding the envelopes for all to see, the experimenter said, “One of these says ‘Speaker’ inside and the other says ‘Listener’.” The experimenter then  —44  -.  Lenz  Response to Audience Agreement  shuffled the envelopes, keeping track of their position while appearing to mix them up. The “huffle” was stopped with the ‘Listener” role in a pre-arranged position, and the experimenter said, “Now I’d like you to choose your roles. ’ The two envelopes were then t held out towards the confederate, who was seated closest to the experimenter. Acting as if considering the choice carefully, the confederate chose the envelope in the pre-arranged position, and the other was offered to the subject. Both were asked to open their envelopes and show the contents. At this point, the experimenter said to the subject, “You got the harder role today,” and to the confederate, “and you got the easy one. Now, I need to give you both some additional instructions before you begin.” Speaking task. The confederate “listener” was instructed within the hearing of the speaking subject as follows:  Even though you have an easy role, there are some particular things I would like you to do. First, please put your clipboard down and try to put the speech you have been preparing out of your mind entirely. Your job is to listen closely to what  —  (subject’s name) has to say. Please try to listen as naturally as you can.  As we listen, most of us feel the urge to comment from time to time, and if this happens, you are asked to keep your comments brief and to the point. Remember, this is _‘s (subject’s name) talk, and we don’t want you to take the floor from her (or him) or turn the speech into a discussion. The speaking subject was instructed as follows:  —45—  Lenz  Response to Audience Agreement  Your task, as I described before, is to speak for five minutes on the topic of_ (the chosen topic is named). You are to start out by declaring your position very clearly and to continue for the rest of the speech by expressing your opinions and your feelings on this topic as clearly and as forcefully as possible. You may use any notes you have written. Please speak directly to  —  (confederate’s name),  telling her (or him) exactly how you think and feel. If you find yourself running out of things to say, it is fine to repeat something you have already said. It is more important for our purposes that you continue to speak than for the speech to be perfectly structured. If you find yourself at a loss for words, repeating your opening statement would be a good idea. Both were given these final instructions:  Because this monitoring equipment is sensitive to movement, we ask that you both remain as still as possible while you are speaking and listening. It is very important that you do not move the arm or hand that is attached to the monitor—especially while the cuff is deflating.  I am going to turn on this video camera now (she or he turns it on) and let it know where we are in the experiment (experimenter holds in front of the camera a sheet ofpaper with subject number on it). After I leave the room, it will take a few seconds to start the timer. I will signal you on the intercom when to begin speaking. Please wait for the cue; I will let you know again when the speech time is up.  —46  —  Lenz  Response to Audience Agreement  After leaving the room, the experimenter behaved exactly as described, cueing the subject over the intercom at the beginning and end of the speech period. Experimental conditions. Subjects were randomly assigned within gender to one of three experimental conditions differing only in the behavior of the confederate during the subject’s speech (n  15 for each gender in each condition). Confederates were careftully  trained to respond in a prescribed manner. Posture, proximity to the subject, gaze, and initiation of eye contact were controlled through extensive training in order to equate these behaviors across conditions. Descriptions of the three conditions follow:  1) Agreement. Confederates expressed, general agreement with the subject& opinions and approval of the presentation of them. In this condition, confederates listened to the instructions and to the very outset of the subject& speeches with an attentive but neutral expression. As the subjects declared their positions, the confederates expressed pleasant surprise by widening eyes and smiling slightly. Immediately after the initial declaration of position—at first opportunity to speak without intermpting—the confederates nodded in agreement and said, “I feel the same way. ’ Throughout the remainder of the speech, the confederates listened t attentively, and returned eye contact initiated by the subject as in all three conditions. The confederates’ facial expressions ranged from serious and neutral to approving; at no point was disagreement portrayed. Major points in the • subjects’ speeches or illustrative examples were met with nods, smiles and on three occasions with situationally appropriate comments chosen from the following list: “Yes. Right. Exactly. That’s really good. I never thought of that  —47  —  Lenz  Response to Audience Agreement  before. So true. Good point. “If subjects asked for help at any point, the confederates replied, “You’re doing fine on your own,” or “You’re saying it better than I could.”  2) Neutral. In this condition, confederates did not speak and expressed neither  agreement nor disagreement with non-verbal signs. Confederates were trained to exhibit a generally impassive facial expression in this condition and to remain neutral and non-committal throughout the speeches. The intent of the training was to convey neither pleasure nor displeasure while listening. Eye contact was reciprocated as in the other two conditions. In this condition, if the subjects asked or made a non-verbal bid for help or assistance, the confederates siñiply shrugged or said, “I don’t know.”  3) Disagreement. In this condition, confederates expressed disagreement with the subjects’ opinions. As in all conditions, the confederates listened to the instructions and to the very outset of the subjects’ speeches with an attentive but neutral expression. As the subjects declared their positions, the confederates expressed displeasure and surprise by squinting slightly and frowning. Immediately after the initial declaration of position—at first opportunity to speak without interrupting—the confederates shook their heads slightly and said, “I couldn’t disagree more.” Throughout the remainder of the speeches, the confederates listened attentively and returned eye contact as in all three conditions. The confederates’ facial expression ranged from serious and neutral to  —48--  Lenz  Response to Audience Agreement  disapproving; at no point was agreement portrayed. Major points or illustrative examples in the subjects’ speeches were met with head shakes, slight frowns, and on three occasions with situationally appropriate comments chosen from the following list: “No. That’s wrong. That can’t be. You’re not looking at the big picture. That doesn’t make sense. I can’t believe that.” If subjects asked for help at any point—an infrequent event in this condition—the confederates replied, “You’re on your own here,” or “I can’t help with this.” Recovery period. At the end of the five-minute speech period, the experimenter said over the intercom, “OK, that’s fine. You may stop now.” Re-entering the room, the experimenter turned off the video camera and said, “Now I’d like you both to fill in a form right away.” The second Affect Grid was presented, and the subjects were instructed to fill it in describing how they felt while speaking. The confederates were instructed to fill it in describing how they felt while listening to the speech. Both participants were then instructed to try to put the speech out of their minds and to rest for the next five minutes by sitting still or by reading cartoon books as at the beginning to the study. Post-speech questionnaires. At the end of the five-minute recovery period, the experimenter returned to the room and announced that no more blood pressure measures would be needed. Cuffs were removed from the subject’s and confederate’s arms. At this time, the self-esteem scale (SSES) and the Reactions to Speaking questionnaires were introduced. While instructions for filling out the last questionnaire were given, the confederate was told, “Since you did not speak, we have a different questionnaire for you called Reactions to Listening. It is on the table in the other room, and you can fill it out  —49—  Lenz  Response to Audience Agreement  there since you no longer have to be fastened to the blood pressure machine. Your form is just like this one, so please listen to these instructions,” Instructions on marking the analog scale were then given. At the end of the instructions, the confederate was led to the adjoining room and the subject left to complete the final set of questionnaires alone. Debriefing. Subjects were debriefed orally and given a printed debriefing that explained the purposes of the study. (See Appendix A-il for the written study explanation form and Appendix A-12 for the complete script for the oral debriefing.) The form informed subjects how to contact the laboratory at a later time if they so desired and explained how to write or phone for a summary of the findings when data analysis was complete. Subjects were informed that a major hypothesis of the study is that blood pressure and heart rate are affected by our perception of others’ responses to us in social settings. It was then explained that the “other subject” was in fact an employee of the laboratory who was trained to act in a particular way. All subjects were asked whether they suspected the deception or the purposes of the study. At the end of the debriefing, subjects were introduced to the confederates and were encouraged to ask questions about the way they really felt about the speech topic or about the subject’s performance. Before they left the laboratory, subjects were told the general pattern of their own cardiovascular responses. In so far as possible, subjects were assured of the normality of their reactions during the study.  —50—  Lenz  Response to Audience Agreement  Analytic Strategy ANOVA designs. The major analysis was conducted with a repeated measures analysis of variance with two between-subjects factors, sex and experimental condition (agree, neutral, and disagree), and one within-subjects factor referred to as “task” (baseline versus speech). Means of CV measures at two time periods constituted the repeated measures: The first (baseline) was the mean of the last two measures during the rest period; the second (speech) was the mean of the three measures taken during the speech task. For analyses of self-report data (affect and self-esteem), pre- and post-task measures constitute the within-subjects factor (task) rather than baseline and speech. Other selfreport data were analyzed with oneway ANOVAs using the experimental condition as the sole independent variable. Evaluation of assumptions for statistical procedures. Data were evaluated for conformity with assumptions of statistical tests following procedures recommended by Stevens (1992) for the repeated measures ANOVAs and Howell (1987) for the oneway ANOVAs. The procedures specified in this section were followed as appropriate for each analysis, and only violations of assumptions are reported in the analyses that follow. For all analyses, independence of observations was assumed on the basis of the integrity of the experimental design and its controlled implementation. For the repeated measures ANOVAs, univariate rather than multivariate normality was examined due to sample sizes under 20 (Stevens, 1992). For all analyses, normality of distributions of the dependent variables at various levels of each independent variable was —51—  Lenz  Response to Audience Agreement  screened by (a) visually inspecting normal and detrencled normal probability plots and by (b) computing the Shapiro-Wilk statistic. Where deviations from normality were indicated by the Shapiro-Wilk test (that is, when p  <  .0 1), skewness and kurtosis coefficients were  examined in order to specify characteristics of non-normal distributions. Deviations from the assumption of normality, expected impact on consequent inferential statistics, and any steps taken to adjust distributions are reported with each analysis in which they occur. For repeated-measures ANOVAs, the assumption of homogeneity of covariance matrices across levels of the between-subjects factors was tested for each of the dependent variables utilizing the F approximation to the Box test (Stevens, 1992). The equality of cell sizes in the present experiment reduces concern with homogeneity of variance (Tabachnick & Fidell, 1989), but deviations from the assumption and expected impact on results are reported with the analyses in which they occur. Because there are only two levels for the repeated-measures factor in the present design, sphericity and homogeneity of mean square errors are not of concern (Glass & Hopkins, 1984). For oneway ANOVAs, homogeneity of variance was tested by the Levene test. Although equal cell sizes make ANOVAs quite robust to violations of homogeneity of variance (Howell, 1987), positive results of the Levene test and the expected impact on the ANOVA are presented with each analysis where lack of homogeneity is a potential problem. Mukivariate versus univariate analysis. Wherever multivariate analysis appeared a viable alternative to univariate analysis, suitability of the data for multivariate analysis was assessed. First, intercorrelations among the dependent variables were examined. Where  —52—  Lenz  Response to Audience Agreement  logic and correlations suggested a multivariate approach, the suitability of the data for multivariate analysis was assessed by testing homogeneity of the covariance matrices using the F approximation of the Box test. Correlations among dependent variables and other reasons for resulting decisions are presented with each analysis. Control of Type I error. Because univariate treatment of the cardiovascular data produces three separate ANOVAs, it was decided to control the familywise rate for Type I error for cardiovascular data by applying a correction to alpha. Accordingly, alpha was set at .0 18, maintaining an approximate familywise alpha of .05 for the three omnibus ANOVAs and for simple effects analysis on cardiovascular data. For single degree-offreedom post hoc analyses, Tukey’s honestly significant difference was utilized throughout with a familywise alpha of .05. Similar familywise corrections were utilized in the analyses of self-report data with adjusted alpha levels reported in each. While the strategy utilized maintains Type I error for groups of closely related analyses at approximatelyp = .05, the number of separate analyses increases experimentwise error well beyond that level. It was determined that stringent experimentwise error control would result in a tendency to miss potentially meaningftil effects. To preserve information, exact p values are presented whenever available in the analyses that follow. Missing data. Because additional subjects were run when major errors occurred during data collection, there were very few missing data. In most situations, cases with missing data were excluded from related sets of analyses. Handling is reported separately in each analysis where data were missing.  —53—  Lenz  Response to Audience Agreement  RESULTS  Manipulation Check Subjects’ ratings of the supportiveness of the confederates’ behavior were analyzed to compare the intended level of experimentally manipulated support with perceived supportiveness of the confederate. The mean of the last six items from the “Reactions to Speaking” questionnaire were used as the dependent variable for this analysis. In these items, subjects were asked, “How would you rate the student who listened to your speech?” Response was indicated by marking a position on a visual analog scale. The score for items was the distance between the left end of the line and the center of the subject’s mark. Response direction was mixed on the questionnaire, so responses were arithmetically adjusted so a higher number indicated the more supportive (more positive) end of the scale for all items. The six scores were then averaged. Responses ranged from 16.0 to 98.5 with the mean rating across all experimental conditions at 56.39 (SD  21.32). Means and standard deviations of responses for individual items (after  equating response direction) and for the composite scale are presented for each experimental condition in Table 1. A score of 100 is the maximum and indicates uniform reporting of a perfectly warm and supportive listener. Note that each individual item exhibited the same pattern of means as the composite scale: That is, confederates in the Agree condition were rated as most supportive, those in the Disagree were least supportive and Neutral was between the extremes.  —  54—  Lenz  Response to Audience Agreement  Table. 1.  Means and (Standard Deviations) of Ratings of Confederate Supportiveness by Experimental Condition  ITEMS*.. Unsupportive/Supportive  Distant/Close  Rejecting/Accepting  Unfriendly/Friendly  Cold/Warm  Unhelpful/Helpful  Mean of6 items:  N = 30 in all groups.  .  AGREE  NEUTRAL  DISAGREE  84.80  54.67  26.70  (8.29)  (17.22)  (18.67)  72.83  48.63  39.13  (11.77)  (16.71)  (23.20)  79.90  53.87  19.73  (10.75)  (14.13)  (15.81)  83.27  58.83  51.30  (9.82)  (16.32)  (14.61)  78.37  54.43  41.30  (13.47)  (16.13)  (14.10)  75.97  47.53  43.73  (14.66)  (18.55)  (21.38)  79.19  52.99  36.98  (9.66)  (13.92)  (13.00)  Response direction equated for all items  —  55  —  Lenz  Response to Audience Agreement  The rated supportiveness of the confederate was analyzed in a oneway ANOVA with experimental condition as the independent variable, and subjects were found to rate the supportiveness of their listener in accordance with their assigned experimental condition. The ANOVA indicated significant differences between conditions (F(2,87) p  <  =  89.62,  .00 1), and post hoc comparisons using Tukey’s honestly significant differences  indicated that all groups differed (p <.05; see Appendix B-i for the ANOVA table.) The extent of the differences in ratings is apparent in the spreads between the 95% confidence intervals for group means that follow: the Agree condition was 75.6 to 82.8; the Neutral condition was 47.8 to 58.2; the Disagree condition was 32.1 to 41.8.  Other Self-report ofResponse to the Speaking Task To examine whether subjects’ subjective response to the speaking task varied by experimental condition, oneway ANOVAs were conducted on two other indices calculated from responses on the “Reactions to Speaking” questionnaire. These indices are named “Distress,” and “Performance.” Scales were constructed on the basis of face validity and refined by examining interitem correlations. The Distress scale was the mean of eight items with interitem correlations ranging from .31 to .76. (See Appendix B-2 for item descriptions and for intercorrelations among the eight items comprising the scale.) The Performance scale was the mean of two closely related items. One asked “Did you explain your opinions and your feelings clearly?” and  —56—  Lenz  Response to Audience Agreement  the other, “Did you present your arguments and opinions in an interesting manner?” These items were correlated .60 (p  <  .001). A high score indicated better performance.  Results indicated that subjects in the Agree condition reported less distress and rated their own performance as better than subjects in the Disagree condition. Ratings of distress and performance did not differ between Neutral and Disagree. Because the Performance and Distress scales correlated —.60 (p  <  .00 1) and produced homogenous  variance/covariance matrices, the multivariate test was used; it showed differences between the experimental conditions (Wilk’s F(4,172)  =  3.56, p  =  .008). Two univariate  tests (with alpha at .025) were then conducted with Tukey’s honestly significant difference test to locate differences between pairs of cells. For distress, the univariate test was significant (F(2,87)  =  6.88, p  =  .002), and Tukey’s test indicated that distress ratings in the  Agree condition were significantly lower than in the Neutral and Disagree conditions (p  <  .05). Group means and standard deviations follow: AgreeM= 36.70, SD  NeutralM=49.57,SD= 19.21; DisagreeM  15.03;  52.12, SD= 17.27.  The univariate ANOVA for performance also indicated significant differences between conditions (F(2,87)  =  4.36, p  =  .016). The Tukeypost hoc showed that  performance ratings in the Agree and Disagree condition differed (p  <  .05) significantly  while ratings in the Neutral condition did not differ from either. Group means and standard deviations follow: AgreeM= 54.12, SD Disagree M =41.72, SD  =  19.69; NeutralM= 44.48, SD  14.69;  16.46. Multivariate and univariate ANOVA tables are  presented in Appendix B-3.  —57—  Lenz  Response to Audience Agreement  Cardiovascular Reactivity to the Task Umvariate versus multivariate analysis. Correlations between the three cardiovascular measures were examined to determine whether data were more suitable for univariate or multivariate analysis. Correlations among SBP, DBP, and HR were calculated at baseline and speech. At baseline, SBP and DBP were moderately correlated, and HR was correlated slightly with DBP but uncorrelated with SBP. During the speech task, however, the pattern of correlations was different: Systolic and diastolic blood pressure were more highly correlated than at baseline (t(87)  =  3.67, p  <  .01) and heart rate was uncorrelated  with either SBP or DBP. (For comparisons between correlations, the t-test for differences between non-independent correlations was calculated in accordance with Williams. revision to Hotelling’s procedure (Howell, 1987). Since the correlation between SBP at baseline and speech was higher than that of DBP (.63 versus .43, respectively), the SBP intercorrelation was used as the correction factor.) Correlations among CV measures at baseline and speech are presented in Table 2. Although absolute levels of SBP and DBP are moderately correlated, hypotheses of the present investigation concern task-related changes in CV indices, and there is evidence that changes in CV indices are substantially less correlated than absolute levels. Lamensdorf and Linden (1992) examined the effects of stress tasks on CV measures and found that when task-related changes are adjusted for baseline levels, the highest correlations between changes in CV measures are less than r = .40. This would indicate that no more than 20% of the variance of task-related change is shared between any two CV measures. On a more theoretical level, recent work utilizing autonomic blockade  —  58  —  Lenz  Response to Audience Agreement  techniques has identified independent pathways of neural activation and modulation that begin to explain the frequently observed desynchrony between stress related changes in blood pressure and heart rate (Berntson et al., 1994; Cacioppo et al., 1994). Although the relationships between cardiovascular measures are far from understood at this time, it is clear that cardiovascular arousal cannot be treated as a unitary phenomenon. For these reasons, and because generally low correlations and inconsistent patterns of correlations across measures and across time were observed in the present data, it was determined that the data were better suited for univariate analyses than for multivariate. This decision is consistent with typical handling of cardiovascular measures in the cardiovascular psychophysiology literature.  Table  2. Correlations among Cardiovascular Measures BASELINE Systolic  Diastolic .42 (p=.000)  Diastolic  Heart Rate .13 ) 23 (p=. .34 ) 1 (p=.OO  DURING SPEECH Systolic  Diastolic  Heart Rate  .67  .15  ) 000 (p=. Diastolic  ) 7 (p=.l  .03 (p=. 80)  Pearson product-moment correlations. N 90 in all cases.  —59—  Lenz  Response to Audience Agreement  CV response to the experimental tasks. A total of 14 measurements were made of each of the three CV indices. Because hypotheses of the present study involve only taskrelated .change from baseline, the analyses that follow include only five of the measures— two at the end of the baseline and three during the speech. Measures were also taken when the subject first began filling in forms, during the speech preparations period, and during the recovery period following the speech. The pattern of cardiovascular response observed during the experiment was similar to that observed during most laboratory stress procedures. Figure 1 shows SBP for men and women at each of the fourteen time points.  Figure 1. Systolic Blood Pressure at all Measurement Points by Sex 150  :  145 140 135  125 • 120 C 115 ; 110 C 105 100 Al A2 Bi B2 B3 P1  Sl  P2 P3  S2  S3 Ri R2 R3  Measurement Points Measurement points: A  =  Adaptation period, B  =  Baseline, P  =  Speech preparation period,  S  =  During the speech, R = Recovery period. Measurements initiated as follows:  1  =  minute 0:30, 2  =  minute 2:30, 3  =  minute 4:00.  —  60  —  Lenz  Response to Audience Agreement  Note that SBP levels were lowest in the baseline and recovery periods, highest during the speech, and between these extremes during the preparation period. Note also the frequently observed adaptation response within each time period: SBP was highest at the beginning of each new time segment and fell as subjects adapted to the new task or situation. Although responses varied somewhat for DBP and HR. these measures exhibited patterns similar to SBP with respect to relative elevation of the time segments and adaptation within each time-period. Systolic blood pressure. Systolic blood pressure was found responsive to the speech task but unaffected by the supportiveness of the experimental confederate. The predicted interaction between the repeated-measures factor, task, and experimental condition failed to reach significance (F(2,84)  =  1.93, p  =  .  15). As generally reported, men had higher  systolic blood pressure than women. The main effect for sex (F(1,84)  =  37.40, p  <  .00 1)  indicated that when subjects’ systolic blood pressure was pooled across experimental condition and across baseline and speech measures, men’s systolic blood pressure was 10.12 mmHg higher than women’s (MenM= 133.56, SD = 7.66; WomenM= 123.44, SD  =  7.92). Means and standard deviations of SBP for both sexes in each condition are  presented in Table 3. The significant main effect for the within-subjects factor task (F(1,84)  =  681.60, p  <  .00 1), was due to a rise in systolic blood pressure from baseline to  speech task of over 24 mmHg (Baseline M = 116.34, SD M= 140.65, SD  =  =  9.94; During speech  10.58). (The ANOVA table is presented in Appendix B-4.)  —61--  Lenz  Response to Audience Agreement  Table 3.  Means and (Standard Deviations) of Systolic Blood Pressure by Experimental Condition and Sex.  Condition Sex Agree  Neutral  1 BASEIJNE  2 SPEECH  CHANGE  Men  122.50  (8.88)  143.40  (10.46)  Women  113.23  (9.23)  136.91  (8.84)  23.68  (6.49)  Men  120.97  (9.83)  144.73  (7.50)  23.77  (6.90)  Women  109.07  (7.77)  133.11  (11.30)  24.04  (11.23)  121.73  (7.15)  148.00  (10.11)  26.27  (9.95)  110.57  (7.61)  137.73  (8.33)  27.17  (6.11)  Disaree Men Women  Mean of the last two measures in the rest period. during the speech task. 3 N= 15 in all cases.  2  20.90 (10.76)  Mean of the three measures taken  Change scores may vary from cross-  table computations due to rounding.  Diastolic blood pressure. Diastolic blood pressure also increased during the speech task and—like SBP—was unaffected by the supportiveness of the experimental confederate. Men’s DBP increased slightly but significantly more in response to the speech task than women’s. The ANOVA for DBP indicated significant main effects for sex and task. In addition, there was a significant interaction between sex and task. The predicted interaction between task and experimental condition was not significant (F(2,84) p  =  1.06,  .35). As with SBP, diastolic increase from baseline to speech was large (over 23  —  62  —  Lenz  Response to Audience Agreement  mm/Hg) and produced a significant main effect for task (F(I,84) BaselineM= 68.60, SD  6.86; During speechM= 92.36, SD  =  =  853,54, p  <  .001;  7.90). Means and  standard deviations of DBP for both sexes in all experimental conditions are presented in Table 4. Overall differences between men’s and women’s diastolic blood pressure were small and not statistically significant (F(1,84)  =  5.02, p  =  .028). The presence of a  marginally significant interaction between sex and task (F(1 ,84)  =  5.78, p  <  .018) indicated  that men and women may have responded somewhat differently to the speech task; or, more specifically, differences between men’s and women’s diastolic blood pressure were greater during the speech than at baseline.  Table 4.  Means and (Standard Deviations) of Diastolic Blood Pressure by Experimental Condition and Sex.  Condition Sex Agree  Neutral  1 BASELiNE  2 SPEECH  CHANGE  Men  70.70  (6.07)  96.58  (9.07)  25.88k  (6.51j  Women  69.53  (6.40)  93.04  (7.06)  23.51  (7.05)  Men  68.23  (9.17)  93.93  (7.46)  25.70  (6.98)  Women  68.03  (6.54)  86.51  (6.92)  18.48  (9.68)  68.23  (6.40)  93.80  (7.05)  25.57  (7.55)  66.87  (6.63)  90.29  (6.74)  23.42  (8.11)  Disagree Men Women  Mean of the last two measures in the rest period. 2 Mean of the three measures taken  during the speech task. 3 N = 15 in all cases. Change scores may vary from crosstable computations due to rounding.  —  63  —  Lenz  Response to Audience Agreement  A simple effects analysis was conducted to determine whether men and women exhibited different DBP levels at baseline and during the speech task. At baseline, DBP was not different by sex (F(1,88)  =  0.39, p  .53). In fact at baseline, DBP for men was  only 0.92 mmHg higher than that for women (MenM= 69.06, SD M= 68.14, SD  =  =  7.28; Women  6.47). During the speech task, however, men’s DBP was 4.82 mmHg  higher than women’s (MenM= 94.77, SD  =  7.83; WomenM= 89.95, SD  =  7.27), and  although this difference is not of practical importance, it was statistically significant (F(1,88)  9.16, p  <  .003). This interaction is presented in Figure 2. Visual inspection  indicates that the magnitude of task-related change in DBP is far greater than sex differences. (ANOVA tables and simple effects analysis are presented in Appendix B-5.)  Figure 2. Diastolic Blood Pressure at Baseline and Speech by Sex. 95 90  85  80 2 75 -Men  —  C  —0—  Baseline  Experimental Task  Wonn  Speech  Heart rate. Examination of the distribution of the HR data produced a significant Shapiro-Wilk test (p  =  .049) for women’s HR during the speech task; as a result,  —  64—  Lenz  Response to Audience Agreement  skewness, kurtosis, and stem and leaf plots were examined. Departure from normality was due to positive skewness with no kurtosis evident. Because the repeated-measures ANOVA is generally robust to skewed distributions and the non-normality was found in only in one level of one of the independent variables, impact on power and error rate was determined to be negligible (Stevens, 1992). No adjustment was made to the data. Heart rate increased during the speech task in all conditions but increased less when confederates agreed with subjects than when they remained neutral or disagreed. There was no difference between the Neutral and the Disagree conditions. The ANOVA for FIR produced main effects for sex and task and an interaction between task and experimental condition. Means and standard deviations of HR for both sexes are presented in Table 5.  Table 5.  Means and (Standard Deviations) of Heart Rate by Experimental Condition and Sex.  Condition Sex Agree  Neutral  BASELINE’  2 SPEECH  CHANGE  Men  68.17  (7.82)  81.36  (11.98)  13.19k  (5.87)  Women  72.57  (12.55)  89.60  (14.40)  17.03  (8.76)  Men  71.57  (12.64)  92.67  (16.73)  21.10  (16.29)  Women  77.03  (15.19)  94.91  (17.77)  17.88  (14.16)  64.67  (10.15)  88.78  (13.71)  24.11  (12.75)  71.70  (10.65)  105.53  (14.73)  33.83  (17.14)  Disagree Men Women  ‘Mean of the last two measures in the rest period. during the speech task. 3 N = 15 in all cases. table computations due to rounding.  —65—  2  Mean of the three measures taken  Change scores may vary from cross-  Lenz  Response to Audience Agreement  Averaging across baseline and speech task, women’s HR were 7.35 beats per minute higher than men’s (F(1,84)  =  8.80,p < .004; MenM= 77.87, SD  minute faster than at baseline (BaselineM= 70.95, SD  p  <  =  12.73; Women  12.73). Heart rates during the speech were on average 21 beats per  M= 85.22, SD  SD  =  =  12.03; SpeechM  92.14,  16.33). Although this difference produced a significant main effect (F(1,84)  =  234.63,  .00 1), the presence of an interaction between task and experimental condition  indicated that the amount of increase in heart rate from rest to speech task varied across groups (F(1,84)  =  8.74, p  <  .001). Figure 3 illustrates this differences by showing mean HR  for each experimental condition at baseline and during the speech task.  Figure 3. Heart Rate at Baseline and Speech by Experimental Condition 100 95  -  -  8580751  70 65  Speech  Baseline Experimental Task  A simple effects analysis was conducted to determine if the HR varied by experimental condition at baseline and during the speech. These analyses indicated that no group differences existed at baseline (F(2,87)  =  —  2.04, p  66  —  =  .13.6) but that the groups differed  Lenz  Response to Audience Agreement  significantly during the speech (F(2,87)  =  4.37, p  =  .0 16). Tukeys procedure was used to  compare means of experimental conditions during the speech. Observed mean differences were compared to a critical difference of 5.69 (p  <  .05). Mean HR during the speech was  found to be lower for subjects in the Agree condition than for subjects in the other two conditions; Neutral and Disagree did not differ significantly from each other. (The ANOVA table and simple effects analysis are presented in Appendix B-6.)  Pre- and Post-task Subjective Measures Analyse of pre- and post-task subjective measures followed the same design as the  analysis of CV measures: measurement at two time points (pre- and post-task) constituted the repeated-measures factor task, and the two between-subjects factors were sex and experimental condition. Affect Grid. The Affect Grid produces separate scores for arousal and pleasantness. Correlations between the two scores were computed at both pre- and post-task to assess suitability of the data for multivariate analysis. In keeping with reports by the developers of the measure (Russell et a!., 1989), the two scores were uncorrelated. At pre-task, correlation between the scores was r = —.02 (p  r = —.12 (p  =  =  .87), and the post-task correlation was  .27). Because of lack of correlation and because neither arousal nor  pleasantness was substantially correlated with self-esteem subscales, the Affect Grid scores were analyzed in two separate univariate ANOVAs. Intercorrelations among all pre- and post-task subjective measures are presented in Table 6.  —  67  —  Lenz  Response to Audience Agreement  Table 6.  Correlations among Self-report Measures at Pre-task  Measure  Subscale  Affect Grid  1. Pleasantness  2  3  —.02  2. Arousal  SSES  4  5  .09  .29**  25*  .01  .13  .11  54***  .50***  3. Appearance 4. Performance  5. Social Pearson product-moment correlations. N = 89 for all correlations. * p  <  .05,  **  p<.o’, ***p<.OOl  Table 7.  Correlations among Self-report Measures at Post-task  Measure  Subscale  Affect Grid  1. Pleasantness  2  3 .22*  —.12  2. Arousal  SSES  —.02  4  5  .19  .22*  —.06  —.14 •55***  .65***  3. Appearance  74***  4. Performance  5. Social Pearson product-moment correlations. N = 89 for all correlations. p  <  .001  —  68  —  *  p  <  .05,  Lenz  Response to Audience Agreement  Subjects reported more arousal after the speech task than before, but the experimental condition did not affect level of reported arousal. Pre- to post-task change was indicated by the main effect for task (F(1,84)  =  105.71; p  <  .001; see Appendix B-7 for the ANOVA  table). Means and standard deviations follow: pre-taskM= 5.29, SD M = 7.11, SD  =  1.49; post-task  1.32. Lack of effect of experimental condition on response was evidenced  by the non-significant task by condition interaction (F(2,84)  =  0.69, p  =  .51).  Subjects also reported less pleasant affect after the speech task than before, and this tendency was more pronounced in the Neutral and Disagree conditions than in the Agree condition. There was a significant main effect for the repeated-measures factor task (F(1 ,84)  =  10.75; p  =  .002); ratings of pleasantness of affect were higher prior to the task  (Pre-taskM= 5.60, SD  =  1.53; Post-taskM= 4.79, SD  experimental condition was also significant (F(2,84)  =  =  2.03). The main effect for  6.25; p  =  .003). Although the  interaction between task and condition was not significant (F(2,84)  =  6.67, p  =  .095), it  suggested a trend that affected decisions regarding post hoc analysis. Means and standard deviations of the pleasantness scores for each of the conditions are presented in Table 8. Note the differences between groups in magnitude of change. In order to examine whether differences between conditions were evident at both pre and post-task, a simple effects analysis was conducted. Results indicated that mean ratings of pleasantness differed by experimental condition at post-task (F(2,87) but not at pre-task (F(2,87)  =  1.13; p  =  =  6.82 ; p  =  .002)  .3 3). The tables for the omnibus ANOVA and  simple effects analysis are presented in Appendix B-8. Tukey s procedure was used to t compare mean ratings between conditions at post-task; observed mean differences were  —  69  —  Lenz  Response to Audience Agreement  compared to a critical difference of 1.12 (p  <  .05). Mean reported pleasantness at post-  found to be lower for subjects in the Disagree and Neutral conditions than for  task  those in the Agree condition; reports for Disagree and Neutral did not differ significantly. These results and group means are presented in Table 8.  Table 8.  Means and (Standard Deviations) of Pleasantness Rating on the Affect Grid by Experimental Condition  Pre-task  Post-task  Change  AGREE  NEUTRAL  DISAGREE  5.93  5.37  5.50  (1.68)  (1.33)  (1.55)  5.80a  b 453  (2.02)  (1.85)  (1.85)  —0,13  —0.84  —1.47  (2.49)  (2.29)  (2.21)  Means in the same row with different superscripts differ, p < .05.  State Self-Esteem Scale. The State Self-Esteem Scale (SSES) yields a total score and subscale scores for appearance esteem, social esteem, and performance esteem. Because analysis of the total scores could obscure potentially meaningful differences in patterns of subscale scores, subscale scores were analyzed. Moderate to high intercorrelations between subscales of the SSES suggested that multivariate analysis was more appropriate  —70—  Lenz  Response to Audience Agreement  than univariate. As shown in Tables 6 and 7, correlations between SSES subscales at pre task ranged from r = .50 to .61, and at post-task correlations were between .55 and .74 (for all correlations, p  <  .001).  Tests of assumptions for the MANOVA indicated two problems with scores from the performance subscale: The scores were not normally distributed in three cells, and covariance matrices for the planned MANOVA were non-homogenous (Box M approximate F = 1.33, p  =  .01). It was decided to analyze subscale scores separately in a  univariate model using an alpha with a familywise correction (.018) for all three tests. Because one SSES questionnaire was missing, N = 89 for all analyses of SSES data. When analyzed separately, subscale scores for SSES met criteria for homogeneity of variance/covariance matrices (i.e., for Box M approximate F, p> .05). For the performance subscale only, there were indications of non-normal distribution of scores. The screen for normality of distributions using the Shapiro-Wilk test indicated potential violations of the assumption for three groups (p  <  .0 1). Stem and leaf plots and skewness  and kurtosis coefficients were examined to specify types of deviations from normality. For men in the disagree condition at pre-task, performance esteem scores were distributed bimodally, but skewness and kurtosis were within normal limits. Significant skewness or kurtosis was found in only two groups. Men’s post-task performance esteem scores in the neutral condition were found to be negatively skewed (skewness = —1.57, p  <  .01), and  although the distribution was leptokurtotic, this characteristic was not significant (kurtosis  =  2.21, p> .0 1). Women’s post-task performance scores in the neutral condition  were found to be significantly negatively skewed and leptokurtotic (skewness  —71—  —2.56,  Lenz  p  <  Response to Audience Agreement  .01, kurtosis = 8.58,p  <  .01). Following the rationale of Stevens (1992), it was  decided not transform the distributions. The negative skewness was expected to have little impact on power in the ANOVA. Although kurtosis may have more potential impact on power, it occurred in only one group, and leptokurtosis is in general less problematic than platykurtosis (Stevens, 1992). Although impact of non-normal distributions on power in an ANOVA can be difficult to predict, the anomalies in distributions of SSES performance subscale were expected to result in slightly increased power in the analysis. The subsequent univariate analyses indicated no significant effects for any of the SSES subscales. (See Appendices B-9, -10, and -11 for complete ANOVA tables.) Task by condition interactions were non-significant for all three scales. Statistics for the task by condition interactions are as follows: Social esteem F(2,83)  =  0.80, p  =  .45  Appearance esteem  F(2,83)  =  2.06, p  =  .14  Performance esteem  F(2,83)  =  2.37, p  =  .10.  Because the power of the ANOVA may have been affected by lack of normal distribution of the performance scores, means and standard deviations of the SSES performance scores for each of the conditions are presented in Table 9. Visual examination shows that pre- to post-task changes in mean ratings vary only slightly across experimental conditions and that mean changes from pre- to post-task are small relative to pre- and post-task standard deviations.  —  72  —  Lenz  Table 9.  Response to Audience Agreement  Means and (Standard Deviations) of Performance Rating on the SSES by Experimental Condition AGREE Pre-task  Post-task  Change  No means differ,p  <  NEUTRAL  DISAGREE  27.80  27.20  26.47  (3.60)  (5.22)  (5.35)  28.24  27.20  25.63  (3.81)  (5.28)  (6.20)  +0.44  0.00  —0.84  (2.06)  (2.99)  (2.56)  .05.  Correlates ofSelf—reported Social Support The final set of analyses was exploratory in nature. It involved examining existing social support as measured by the Interpersonal Support Evaluation List (ISEL) and its relationship with response to the experimental procedure. The general question guiding this exploration was whether report of pre-existing social support had any discernible impact on subjects’ response to the support provided during their speech. For this exploration, change scores were computed for the CV and self-report indices. (See Appendix B- 12 for CV change scores and standard deviations by experimental condition.) Fisrt the ISEL scores themeselves were examined. The distribution of ISEL scores was normal with a tendency towards a slight negative skew. Women reported higher levels  —  73  —  Lenz  Response to Audience Agreement  of social support than men. Women’s mean ratings on the ISEL were 36.22 (SD and men’s were 32.20 (SD p  =  =  =  6.5); these scores were significantly different (t(62)  =  298) -2.84,  .006, df adjusted for non-homogeneity of variance). There was a wider variation in  men’s ISEL scores than in women’s.  -  Correlations between CV change scores and the ISEL were then calculated. Because subjects rated the supportiveness of the confederates differently in each experimental condition, it was necessary to examine differences between correlations in the experimental conditions. While this approach resulted in small number of subjects contributing to each correlation (n  =  30) and in an increased number of correlations, it was  determined to be necessary to avoid overlooking potentially meaningful relationships. More specifically, it was of interest whether existing social support correlated differently with response to the speech task in the Agree condition (when support was present) than in the other two conditions (when there was no support). Although there were numerous other potential effects, it could be anticipated that someone used to high levels of support may have decreased arousal in the somewhat familiar Agree condition and increased arousal in the unfamiliar non-supportive conditions. Correlations between the ISEL and changes in CV and self-report measures are presented in Table 10. There is only one significant correlation, and that is between the ISEL and social self-esteem in the Agree condition (r  .36, p  =  .05). This is a relatively  small correlation and is marginally significant even without error correction. With 24 correlations, chances are above 1.00 that one would be found significant at p  —74—  =  .05.  Lenz  Response to Audience Agreement  Table 10. Correlations between ISEL and Change in Cardiovascular and Selfreport Measures by Experimental Condition AGREE  NEUTRAL  DISAGREE  SBP  —.06  —.01  .15  DBP  —.18  .03  —.18  HR  .23  .12  .20  AFFECT GRID-Arousal  -.18  .07  -.13  -Pleasantness  —.19  .26  —.01  .09’  —.14  —.04  .3612  —.01  .04  .23’  —.14  —.09  SSES-Performance -Social -Appearance  Pearson product-moment correlations. n  =  P = .05; for all other correlations, p> .05.  —75—  29; for all other correlations, n  =  30.  2  Lenz  Response to Audience Agreement  DISCUSSION  Summary ofFindings Ninety undergraduate students gave a speech on a topic they cared deeply about. Subjects were randomly assigned to receive one of three types of feedback from a laboratory confederate: (a) agreement with the. sujject, (b) non-responsive listening, or (c) disagreement. Blood pressure and heart rate were monitored before and during the speech task, and measures of affective state and self-esteem were taken before and after the speech. Subjective response to the speech task was measured post-task. Findings indicated that subjects whose audience agreed with them had significantly less increase in heart rate during the speech task than subjects whose audiences disagreed or remained non-responsive. No condition-related effects were observed on measures of blood pressure, but men in all conditions were found slightly more responsive to the speech task than women. A number of self-report measures indicated that subjects found audience disagreement to be more aversive than agreement. Reaction to the non— responsive audience varied by measure but it was generally more aversive than agreement and less aversive than disagreement.  Internal Validity Because the hypotheses of the present study postulate a stress buffering mechanism for social support, evaluation of the adequacy of the experimental procedure must address two related manipulations, both of which are necessary to the integrity of the experiment. First, the stress task itself—the speech task—must be shown adequately evocative of  —  76  —  Lenz  Response to Audience Agreement  stress responses in subjects. Secondly, the behavior of the confederates must vary in a manner that leads to different levels of perceived supportiveness between experimental conditions. The relevant manipulation checks converge to indicate that the experimental procedures clearly accomplished both of these goals. Response to the task. First, the speech task evoked a strong and consistent stress response. The repeated-measures analyses showed significant task-related changes in SBP, DBP, HR, and in subjective report of increased arousal and decreased pleasantness of affect. The consistency and the magnitudes of these effects across measurement modality argues for the adequate provocation of stress response. The effect sizes (using baseline or pre-task SD) for task-related changes for the pooled data for all subjects follow: SBP d= 2.45, DBP d= 3.46, HRd= 1.76, Affect Grid Arousald= 1.22, Affect Grid Pleasantness d = 0.53. (CV effect sizes for Agree and Neutral are compared to effect sizes in similar experiments below.) There is little doubt that consistent task-related changes of these magnitudes demonstrate stress induction at a level adequate to detect experimentally related attenuation if it is present. Ratings of support. Secondly, there are strong indications that the experimentally provided support was both credible and socially appropriate. As for credibility of the manipulation, only two subjects (both excluded from analysis) reported suspecting that the confederates were assigned a role in the study. Most subjects responded with genuine surprise when told of the experimental deception. They generally mentioned the drawing of lots and asked how they could have been preassigned as speaker when they drew their role themselves “at random.” It appears that the process of assigning roles was particularly  —  77  —  Lenz  Response to Audience Agreement  convincing to subjects. Finally, subjects rated the supportiveness of the confederates differently in each of the three experimental conditions, and their ratings were in keeping with the intended levels of support. This indicator was compiled from six individual items, all of which were related semantically to supportiveness and all of which produced the same ordering of mean ratings between conditions: Agree was most supportive, Disagree was least, and Neutral was between the other two. This pattern of findings indicates that subjects in different experimental conditions perceived confederates’ behavior in accordance with the intent of the experimental design and that they believed these differences to have occurred naturally.  Evaluation ofSpecfic Hypotheses Conclusions relating to each area of investigation in the present study are presented separately below. Discussion of each hypothesis includes (a) a synopsis of relevant findings, (b) analysis of the strength and reliability of the findings, (c) generalization of findings, and (d) comparisons with findings in closely related studies. The effects of neutral controls. The primary question addressed in the present study was whether the use of a neutral or non-responsive control constitutes an adequate comparison group in studies of response to experimentally provided support. The issue was defined as follows: The ambiguity of a non-communicative social interaction could conceivably increase stressfulness of an interaction and thereby inflate the relative protective effect of a supportive interaction. This question was addressed by including a Disagreement condition in the present study—a condition intended to be both non  —  78  —  Lenz  Response to Audience Agreement  supportive and unambiguous. Direct comparison of the effects of audience disagreement with both supportive and non-responsive interactions allowed assessment of the relative contributions of supportiveness and clarity of communication. The data show clearly that non-responsive social interaction does not increase stressrelated reactivity more than unambiguous disagreement. For the variables on which differences in reactivity between experimental conditions were apparent, audience agreement was associated with less stress response than disagreement. Responses to the more ambiguous, non-communicative audience were mixed. Non-communicative and disagreeing audiences were found to have similar effects on HR, and both were associated with greater HR response than an agreeing audience. Findings were slightly different for subjective report of pleasantness of affect. Subjects reported more pleasant affect when their audiences agreed with their opinions than when the audience disagreed, but affect in the face of a non-communicative audience was not different from either agreeing or disagreeing. Contrary to hypotheses based on postulated ambiguity in non-communicative social situations, the non-responsive audience evoked equivalent or less stress response than unambiguous disagreement. Findings show clearly that objections to the use of neutral or non-communicative audiences as controls by Germ et al. (1992) and by Lepore et al. (1993) were unfounded. Sex differences in cardiovascular reactivity. Previous studies have frequently used exclusively female subjects (e.g., Allen et al., 1991; Germ, Pieper, Levy, et al., 1992; Kamarck et al., 1990). When both men and women have been included as subjects, ambiguous patterns of sex-related response have been observed (e.g., Lepore et al., 1993;  —  79  —  Lenz  Response to Audience Agreement  Sheffield & Carroll, 1995). Consequently, the present study directly examined sex differences in response to experimentally provided social support from a same-sex confederate. Consistent with numerous other reports (e.g., Cohen, S. et al., 1985; Linden et al., 1993; Sarason, Levine, Basham, & Sarason, 1983), women reported significantly more support in their daily lives than men. However, men and women did not respond differently to the manipulation of social support in the present experiment. There were no significant sex by condition interactions for any of the dependent variables in the study, and the only sex differences observed were independent of the manipulation of support. Although the data in the present study fail to indicate any sex differences in response to experimentally provided support, the pattern of findings cautions against drawing definitive conclusions. Conclusions regarding patterns of response on SBP and DBP are unwarranted. Without clear evidence that responsivity is attenuated in the presence of a supportive audience, sex differences cannot be observed. For FIR, caution in generalization is recommended on the basis of a different argument—on the general difficulty in interpreting the failure to reject a null hypothesis. Although there was a significant support-related attenuation of HR reactivity and no significant interaction of this effect with sex, the pattern of findings indicates a weak trend towards sex differences. This trend is most apparent in the comparison of HR changes (baseline to task) for men and women in the three experimental conditions as presented in Figure 4. Although the interaction represented by this graph fell far short of statistical significance (p  =  17), the  data hint that men and women may respond differently to a non-communicative audience. The change scores indicate a tendency for women to respond to lack of communication as  —  80—  Lenz  Response to Audience Agreement  if it means agreement whereas men may respond to it more like they respond to disagreement. Such an effect could be of interest in defining sex differences in response to social cues, and the present findings provide no basis to predict whether such a tendency may be found significant in research designed specifically for investigating such an effect.  Figure 4. Change in Heart Rate by Sex and Experimental Condition 35  Agree  • Neutral D Disagree  E 30 25  20  15 10  -FMen  Women Sex  Sex differences unrelated to experimentally provided support were more clear and replicated earlier findings in a number of studies. Men exhibited higher SBP and lower HR than women at baseline and during the speech task, and men showed a slightly (but significantly) greater response to the speech task than women on DBP. The sex differences in SBP and HR are frequently observed in both epidemiological studies and in laboratory stress research and are unrelated to the hypotheses of the present study. The greater response of men to the stress provocation was also reported by Lepore et al.  —81  —  Lenz  Response to Audience Agreement  (1993) who found men’s reactivity to be higher than women’s on both SBP and DBP. Similarly, Sheffield and Carroll (1995) found men more reactive than women on SBP and a trend towards more reactivity on FIR. Self-report measures. One of the goals of the present study was to reduce the commonly observed desynchrony between CV response and self-report. This goal was addressed in two ways. First, the Affect Grid was included as a measure of affective state because it is simple to administer, highly sensitive, and psychometrically sound. Secondly, items previously found to show differences between supported and unsupported subjects were refined and presented in a uniform visual analog format. The arousal sub scale of the Affect Grid indicated large increases in arousal during the speech task, but arousal did not differ by experimental condition. Three other self-report measures were found to distinguish between experimental conditions: (a) Affect Grid Pleasantness, (b) a composite scale of distress during the speech, and (c) a two-item scale indicating self-assessment of performance level during the speech. In keeping with expectations, subjects whose audiences disagreed with their positions reported more distress, less pleasant affect, and lower performance than subjects whose audiences agreed with them. For those subjects whose audiences were non-communicative, findings varied. Performance ratings for these subjects were not significantly different from either of the other two groups, but reports of distress and pleasantness of affect were much like those whose audiences disagreed with them—that is, Neutral subjects reported more distress and less pleasant affect than Agree subjects.  —  82  —  Lenz  Response to Audience Agreement  Because findings on several self-report scales paralleled heart rate response during the experiment, the general goal of improving synchrony between self-report and physiological measures was accomplished. These findings suggest that desynchrony between CV response and self-report has been to a large extent due to measurement strategies. One result is of particular interest for the development of self-report measures in cardiovascular stress research. The distress scale developed for this study from Lepore et al’s (1993) items resulted in the same findings as the Pleasantness scale of the Affect Grid. Indeed, pleasantness and distress appear to be polar opposites and probably tap a similar underlying construct. In contrast, the Arousal scale of the Affect Grid did not distinguish between experimental conditions. These results suggest that cardiovascular stress researchers may miss potentially meaningfiil effects if they query subjects only about arousal. As successful as the self-report measures were in the present experiment, they are not without limitations. First, support-related attenuation of CV response was observed only on HR and not on blood pressure. Because CV measures were desynchronous with each other, it is impossible to show complete synchrony between CV measures and self-report. For this reason, interpretation of self-report data has focused only on its relations to HR response. Secondly, the “Response to Speaking” questionnaire included several items that were discarded due to low correlations with other items originally conceived to be similar. It is uncertain whether responses would change if these items were excluded from future versions of the questionnaire.  —  83  —  Lenz  Response to Audience Agreement  Self-esteem as a moderating variable. One of the hypotheses of the present study was that self-esteem may moderate cardiovascular response to social support. Because one of the major benefits of social support has been reported on a theoretical basis to be the enhancement of self-esteem (Cohen, S., Evans, Stokols, & Krantz, 1986; Cohen, S. & Hoberman, 1983; Cohen, S. & Wills, 1985; Wills, 1985), it followed that changes in selfesteem may accompany laboratory stress tasks and may vary according to the provision of support. Further, it was of interest whether changes in self-esteem may correspond to indices of cardiovascular response. Tests of these hypotheses uniformly indicated that the experimental procedure had no significant effect on self-esteem as measured by the subscales of the State Self Esteem Scale. As with the data on sex differences, interpretation of the self-esteem findings is complicated by the lack of a strong cardiovascular effect due to social support. To the extent that changes in self-esteem are correlated with cardiovascular responsiveness, self-esteem changes may only be observable in studies where stronger effects on CV indices are also present. A couple of comments may be in order with regards to performance esteem on which a weak trend towards the hypothesized effect was observed. Subjects with an agreeing confederate showed slight (but not statistically significant) increase in performance esteem and those with disagreeing confederates showed a slight decrease. Subjects with a non communicative audience reported exactly the same levels of performance esteem before and after their speeches. The magnitude of these pre- to post-task changes was quite small (d = 0.15 for the Disagree condition). However, the inter-test interval in the present study  was no longer than 20 minutes, and the stressors themselves lasted no longer than 10  —  84—  Lenz  Response to Audience Agreement  minutes including both speech preparation and speech delivery. Heatherton and Polivy (1991) reported relatively small changes in performance esteem as significant both statistically and practically. In their study, an effect size of d  =  0.26 was observed on  performance esteem over a one-week period after students were warned about a particularly difficult midterm exam or after receiving midterm grades. Comparison with Heatherton and Polivy’s study—with its longer time frame and more personally relevant stressors—suggests that the lack of significant findings for performance self-esteem in the present study could be due to the brevity of the stress induction or to the lack of strong personal investment in the outcome of the speech. Subjects in different conditions in the present study reported statistically significant differences in their assessment of their own performance levels, but not in their performance esteem. Effects of environmental social support. The relationship between self-report of ongoing social support and response to the experimental procedures was explored by examining correlations between ISEL scores and task-related changes in CV measures, affect, and self-esteem. There was no indication that report of pre-experimental support levels was related to response during the experiment, Because the examination was exploratory, the findings are less than conclusive but they have methodological implications. Several investigators have found report of higher levels of environmental social support to be associated with lower CV levels when ambulatory measurement strategies are used (Linden et al., 1993; Spitzer, Llabre, Ironson, Geliman, & Schneiderman, 1992; Unden, Orth-Gomer, & Elofsson, 1991). If social support attenuates CV response in the workplace, at home, and at university (respectively) and is not related  —  85  —  Lenz  Response to Audience Agreement  to CV response in the laboratory, one must question the relationship between laboratory stressr and real-life stressors and the relation of experimentally provided support to that provided in subjects’ everyday lives. Sheffield and Carroll (1994, 1995) make a similar point and conclude that social support research must be conducted in settings that are more realistic than the laboratory.  Cardiovascular Response to Support Subjects in the Agree condition showed less increase in heart rate than those in the Neutral and Disagree conditions. The response difference for conditions on systolic blood pressure was not statistically significant but the ordering of group means was the same as that for HR—that is, mean response was greatest in the Disagree condition and least in the Agree condition. For diastolic blood pressure, there were no indications of support-related attenuation of CV response. Comparison of findings with earlier studies. To date, there are four published studies that investigate the effects of experimentally provided support on cardiovascular response to a social interaction stress task. The general pattern of findings in the present study provides partial support of the protective effects of social support observed by Germ et al. (1992) and by Lepore et al. (1993). While both earlier studies found experimentally provided support to attenuate SBP and DBP response to a social stress task and Genn et al. found it to attenuate HR response as well, the present data demonstrate a significant effect only on HR with a weak trend appearing in the SBP data. Sheffield and Carroll (1995) also found HR to be the only CV measure attenuated by the presence of a  —86--  Lenz  Response to Audience Agreement  supportive confederate, but it must be noted that in that study, HR response in the disagreement condition (which they called ‘undermining”) was significantly different from the alone condition and did not differ from the Agree condition. Effect sizes. In the interest of examining the major findings of the four closely related studies in a quantitative fashion, estimated effect sizes for cardiovascular response in each of the four studies are presented in Table  11.  Table 11. Estimates of Task- and Support-Related Effect Sizes for Cardiovascular Measures in Studies Using Social Interaction Tasks. Without Support’ STUDY Present study  With Support 2  3 Difference  SBP  DBP  HR  SBP  DBP  HR  SBP  DBP  HR  2.90  3.80  2.67  2.21  4.01  1.44  0.69  —0.21  1.23  1.73  1.53  1.43  0.79  0.81  0.48  0.94  0.72  0.95  2.75  3.14  1.50  2.00  1.25  1.14  1.39  1.11  1.05  0.89  0.34  0.22  Germ et aL, 1992  Leporeetal., 1993 —  —  Sheffield and Carroll, 1995  0.78  0.74  0.04  ‘Mean of largest task-related change in each study divided by baseline SD. 2 Mean task-related change for groups with support divided by baseline SD. 3 Difference between responses to the task in supportive and non-responsive conditions. 4 Effect sizes based on estimated baseline standard deviations. For the purposes of these estimates, the effect size for support-related response attenuation is taken to be the difference between the task-related effect sizes in supported  —  87  —  Lenz  Response to Audience Agreement  and unsupported conditions. Note that the largest responses on all cardiovascular measures—both with and without support—are observed in the present study. The smallest effect found statistically significant is that for DBP in the Germ et al. (1992) study (d  =  0.72).  Explanations ofDfferences Between Findings There are a number of possible explanations for the different findings in the four related studies. Each will be addressed and evaluated separately; however, there is no reason to assume that there is one single cause of variation between studies or that causes are mutually exclusive. Instead, it is expected that a variety of causes may be acting simultaneously and may have additive or interactive effects. Population differences. Differences in findings between similar studies conducted at different locations necessitates examination of the populations sampled. In the present case, population differences across studies are apparent, but there is no reason to suspect that population differences would lead to systematic differences in response to the provision of social support. All four of the closely related studies were conducted at universities using undergraduates as subjects. To this extent, populations appear similar across studies. However, there are numerous cultural and ethnic differences among the samples. The Germ et al. (1992) study was conducted at a small all-women’s college located in the metropolitan East Coast of the United States. Sheffield and Carroll (1995) conducted their study at a major university in Scotland. Neither Germ et al. nor Sheffield and Carroll reported ethnicity of their samples. Both the present study and the Lepore et  —88—  Lenz  Response to Audience Agreement  al. (1993) study were conducted at large West Coast universities with multicultural populations—Lepore’s in the United Stated and the present study in Canada. Lepore’s sample was approximately 48% Caucasian, 27% Asian, and 16% Latino. Subjects in the present study were approximately 59% Caucasian and 37% Asian. It is apparent that populations vary actoss studies, but at the present time there are inadequate data to predict the impact of these differences on CV response to the stress tasks or to experimentally provided support. Very little research has been reported on cross-cultural differences in CV response and none systematically addresses differences between localities in North America—an area generally, though probably mistakenly, treated as a cultural unit for the purposes of cross-cultural research. The most salient data on this issue are probably the preliminary analyses in the present study that compare Asian and Caucasian subjects on the major CV and self-report indices. On a brief series of exploratory analyses, no differences were found related to ethnicity. Because these analyses necessitated cutting cell sizes in half and resulted in serious reductions in statistical power, no firm conclusions should be drawn on the basis of these negative findings. The best that can be said is that there appear to be no data indicating a systematic impact of cultural and ethnic factors on the results of these studies. However, the only published study of social support and health in Asian men (Reed et al., 1984) produced different findings than similar studies in Caucasian populations (e.g., Berkman & Syme, 1979; Blazer, 1982; House et al., 1982; Orth-Gomer & Johnson, 1987). On this basis, continuing suspicion about as yet undetected cultural or ethnic effects may well be warranted.  —89—  Lenz  Response to Audience Agreement  Level of arousal. Differences between task-related effect sizes in the four studies suggest that there may be an optimal stress level for the observation of the effects of social support. This explanation hypothesizes a curvilinear relationship between stress-related arousal and the attenuating effects of support. The idea would be that an inadequate amount of stress would produce too little arousal to observe attenuation and too much arousal may override the potential benefits of supportive interaction and make attenuation more difficult. According to this explanation, the present study may have evoked too much arousal and Sheffield and Carroll’s (1995) too little for observation of supportrelated attenuation. Although this explanation is consistent with the general patterns of effect sizes across studies, the Lepore et al. (1993) data do not conform to expectations based on this explanation. In the Lepore study, the magnitudes of task-related response in the unsupported condition are very close to those observed in the present study, but this effect is confounded with position change since baseline measures were taken sitting and measures durint the speech standing. Support-related attenuation effects in the Lepore study are the largest observed in all four studies. Thus, the argument that a moderate level of stress is necessary for the observation of the protective effects of support is at best an incomplete explanation of differences between findings—the impact of postural change not withstanding. The logic of detecting a buffering effect requires an adequate level of stress induction, and the Sheffield and Carroll study may well have fallen below this essential level. The Lepore data suggest, however, that support-related attenuation can be detected even in the presence of strong stressors.  —  90  —  Lenz  Response to Audience Agreement  Type of support. A third possible explanation rests on the proviso explained by Cohen and Wills (1985) that in order to be effective, support must provide resources that are salient to the challenges imposed by a specific stressor. This explanation suggests that the type of support provided in the present study may have been less well matched to the stresses of the immediate situation than the types of support provided in the Lepore et al. (1993) and the Germ at al. (1992) studies. In the present study, support was operationalized as audience agreement with an expressed opinion, and this type of support is essentially the same as that provided in the Germ study. In contrast, support in the Lepore et al. study, was focused on coping with the speech task rather than on the topic of the speech—presumably because the topic for the speech was assigned. Confederates in the Lepore study provided encouragement for getting through the task, reminding the speakers that they were doing fine and that they were nearly finished. Why would audience agreement constitute stronger support in the Germ et al. (1992) study than in the present one? In the Germ et al. study, all subjects were speaking in the presence of two disagreeing laboratory confederates. In the Support condition, a third confederate agreed with the subject, while in the No-support condition, the third confederate was silent. It is possible that the argumentative tone and the atmosphere of dispute in the Germ et al. study may have set up a situation in which agreement could be regarded as a scarce—and therefore more highly valued—resource. If this is true, then the Cohen and Wills (1985) proviso would predict that any agreement with subjects may have filled an acute need and may, therefore, have been flinctionally supportive. The same level of agreement may constitute less salient support in a less conflictual setting. In the present  —  91  —  Lenz  Response to Audience Agreement  study and in the Lepore et al. (1993) study, speakers in the supported condition did not have to contend with audience disagreement from other confederates. In the absence of challenges to their opinions, subjects may have been more likely to focus on completing their speech and less likely to see speech content as essential. The type of support in the present study did not directly address this focus on simply completing the speech, whereas support in the Lepore et al. study did. There is some evidence that subjects in the present study were in fact focusing on simple completion of the speech task. Ratings suggest that 39 subjects during the last half of data collection found the most upsetting aspect of the experiment to be the need to speak for a full five minutes. Mean ratings of the 12 items added to the “Response to Speaking” questionnaire were compared. The two items with the highest ratings both had to do with speaking for a full five minutes (Means  =  46.30 and 41.54, SDs  =  29.03 and  28.84). In contrast, subjects rated the requirement of speaking on a topic they cared about ninth of the 12 items (Mean  =  21.69, SD  =  21.52). Thirty-two subjects (of 39) rated both  of the speaking duration items higher than the speech content item, and none rated the content item higher than both speech duration items. Although these data provide a plausible explanation of response patterns, it must be noted that they are far from definitive. Fewer than half of the subjects were surveyed in this manner, similar data are not available for the other studies, and systematic assessment of subjects’ attentional focus was not undertaken. Although this evidence is weak, the only indicators available suggest that subjects in the present study may have been more concerned with and therefore more focused on just  —  92  —  Lenz  Response to Audience Agreement  getting through the speech task than on the content of their speeches. To the extent that subjects were simply trying to complete the task, Cohen and Wills’ (1985) specificity principle would suggest that confederates’ agreement with their opinions may have been seen as less than salient support for dealing with that particular stressor. Although the specificity of support argument appears to explain the differences between the present results and those of Lepore et al. (1993) and Germ et al. (1992), it does not explain why agreeing and disagreeing confederates did not result in different CV responses in the Sheffield and Carroll study (1995). In this case, the task was clearly identified as. one requiring accuracy, and it is difficult to imagine why subjects would not find it supportive to hear that a more expert rater agreed with their answers. Divergence of Sheffield and Carroll’s findings from those of Germ and of Lepore is, therefore, more likely due to the fact that their task induced lower levels of CV response. Robustness of the phenomenon. This final explanation for differences in findings across related studies is at a more abstract level of analysis and generally subsumes the other explanations offered above. As suggested by Sheffield and Carroll (1995), divergence among findings suggests that the protective effects of social support on cardiovascular reactivity may be less robust and more situationally sensitive than was earlier believed. If this is in fact the case-as it appears to be—then the task of determining what factors affect the impact of social support on CV reactivity will depend on comparing findings from numerous studies in which support both attenuates and fails to attenuate CV response.  —  93  —  Lenz  Response to Audience Agreement  Directionsfor Future Research Results of the present study suggest several avenues for continued research. The most direct and immediate has to do with methodology. Investigation of how socially supportive behavior leads to attenuation of cardiovascular response depends on the ability to evoke and observe the phenomenon. To this end, it may be of use to explore the differences between subjects’ cardiovascular responses in the present study and those in studies by Germ et al. (1992) and Lepore et al. (1993). Specifically, it would be of interest whether slightly different support—more like that which Lepore used—would result in different outcome with laboratory procedures that are otherwise unchanged. Because variation from the present protocol would be quite minor, if subjects were to respond differently to slightly modified support, this would underscore the sensitivity of the phenomena of support-related attenuation of cardiovascular reactivity. Whether response to supportive behavior is studied in the laboratory as in the present study, or whether it is done in a subject’s natural environment as recommended by Sheffield and Carroll (1994, 1995), the mechanisms of action remain of primary interest. The results of the present investigation suggest that we know very little about the subtleties of the phenomena. It may be warranted to work on the problem with exploratory techniques by conducting discussions with subjects concerning their reactions to stress and their reactions to supportive and non-supportive behavior in others. The target of such an endeavor could be finding the degree to which subjects themselves could articulate the basic regulatory responses Taylor (1991) describes as attendant to reactions to negative events. If successful, an approach like this could be useful in developing  —94—  Lenz  Response to Audience Agreement  hypotheses about the types of social interactions subjects perceive as supportive and hypotheses about social cognitions related to cardiovascular response to support. Just as Tolman once advocated that a researcher try to see the maze from the point-of-view of the rat (Tolman, 1948), it may be of use to modern stress researchers to try to see our laboratory procedures through the eyes of our subjects. Similar exploration could be undertaken in a slightly different vein, by building upon previously successful identification of linkage between social support and ambulatory blood pressure (Linden et al., 1993). Although others have reported connections between ambulatory CV measures and social support (Unden et a!., 1991), it has not been investigated in detail what specific day-to-day events or interactions are perceived by subjects as supportive or helpful. The drawback is that exploration of these issues would require sophisticated event-monitoring techniques which would probably need to be implemented in conjunction with extensive interview.  Conclusions Both report of the supportiveness of the laboratory confederate and the strong cardiovascular response to the experimental procedures indicated a valid experimental procedure. The support-related attenuation of CV response to the stress task was apparent on heart rate, providing partial support for earlier findings in research that provoked adequate levels of CV stress response. The lack of significant findings on blood pressure measures—together with strong indications of internal validity—argues for viewing the  —95—  Lenz  Response to Audience Agreement  effects of social support on CV responsivity as more volatile and probably more situationally determined than was previously believed. Several minor goals were addressed by the present study. As in earlier research, men were found slightly more responsive to the stress-induction task, but no sex differences were found in response to supportive audience behavior. The present study contributes by providing additional evidence against the existence of important sex differences in cardiovascular response to manipulations of social support in laboratory research. Fluctuation in self-esteem was not a significant explanatory factor for the observed effects of social support. The refined self-report instruments produced generally interpretable results with better synchrony between self-report and physiological measures than previously observed. Results suggested that self-report of simple arousal level is probably an inadequate measure of subjective response to procedures like those used in the present study. The major goal of the present study—to examine the appropriateness of neutral controls in social support research—appears to have been accomplished with little ambiguity. A neutral or non-communicative audience response is clearly not more upsetting or arousing than straightforward disagreement. The present study has led to unanticipated conclusions. The identified reason for questioning the interpretation of laboratory social support research—that the type of control group was inadequate—is clearly wrong. However, instead of bolstering that research, the present findings present a challenge of a different type: The effects of support on cardiovascular reactivity appear less reproducible than anticipated and are  —  96  —  Lenz  Response to Audience Agreement  likely to be dependent on subtle situational factors that have not been clearly identified. The most likely cause of differences between closely related studies is that supportive behavior may need to be finely tuned to stress-related needs of a situation if it is to be effective. This notion, which is drawn from Cohen and Wills (1985), has potentially important implications for the design and implementation of supportive interventions, and it means that it is essential to tailor support to a recipient’s current needs. This principle was clearly articulated in applied writings in the field of social support a decade ago (e.g., Cohen, S. & Syme, 1985; Cohen, S. & Wills, 1985; Cohen, W. S., 1985; Wills, 1985). The present data add empirical support for the importance of this relatively common-sense principle and also suggest that determining the exact nature of those needs may at times be difficult.  —  97  —  Lenz  Response to Audience Agreement  REFERENCES  Allen, K. M., Blascovich, 3., Tomaka, J., & Kelsey, R. M. (1991). Presence of human friends and pet dogs as moderators of autonomic responses to stress in women. Journal of Personality and Social Psychology, ]., 582-589. Berkman, L. F. (1985). The relationship of social networks and social support to morbidity and mortality. In S. Cohen & S. L. Syme (Eds.), Social support and health (pp. 241-262). Orlando,FL: Academic Press. Berkman, L. F., & Syme, S. L. (1979). Social networks, host resistance, and mortality: A 9-year follow-up study of Alameda County residents. American Journal of Epidemiology, i, 186-204. Berkman, L. F., Vaccarino, V., & Seeman, T. (1993). Gender differences in cardiovascular morbidity and mortality: The contribution of social networks and support. Annals of Behavioral Medicine, j.., 112-118. Berntson, G. G., Cacioppo, J. T., Binkley, P. F., Uchino, B. N., Quigley, K. S., & Fieldstone, A. (1994). Autonomic cardiac control. III. Psychological stress and cardiac response in autonomic space as revealed by pharmacological blockades. Psychophysiology,  II, 599-608.  Blazer, D. G. (1982). Social support and mortality in an elderly community population. American Journal of Epidemiology, jj., 684-694.  —  98  —  Lenz  Response to Audience Agreement  Blumenthal, J. A., Jiang, W., Robinson, L., Budinger, S., Hayano, J., Waugh, R., Phillips, B., Thurber, D., Coleman, E., Frid, D., & O’Connor, C. (1992, March). Cardiovascular reactivity during mental stress predicts myocardial ischemia in the laboratory and during daily life. Paper presented at the 13th annual scientific sessions of the Society of Behavioral Medicine, New York. Boyce, W. T., & Chesterman, E. (1990). Life events, social support, and cardiovascular reactivity in adolescence. Journal of Developmental and Behavioral Pediatrics, II, 105-111. Burger, J. M. (1992). Desire for control and academic performance. Canadian Journal of Behavioural Science, 4, 147-155. Cacioppo, J. T., Berntson, G. G., Binkley, P. F., Quigley, K. S., Uchino, B. N., & Fieldstone, A. (1994). Autonomic cardiac control. II. Noninvasive indices and basal response as revealed by autonomic blockades. Psychophysiology, j, 586-598. Cassel, J. (1976). Contributions of the social environment to host resistance: The fourth Wade Hampton Frost lecture. American Journal of Epidemiology, 1Q4, 107-123. Cobb, 5. (1976). Social support as a moderator of life stress. Psychosomatic Medicine, 38, 300-314. Cohen, S., Evans, G. W., Stokols, D., & Krantz, D. 5. (1986). Behavior, health, and environmental stress. New York: Plenum Press.  —  99  —  Lenz  Response to Audience Agreement  Cohen, S., & Hoberman, H. M. (1983). Positive events and social supports as buffers of life change stress. Journal of Applied Social Psychology, i.., 99-125. Cohen, S., & Matthews, K. A. (1987). Social support, Type A behavior, and coronary artery disease. Psychosomatic Medicine, 4, 325-330. Cohen, S., Mermeistein, R., Kamarck, T., & Hoberman, H. M. (1985). Measuring the functional components of social support. In I. G. Sarason & B. R. Sarason (Eds.), Social support: Theory, research, and applications (pp. 73-94). Netherlands: Martinus Nijhoff. Cohen, S., & Syme, S. L. (1985). Issues in the study and application of social support. In S. Cohen & S. L. Syme (Eds.), Social support and health (pp. 3-22). Orlando,FL: Academic Press. Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychological Bulletin,  ,  310-357.  Cohen, W. S. (1985). Health promotion in the workplace: A prescription for good health. American Psychologist,  4, 213-216.  Coyne, J. C., & Bolger, N. (1990). Doing without social support as an explanatory concept. Special Issue: Social support in social and clinical psychology. Journal of Social and Clinical Psychology,  ,  148-158.  —100—  Lenz  Response to Audience Agreement  Edens, J. L., Larkin, K. T., & Abel, J. L. (1992). The effects of social support and physical touch on cardiovascular reactions to mental stress. Journal of Psychosomatic Research,  ,  371-382.  Fontana, A. F., Kerns, R. D., Rosenberg, R. L., & Colonese, K. L. (1989). Support, stress, and recovery from coronary heart disease: A longitudinal causal model. Health Psychology,  ,  175-193.  Fredrikson, M., & Matthews, K. A. (1990). Cardiovascular responses to behavioral stress and hypertension: A meta-analytic review. Annals of Behavioral Medicine, j, 30-39. Germ, W., Mimer, D., Chawla, S., & Pickering, T. G. (1995). Social support as a moderator of cardiovascular reactivity in women: A test of the direct effects and buffering hypotheses. Psychosomatic Medicine, 5j, 16-22. Germ, W., Pieper, C., Levy, R., & Pickering, T. G. (1992). Social support in social interaction: A moderator of cardiovascular reactivity. Psychosomatic Medicine, 54, 324-336. Germ  W., Pieper, C,, Marchese, L., & Pickering, T. G. (1992). The multi-dimensional  nature of active coping: Differential effects of enhanced control on cardiovascular reactivity. Psychosomatic Medicine,  54, 707-7 19.  Glass, G. V., & Hopkins, K. (1984). Statistical methods in education and psychology. Englewooci Cliffs, NJ: Prentice-Hall.  —  101  —  Lenz  Response to Audience Agreement  Heatherton, T. F., & Polivy, J. (1991). Development and validation of a scale for measuring state self-esteem. Journal of Personality and Social Psychology, Q, 895910. Heitzman, C. A., & Kaplan, R. M. (1988). Assessment of methods for measuring social support. Health Psychology, 7, 75-109. Helgeson, V. 5. (1993). Two important distinctions in social support research: Kind of support and perceived versus received. Journal of Applied Social Psychology, 23, 825-845. House, J. S., Robbins, C., & Metzner, H. M. (1982). The association of social relationships and activities with mortality: Prospective evidence from the Tecumseh  Community Health Study. American Journal of Epidemiology, jj, 123-140. House, J. S., Umberson, D., & Landis, K. R. (1988). Structures and processes of social support. Annual Review of Sociology, 14, 293-318. Howell, D. C. (1987). Statistical Methods for Psychology (2nd ed.). Boston: Duxbury Press. Janis, I. L., & Field, P. B. (1959). Sex differences and factors related to persuasibility. In C. I. Hovland & I. L. Janis (Eds.), Personality and persuasibility (pp. 55-68). New Haven, CT: Yale University Press.  —102—  Lenz  Response to Audience Agreement  Kamarck, T. W., Manuck, S. B., & Jennings, J. R. (1990). Social support reduces cardiovascular reactivity to psychological challenge: A laboratory model. Psychosomatic Medicine, 52, 42-5 8. Kamarck, T., Peterman, A., & Marco, C. -Social integration and cardiovascular reactivity in a married community sample. Kaplan, G. A., Salonen, J. T., Cohen, R. D., Brand, R. J., Syme, S. L., & Puska, P. (1988). Social connections and mortality from all causes and from cardiovascular disease: Prospective evidence from eastern Finland. American Journal of Epidemiology, j., 370-380. Katz, R., & Wykes, T. (1985). The psychological difference between temporally predictable and unpredictable stressful events: Evidence for information control theories. Journal of Personality and Social Psychology, 4, 781-790. Kiecolt-Glaser, J. K., & Greenberg, B. (1984). Social support as a moderator of the aftereffects of stress in female psychiatric inpatients. Journal of Abnormal Psychology, ,  192-199.  Knox, S. S., Theorell, T., Svensson, J. C., & Waller, D. (1985). The relation of social support and working environment to medical variables associated with elevated blood pressure in young males: A structural model. Social Science and Medicine, j, 525531.  —  103  —  Lenz  Response to Audience Agreement  Kors, D. J., Linden, W., & Germ, W. (1995). Evaluation interferes with social support: Effects cardiovascular stress reactivity in women. Manuscript submitted for publication. Krantz, D. S., & Manuck, S. B. (1984). Acute psychophysiologic reactivity and risk of cardiovascular disease: A review and methodologic critique. Psychological Bulletin, 96, 435-464. Lamensdorf A. M., & Linden, W. (1992). Family history of hypertension and cardiovascular changes during high and low affect provocation. Psychophysiology, 29, 558-565. Lepore, S. J., Allen, K. A. M., & Evans, G. W. (1993). Social support lowers cardiovascular reactivity to an acute stressor, Psychosomatic Medicine,  ,  518-524.  Linden, W., Chambers, L., Maurice, J., & Lenz, J. W. (1993). Sex differences in social support, self-deception, hostility, and ambulatory cardiovascular activity. Health Psychology, j, 376-380. Linden, W., & Zinimermann, B. (1984). Comparative accuracy of two new electronic devices for the noninvasive determination of blood pressure. Biofeedback and Self Regulation, 9, 229-239. Manuck, S. B. (1994). Cardiovascular reactivity and cardiovascular disease: “Once more into the breach”. International Journal of Behavioral Medicine,  —104—  1 4-3 1.  Lenz  Response to Audience Agreement  Manuck, S. B., Kasprowicz, A. L., & Muldoon, M. F. (1990). Behaviorally-evoked cardiovascular reactivity and hypertension: Conceptual issues and potential associations. Annals of Behavioral Medicine, 12, 17-29. Mehrabian, A., & Russell, 3. A. (1974). An approach to environmental psychology. Cambridge, MA: M.I.T. Press. Orth-Gomer, K., & Johnson, J. (1987). Social network interaction and mortality: A six year follow-up study of a random sample of the Swedish population. Journal of Chronic Disease, 40, 949-957. Pickering, T. G., & Germ, W. (1990). Cardiovascular reactivity in the laboratory and the role of behavioral factors in hypertension: A critical review. Annals of Behavioral Medicine, j, 3-16. Rachman, S. J. (1990). Fear and courage (2nd ed,). New York: W. H. Freeman. Reed, D., McGee, D., & Yano, K. (1984). Psychosocial processes and general susceptibility to chronic disease. American Journal of Epidemiology, 119, 3 56-370. Reed, D., McGee, D., Yano, K., & Feinleib, M. (1983). Social networks and coronary heart disease among Japanese men in Hawaii. American Journal of Epidemiology,  11.7, 384-396. Russell, J. A. (1979). Affective space is bipolar. Journal of Personality and Social Psychology, 7, 345-3 56.  —  105  —  Lenz  Response to Audience Agreement  Russell, J. A., Weiss, A., & Mendelsohn, G. A. (1989). Affect Grid: A single-item scale of.pleasure and arousal. Journal of Personality and Social Psychology,  7, 493-502.  Sarason, I. G., Levine, H. M., Basham, R. B., & Sarason, B. R. (1983). Assessing social support: The social support questionnaire. Journal of Personality and Social Psychology, 4, 127-139. Schneiderman, N., & McCabe, P. M. (1989). Psychophysiologic strategies in laboratory research. In N. Schneiderman, S. P. Weiss, & P. G. Kaufmann (Eds.), Handbook of research methods in cardiovascular behavioral medicine (pp. 349-3 64). New York: Plenum Press. Schoenbach, V. J., Kaplan, B. H., Fredman, L., & Kieinbaum, D. G. (1986). Social ties and mortality in Evans County, Georgia. American Journal of Epidemiology, i, 577-591. Sheffield, D., & Carroll, D. (1994). Social support and cardiovascular reactions to active laboratory stressors. Psychology and Health,  ,  305-316.  Sheffield, D., & Carroll, D. (1995, April). Task-induced cardiovascular activity and the presence of a supportive or undermining other. Presented at the poster session of the Society of Behavioral Medicine, San Diego, CA. Spitzer, S. B., Llabre, M. M., Ironson, G. H., Geilman, M. D., & Schneiderman, N. (1992). The influence of social situations on ambulatory blood pressure. Psychosomatic Medicine, 4, 79-86.  —106—  Lenz  Response to Audience Agreement  Stevens, J. (1992). App1ied Multivariate Statistics for the Social Sciences (2nd ed.). Hillsdale, NJ: Lawrence Eribaum Associates. Syme, S. L. (1974). Behavioral factors associated with the etiology of physical disease: A social epidemiological approach. American Journal of Public Health, 4, 1043-1045. Syme, S. L., Hyman, M. M., & Enterline, P. E. (1964). Some social and cultural factors associated with the occurrence of coronary heart disease. Journal of Chronic Disease,  ]J, 277-289. Tabachnick, B. G., & Fidell, L. 5. (1989). Using Multivariate Statistics (2nd ed.). New York: Harper-Collins. Taylor, S. E. (1991). Asymmetrical effects of positive and negative events: The mobilization-minimization hypothesis. Psychological Bulletin, ]IQ, 67-85. Tolman, E. C. (1948). Cognitive maps in rats and men. Psychological Review,  ,  189-  208. Uchino, B. N., Kiecolt-Glaser, J. K., & Cacioppo, J. T. (1992), Age-related changes in cardiovascular response as a function of a chronic stressor and social support. Journal of Personality and Social Psychology,  ,  839-846.  Unden, A.-L., Orth-Gomer, K., & Elofsson, 5. (1991). Cardiovascular effects of social support in the work place: Twenty four-hour ECG monitoring of men and women. Psychosomatic Medicine,  ,  50-60.  —107—  Lenz  Response to Audience Agreement  Waliston, B. S., Alagna, S. W., DeVellis, B. M., & DeVellis, R. F. (1983). Social support and physical health. Health Psychology, 2, 367-391. Watson, D., & Tellegen, A. (1985). Toward a consensual structure of mood. Psychological Bulletin, 9, 2 19-235. Watzlawick, P., Bavelas, J. B., & Jackson, D. D. (1967). Pragmatics of Human Communication: A study of interactional patterns, pathologies, and paradoxes. New York: W. W. Norton & Company. Weiss, J. M. (1968). Effects of coping responses on stress. Journal of Comparative and Physiological Psychology,  ,  25 1-260.  Welin, L., Tibblin, G., Svardsucld, K., Tibblin, B., Ander-Peciva, S., Larsson, B., & Wilhelmsen, L. (1985). Prospective study of social influences on mortality: The study of men born in 1913 and 1923. Lancet, 1, 915-918. Wills, T. A. (1985). Supportive functions of interpersonal relationships. In S. Cohen & S. L. Syme (Eds.), Social support and health (pp. 61-82). Orlando,FL: Academic Press.  —  108  —  Lenz  Response to Audience Agreement  APPENDIX A  —  QuEsTIoNNAmis AN]) SCRiPTS  Appendix A 1.  —  Number 1.  Recruitment Poster  WHAT HAPPENS WHEN YOU. TALK? Find out how your body responds when you speak your mind or listen to the opinions of others. We are looking for volunteers to participate in a study examining changes in blood pressure that occur when people discuss issues that are important to them.  1 /2 Credit Points For more information—or to volunteer—please call the Psychophysiology Lab, Dr. W. Linden, Department of Psychology, U.B.C. (Project # 93-37)  822-3800  Open 8:30—4:30 weekdays.  Please leave a message for a return call if you phone after hours. —109—  Lenz  Response to Audience Agreement  Appendix A 2.  —  Number 2.  Telephone Screening Guide PHONE SCREEN—SPEECH STUDY  Introduce yourself as from Dr. Linden’s Cardiovascular Lab at UBC Department of Psychology. Tell you are calling about the Speaking and Listening study. Inform that the study takes a bit over an hour plus some preparation time. They get 2 (TWO!)credit points for participating. Inform that you have to ask some questions to make sure they are right for the study. If anything questionable is going on with them, fill in a Phone Data screening sheet with detail. (If clear exclusion or clearly an acceptable subject, no sheet necessary.)  Have you ever been told by a doctor that you have any of these diseases? High blood pressure (Get & record any info.) Heart disease (exclude) Kidney or Liver disease (exclude) Diabetes (exclude) Are you taking any medication? What is itfor? (Query BP meds, anxiety meds, antidepressants. Get drug name, dosage & how long) Is English your main language? YES/NO If “NO”--- How old were you when your schooling began to be in English? (Exclude f over 10.) (Record age if under 10) We ask that you avoid afew things before coming to the lab. For 2 hours before, please 1) Do not drink anything with caffeine or alcohoL 2) Do not smoke or use tobacco or nicotine in anyform. 3) Do not exercise strenuously (break a sweat). Also, 4) Try to arrive as relaxed as possible. —110—  Lenz  Response to Audience Agreement  Set appointment time That means that after (apt time less 2 hrs) you won’t be able to drink coffee tea or coke or alcohol, smoke, or exercise. Do you think you will be able to do that all right? Give directions to the Kenny Bldg if necessary. Tell how to find the LAB. Give Lab phone & ask to call if problems.  —111—  Lenz  Response to Audience Agreement Appendix A  3.  —  Number 3.  Consent Form  Note: Original was printed (single-spaced) on Department ofPsychology letterhead  CONSENT FORM I, (Print Name)  agree to participate in a research  project entitled “Physiologic responses during speaking and listening” which is to be conducted in the cardiovascular psychophysiology laboratory in the Psychology Department at the University of British Columbia as a dissertation project by Joseph Lenz, M.S., with Dr. Wolfgang Linden as the faculty advisor (phone 822-4156). This study will involve the completion of various questionnaires during a sixty-minute lab session with an additional 5 to 10 minutes possibly needed for explanations. The study requires me to relax for 10 minutes while my blood pressure is monitored. I will then be asked to prepare a speech on a social topic important to me and to deliver it in front of another subject while the speech is recorded on videotape and blood pressure is monitored. After the speech I will be asked to rest quietly and to fill out some more brief questionnaires. After that, questions about the study will be answered. I understand that multiple cardiovascular functions will be monitored throughout the session. All monitors are non invasive in nature, are harmless, and lead to no foreseeable risks to my health or safety.  I understand that I may refuse to participate and withdraw from the study at any time without influence on my class standing. All information collected is strictly confidential.  —112—  Lenz  Response to Audience Agreen2ent  All information will be recorded in group form and will remain strictly anonymous. There will be no identification of me personally on any permanent records.  If I have a problem with this experiment, I understand that I may contact either Joseph Lenz at 822-7915 or Dr. Linden at 822-4156. In addition, Dr. Jerry Wiggins is in charge of the subject recruitment committee and can be contacted at 822-6536.  Signature below indicates the following: (1) The procedures to be followed have been explained to me. (2) Questions, if any, have been answered to my satisfaction. (3) I have read and have understood the content of this consent form. (4)1 have received a copy of this consent form.  Research Participant  Witness  (Signature)  Date:  Laboratory contact Number: 822-3800  —113—  Lenz  Response to Audience Agreement  Appendix A 4.  —  Number 4.  Collection Form for Demographic Data CULTURE QUESTIONNAIRE  Subject#  Date  Instructions: For each question below, circle the number of the item that describes you best. Some items also ask you to provide additional information. Please answer all of the questions.  1.  What languages can you speak fluently?____________________________________  2.  Which of the languages did you learn first?  3.  Which language do you prefer for conversation?  4.  If you were born outside North America, at what age did you first come to North America?  5.  Please name the country in which you were born. Canada  Other—Where?_________________________  Don’t know  Where was your mother born? Other—Where?________________________ Canada  Don’t know  Where was your father born? Other—Where?________________________ Canada  Don’t know  6.  How do you describe your own cultural or ethnic identity? (For example: Canadian, Japanese, Cantonese, American, East-Indian, First Nations, or French-Canadian, Indo-Canadian, Asian-Canadian, Ukranian-Canadian, etc.)  —114—  Lenz  7.  Response to Audience Agreement  What was the ethnic origin of the friends and peers you had, as a child up to age 6? 1—Almost exclusively from my own cultural group 2—Mostly from my own cultural group 3—About equally from my group and Anglo or other groups (“Anglo” means English-speaking.) 4—Mostly Anglos or from cultural groups other than my own 5—Almost exclusively Anglos or from cultural groups other than my own  8.  Whom do you now associate with in the community? 1—Almost exclusively from my own cultural group 2—Mostly from my own cultural group 3—About equally from my group and Anglo or other groups (“Anglo” means English-speaking.) 4—Mostly Anglos or from cultural groups other than my own 5—Almost exclusively Anglos or from cultural groups other than my own  9. Do you participate in special occasions, holidays, traditions, etc. that are specific to your culture of ethnic origin? 1—Nearly all of them 2—Most of them 3—Some of them 4—A few of them 5—None at all  10. How much do you identify with each culture: (a) Your culture of ethnic origin: 1—Very much 2—Mostly 3—Partially or somewhat 4—A little 5—Not at all (b) Anglo-Canadian culture: 1—Very much 2—Mostly 3—Partially or somewhat 4—A little 5—Not at all —  115—  Lenz  Response to Audience Agreement  Appendix A 5.  —  Number 5.  State Self-Esteem Scale (Current Thoughts) CURRENT THOUGHTS  Subject # Date I/p Instructions: This is a questionnaire designed to measure what you are thinking at this moment. There is, of course, no right answer for any statement. The best answer is what you feel is true of yourself at this moment. Be sure to answer all of the items, even if you are not certain of the best answer. Again, answer these questions as they are true for you RIGHT NOW. Please use thefollowing scale when answering the questions: Not at all 1  A little bit 2  Somewhat 3  Very much 4  1.  I feel confident about my abilities.  2. 3.  I am worried about whether I am regarded as a success of failure. I feel satisfied with the way my body looks right now.  4.  I feel frustrated or rattled about my performance.  5.  I feel that I am having trouble understanding things that I read.  6.  I feel that others respect and admire me.  7.  I am dissatisfied with my weight.  8.  I feel self-conscious.  9.  I feel as smart as others.  10.  I feel displeased with myself  11.  I feel good about myself  12.  I am pleased with my appearance right now.  13.  I am worried about what other people think of me.  14.  I feel confident that I understand things.  15.  I feel inferior to others at this moment.  16.  I feel unattractive.  17.  I feel concerned about the impression I am making.  18.  I feel that I have less scholastic ability right now than others.  19.  I feel like I’m not doing well.  20.  I am worried about looking foolish.  —116—  Extremely 5  Lenz  Response to Audience Agreement  Appendix A 6.  —  Number 6.  Interpersonal Support Evaluation List (ISEL) LS.E.L  Subject #  Date  Instructions: This scale is made up of a list of statements each of which may or may not be true about you. For each statement, we would like you to circle T if the statement is probably TRUE about you or circle F if the statement is probably FALSE about you. You may find that some of the statements are neither clearly true nor clearly false. In these cases, try to decide quickly whether ‘probably TRUE T” or “probably FALSE F” is most descriptive of you. Although some questions will be difficult to answer, it is important that you pick one alternative or the other. Remember to circle only one of the alternatives for each statement. Please read each item quickly but carefhlly before responding. Remember that this is not a test and there are no right or wrong answers. T  Probably TRUE F Probably FALSE  =  T F 1. I have a hard time keeping pace with my friends. T F 2. There are very few people I trust to help solve my problems. T F 3. No one I know would throw a birthday party for me. T F 4. In general, people don’t have much confidence in me. T F 5. Most of my friends are more successful at making changes in their lives than I am. T F 6. If for some reason I were put in jail, there is someone I could call who would bail me out. T F 7. I have someone who takes pride in my accomplishments. T F 8. I feel that I’m on the fringe in my circle of friends. T F 9. When I need suggestions for how to deal with a personal problem, I know there is someone I can turn to.  —117—  Lenz  Response to Audience Agreement  T F 10. If I were sick and needed someone to drive me to the doctor, I would have trouble finding someone. T F 11. If I wanted to go out of town (e.g., to the coast) for the day, I would have a hard time finding someone to go with me. T F 12. Most of my friends are more interesting than I am. T F 13. If I needed a quick emergency loan of $100, there is someone I could get it from. T F 14. There is really no one who can give me objective feedback about how I’m handling my problems. T F 15. There is at least one person I know whose advice I really trust. T F 16. If I needed some help in moving to a new home, I would have a hard time finding someone to help me. T F 17. When I feel lonely, there are several people I could easily call and talk to. T F 18. I feel there is no one with whom I can share my most private worries and fears. T F 19. There are several different people with whom I enjoy spending time. T F 20. There is someone who I feel comfortable going to for advice about sexual problems. T F 21. Most people I know don’t enjoy the same things that I do. T F 22. If I wanted to have lunch with someone, I could easily find someone to join me. T F 23. If I had to go out of town for a few weeks, someone I know would look after my home (the plants, pets, yard, etc.). T F 24. If I needed a ride to the airport very early in the morning, I would have a hard time finding anyone to take me. T F 25. I don’t often get invited to do things with others.  —118—  Lenz  Response to Audience Agreement  T F 26. I think that my friends feel that I’m not very good at helping them solve problems. T F 27. I regularly meet or talk with members of my family or with friends. T F 28. I am more satisfied with my life than most people are with theirs. T F 29. ff1 decide on a Friday afternoon that I would like to go to a movie that evening, I could find someone to go with me. T F 30. If I had to mail an important letter at the post office by 5:00 and couldn’t make it, there is someone who could do it for me. T F 31. Most people I know think highly of me. T F 32. I am able to do things as well as most other people. T F 33. There is really no one I can trust to give me good financial advice. T F 34. There is someone I can turn to for advice about handling hassles over household responsibilities. T F 35. There is no one I could call on if I needed to borrow a car for a few hours. T F 36. If a family crisis arose, few of my friends would be able to give me good advice about handling it. T F 37. If I were sick, there would be almost no one I could find to help me with my daily chores. T F 38. There is someone I could turn to for advice about changing my job or finding a new one. T F 39. If I got stranded 10 miles out of town, there is someone I could call to come and get me. T F 40. I am closer to my friends than most other people.  —119—  Lenz  Response to Audience Agreement  Appendix A 7.  —  Number 7.  Affect Grid GrId 2  Subject #  Date High Arousal  Stress  Stress  Excitement  High Arousal  Excitement  x x  Sleepiness  Depression  Examule 1:  Pleasant Feelings  Relaxation  Depression Examole 2:  A neutral, everyday feeling  DIRECTIONS:  Stress  Relaxation  A somewhat exciting and moderately pleasant feeling  Please rate how you are feeling riaht now, by placing an X in one of the boxes anywhere on the grid. See example grids above. High Arousal  Excitement  Pleasant Feelings  Unpieaswt Feeling.  Depression  Sleepiness  Sleepiness  —120—  Relaxation  Lenz  Response to Audience Agreement Appendix A  8.  —  Number 8.  Important Topics Questionnaire  IMPORTANT TOPICS Subject#  Date  Instructions: This questionnaire is designed to find out what social or political topics are most important to you. Please select and rank the top three topics. Number them 1 2 3 with number I being your top choice. Base your choices on the following: How much you personally care about the issue (How important it seems to you) How much you know about the issue -  -  You should care strongly about and know a good deal about your number one choice. MARK THE TOP THREE 1-2-3: Abortion rights Animal rights (experimentation using animals) Death penalty First Nations land claims Legal handling of pedophiles (child molesters) Logging clear-cuts (e.g., Clayoquot Sound) Penalties for environmental protesters Physician-assisted suicide Refugee status for immigrants Use of U.N. “Peacekeeping” forces (e.g., in Somalia, Bosnia, Rwanda) Young Offenders’ Act Other—name it Other—name it  —  121  —  Lenz  Response to Audience Agreement  Appendix A 9.  —  Number 9.  Reactions to Speaking Questionnaire  Reactions to Speaking Instructions: For each question below, mark the line to indicate how you think or feel. Use a single line or slash mark to indicate your response. Rate how you felt during your speech by marking positions on the lines below. I. Stressed  Relaxed  Safe  Unsafe  Uncomfortable  Comfortable  Angry  Not angry  Not nervous  Nervous  Calm  Excited  Worried  Content  II. Did you explain your opinions and your feelings clearly? Not at all  Very Much  Did you present your arguments and opinions in an interesting way? Not at all  Very Much  Do most other people you know agree with the opinions you expressed in your speech? Not at all Very Much As you spoke, did your voice express how strongly you feel? Not at all  Very Much  During the speech, did you have an urge to leave the laboratory? Not at all  Very Much  —122—  Lenz  Response to Audience Agreement  Would you have preferred to give your talk with no one else present? Not at all Very Much Did the other student really listen to your talk? Not at all  Very Much  Do you think the other student understands your position well enough to explain it to someone else? Not at all Very Much Do you feel respected by the student who listened to your speech? Not at all  Very Much  ifi. How would you describe the student who listened to your speech? Supportive  Unsupportive  Close  Distant  Rejecting  Accepting  Friendly  Unfriendly  Warm  Cold  Helpful  Unhelpful  NOTE: The following 12 Items were added to the Reactions to Speaking Questionnaire for the last 39 subjects. W. Rate how upsetting (that is, bothersome or anxiety-provoking) each of the following events were during the experiment: Having to talk on a topic you care about— Not at all upsetting  Very upsetting  —  123  —  Lenz  Response to Audience Agreement  Trying to come up with enough material to talk about for 5 minutes— Not at all upsetting Very upsetting Saying your speech to the other student— Not at all upsetting  Very upsetting  Talking in front of an operating video camera— Not at all upsetting  Very upsetting  Having a one-way mirror in the room as you were talking— Not at all upsetting  Very upsetting  Being able to see yourself in the mirror as you were talking— Not at all upsetting  Very upsetting  Making eye contact with the other student while you were speaking— Not at all upsetting Very upsetting Having the other student face you and look at you as you spoke— Not at all upsetting  Very upsetting  The other student’s facial expression while you were speaking— Not at all upsetting  Very upsetting  Trying to continue to speak for the full 5 minutes— Not at all upsetting  Very upsetting  Things the other student said while you were speaking— Not at all upsetting  Very upsetting  Knowing that your blood pressure was being monitored as you spoke— Not at all upsetting Very upsetting  —124—  Lenz  Response to Audience Agreement  Appendix A  —  10.  10. Speech Helper Form  SPEECH HELPER Remember, begin by: Telling your opinion clearly & strongly Tell the important facts. Make arguments in support of your opinion. How has your belief changed your life? Do you act differently because of your belief? Why? How did the change come about? Argue against opposing opinions. Show how inconsistent or uncaring or inaccurate opposing views often are. Don’t hesitate to be really critical. Make a joke or ridicule opposing ideas. Take the opposing view to extreme and reduce it to absurdity. Make definitions. How do others misunderstand the topic or use terms or ideas inaccurately. If others held your opinion, how would the world (Canada, BC, Vancouver) be a better place? Give some examples. Give as many as you can. If your opinion is ignored and its opposite becomes dominant, how would this make the world a worse place? Give examples. Explore the impact of this topic from different angles: Moral Social Legal Financial Logical How did you first learn of this issue? Tell it like a story. What convinced you of its importance? Tell about discussions or arguments you have had on this topic. —  125  —  Lenz  Response to Audience Agreement  Don’t hesitate to tell it as a story. What have you read about this issue? What have you seen in the media about it? What have you experienced on this issue? How about others you know or have heard about? Repeat your main point. Or repeat anything else you have said.  —  126  —  Lenz  Response to Audience Agreement  Appendix A  Number 11.  11. Debriefing Form (Given to all participants)  Original printed (single-spaced) on Department of Psychology letterhead. STUDY EXPLANATION Physiologic Responses during Speaking and Listening A number of studies have shown that speaking in front of an audience leads to increases in blood pressure and heart rate. A few have demonstrated that these increases are less pronounced if the audience expresses agreement and support for the speaker. The type of audience used for comparison has been non-responsive. We believe that it is possible that speaking to a non-responsive listener is particularly difficult and may raise blood pressure. If this is true, the health-protecting effects of having a supportive listener may have been over- estimated in past studies. The study in which you are participating is designed to answer this and other related questions. The goals of the study are as follows: 1)  Examine the effects of warm agreement, opposition, and silence on  cardiovascular arousal to clarify whether social support lowers blood pressure response to speaking in public or whether non-responsive audiences increase blood pressure. 2)  Examine the effects of agreement, disagreement, and non-committal  response on a speaker’s self-esteem and compare changes in self-esteem with blood pressure changes. 3)  Compare responses of men and women in identical social situations.  —127—  Lenz  Response to Audience Agreement  Subjects have each been assigned at random a listener who agrees or disagrees or who remains silent and uncommitted. The other tt subject’ in this study is a U.B.C. student who is employed in this laboratory to play the role of a listener. She (or he) has been trained to give precise verbal and non-verbal response to speakers while maintaining specified eye contact, posture, distance, etc. His (or her) responses during your speech were pre-planned and do not necessarily represent actual opinions and responses. Because other students in your classes may be participating in this study, we ask your cooperation in not discussing what you have learned by participating and not showing other students this sheet. Thank you once again for participating in this study. If you would like to know the results of this study, a copy of the report will be available in six to nine months from the following address: W. Linden, Ph.D. Department of Psychology 2136 West Mall University of British Columbia Vancouver, BC  V6T 1Z4  Phone: 822-3800 i you are interested in this area of research and would like to read more about it, you could start with: Cohen, 5. (1988). Psychosocial models of the role of social support in the etiology of physical disease. Health Psychology, 7, 269-297. Germ, W., Pieper, C., Levy, R., & Pickering, T. (1992). Social Support in social interaction: A moderator of cardiovascular reactivity. Psychosomatic Medicine, 54, 324-336.  —  128  —  Lenz  Response to Audience Agreement  Appendix A  —  Number 12.  12. Debriefing Script  Now that you have completed your participation in our study, we want to explain its full purpose to you. A number of studies have shown that speaking in front of an audience leads to increases in blood pressure and heart rate. A few have demonstrated that these increases are less pronounced if the audience expresses agreement and support for the speaker. The type of audience used for comparison has been non-responsive. We believe that it is possible that speaking to a non-responsive listener if particularly difficult and may raise blood pressure. If this is true, the health-protecting effects of having a supportive listener may be over estimated. The study in which you are participating is designed to answer this and other related questions. The goals of the study are as follows: 1)  Examine the effects of warm agreement, opposition, and silence on  cardiovascular arousal to clarifj whether social support lowers blood pressure response to speaking in public or whether non-responsive audiences increase blood pressure. 2)  Examine the effects of agreement, disagreement, and non-committal  response on a speaker’s self-esteem and compare changes in self-esteem with blood pressure changes. 3)  Compare responses of men and women in identical social situations.  —129—  Lenz  Response to Audience Agreement Subjects have each been assigned at random a listener who agrees or disagrees or  who remains silent and uncommitted. The other “subject” in this study—the listener—is a U.B.C. student who is employed in this laboratory to play the role of a listener. She (or he) has been trained to give precise verbal and non-verbal response to speakers while maintaining constantly controlled eye contact, posture, distance, etc. His (or her) responses during your speech were pre planned and do not necessarily represent actual opinions and responses. We would like to give you an opportunity to talk frankly with this lab assistant now. (The coifederate is invited into the room and introduces self) “Hello, my name is  and I am working in this  study. I was assigned to act as if I agreed (or disagreed or to act as if I had no response) with your speech. If you would like to know how I actually feel about (the topic of the speech), I’d be happy to tell you.” (Briefdiscussion mayfollow  4 the subject desires it.) f  We would like to know how you feel about the experiment now that we have explained the design and purposes to you. Please feel free to share your reactions to the study and to make suggestions if you have any. We appreciate your questions and your participation. (Hands subject the “Study Explanation” paper and explains:) This paper tells you the basic information we have just described to you. In addition, it tells you how to get a copy of the results of the study when it is complete in about six  —130—  Lenz  Response to Audience Agreement  months. It also suggests a couple of related articles you may want to read if you are interested in research on social support. Because other students in your classes may be participating in this study, we ask your cooperation in not discussing what you have learned by participating and not showing other students this sheet. Thank you once again for participating.  —  131  —  Lenz  Response to Audience Agreement  APPENDIX  B  —  STATISTICAL TABLES  Appendix B 1.  —  Number 1.  Oneway ANOVA Table for Rating of Confederate’s Supportiveness by Experimental Condition  (Dependent Variable Source of Variation  Support; Groups  =  Experimental Condition)  SS  Signif. ofF  Between Groups  27238.14  2  13619.07  Within Groups  13220.48  87  151.96  Total  40458.61  89  —  132  —..  89.62  .0000  Lenz  Response to Audience Agreement  Appendix B  —  Number 2.  2. item Descriptions and Correlations among Items in the Distress Scale For Items 1-6: Rate how you felt during your speech by marking positions on the lines below. Anchors were as follows (Response direction equated): 1. Relaxed / Stressed  2. Safe / Unsafe  3. Comfortable / Uncomfortable  4. Not nervous / Nervous  5. Calm! Excited  6. Content / Worried  7. During your speech, did you have an urge to leave the laboratory? (Not at all / Very Much) 8. Would you have preferred to give your talk with no one else present? (Not at all / Very Much) Correlations  ITEM 1  2  3  4  5  6  7  8  .44  .73  .66  .70  .72  .53  .47  .58  .51  .31*  .58  .69  .52  .65  .49  .76  .58  .57  .60  .71  .55  .43  .59  .43  .37  .63  .56  2 3 4 5 6 7  .66 Pearson product-moment correlations. N = 90 for all correlations. *  P  <  .005; for all other correlations,p  —  133  <  —  .001.  Lenz  Response to Audience Agreement  Appendix B 3.  —  Number 3.  Tables for Multivariate and Univariate Oneway ANOVAs of Distress and Performance Ratings  MANOVA Table (Dependent Variables Groups  =  =  Performance and Distress;  Experimental Condition)  Statistic  Value  Wilk’s Lambda  .85  Hvpoth. DF 4.00  3.56  Error DF  Signif ofF  172.00  .008  Oneway ANOVA Table for Rating of Distress during the Speech (Dependent Variable  =  Distress; Groups  =  Experimental Condition)  Source of Variation Between Groups  Signif. ofF 4097.27  2  2048.64  Within Groups  25912.01  87  297.84  Total  30009.28  89  6.88  .002  Oneway ANOVA Table for Rating of Speech Performance (Dependent Variable = Performance; Groups  =  Experimental Condition)  Source of Variation Between Groups  Signif. ofF 2542.16  2  1271.08  Within Groups  25359.93  87  291.49  Total  27902.08  89  —134—  4.36  .016  Lenz  Response to Audience Agreement  Appendix B 4.  —  Number 4.  ANOVA Tables for Systolic Blood Pressure  Tests of Between-Subjects Effects. Source of Variation SEX  Signif. ofF 4607.30  1  4607.30  37.40  .000  COND  217.23  2  108.62  0.88  .42  SEXbyCOND  121.15  2  60.58  0.49  .61  10347.37  84  123.18  WITHIN CELLS  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation TASK SEXbyTASK CONDbyTASK SEX by COND by TASK WITHIN CELLS  Signif. ofF 26580.15  1  26580.15  681.60  19.56  1  19.56  0.50  .48  150.61  2  75.30  1.93  .15  12.70  2  6.35  0.16  .85  3275.70  84  39.00  —  135  —  .000  Lenz  Response to Audience Agreement  Appendix B 5.  —  Number 5.  ANOVA Tables for Diastolic Blood Pressure  Tests of Between-Subjects Effects. Source of Variation  SS  Signif. ofF  SEX  369.80  1  369.80  5.02  .03  COND  365.87  2  182.93  2.48  .09  20.13  2  10.06  0.14  .87  6186.96  84  73.65  SEXbyCOND WITHIN CELLS  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation TASK  Signif. ofF 25402.61  1  25402.61  853.54  .000  172.09  1  172.09  5.78  .018  CONDbyTASK  63.08  2  31.54  1.06  .35  SEX by COND by TASK  61.76  2  30.88  1.04  .36  2499.96  84  29.76  SEXbyTASK  WITHIN CELLS  Simple Effects Analysis of SEX at Baseline and Speech Source of Variation  Signif. ofF  Tests involving ‘TASK at Baseline’ Within-Subject Effect. SEXatBASEL1NE WITHIN+RESIDUAL  18.68  1  18.68  4170.92  88  47.40  0.39  .53  9.16  .003  Tests involving ‘TASK during Speech’ Within-Subject Effect. SEXduring SPEECH  523.21  1  523.21  WITHIN+RESIDUAL  5026.84  88  57.12  —136—  Lenz  Response to Audience Agreement  Appendix B 6.  —  Number 6.  ANOVA Tables for Heart Rate  Tests of Between-Subjects Effects. Source of Variation  Signif. ofF  SEX  2435.91  1  2435.91  8.80  COND  1238.17  2  619.08  •2.24  .11  508.79  2  254.39  0.92  .40  23241.11  84,  276.68  SEXbyCOND WITHINCELLS  .004  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation TASK SEXbyTASK CONDbyTASK SEXbyCONDbyTASK WITHIN CELLS  SS  DF  MS  F  Signif. ofF  20207.14  1  20207.14  234.63  133.76  1  133.76  1.55  .22  1506.14  2  753.07  8.74  .000  315.06  2  157.53  1.83  .17  7234.42  84  86.12  .000  Simple Effects Analysis of CONDITION at Baseline and Speech Source of Variation  SS  Signif. ofF  Tests involving ‘TASK at Baseline’ Within-Subject Effect. COND at BASELINE  576.52  2  288.26  WITIIIN+RESIDTJAL  12297.51  87  141.35  2.04  .136  4.37  .016  Tests involving ‘TASK during Speech’ Within-Subject Effect. COND during SPEECH WITHJN+RESIDUAL  2167.80  2  1083.90  21571.53  87  247.95  —  137  —  Lenz  Response to Audience Agreement  Appendix B 7.  —  Number 7.  ANOVA Tables for the Arousal Subscale of the Affect Grid  Tests of Between-Subjects Effects. Source of Variation  Signif. ofF  SEX  0.20  1  0.20  0.08  .78  COND  3.03  2  1.52  0.57  .56  SEXbyCOND  6.63  2  3.32  1.26  .29  221.93  84  2.64  WITHIN CELLS  Tests involving t 1 T ASK Within-Subject Effect. Source of Variation TASK  SS  DF  MS  F  Signif. ofF  149.42  1  149.42  105.71  SEXbyTASK  0.56  1  0.56  0.39  .53  CONDbyTASK  1.94  2  0.97  0.69  .51  SEX by COND by TASK  0.34  2  0.17  0.12  .89  118.73  84  1.41  WITHiN CELLS  —  138  —  .000  Lenz  Response to Audience Agreement  Appendix B 8.  —  Number 8.  ANOVA Tables for the Pleasantness Subscale of the Affect Grid  Tests of Between-Subjects Effects. Source of Variation SEX COND SEXbyCOND WITHIN CELLS  £S.  MS.  Signif ofF  0.45  1  0.45  0.14  .71  41.68  2  20.84  6.25  .003  2.63  2  1.32  0.40  .67  279.93  84  3.33  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation TASK SEX by TASK COND by TASK SEX by COND by TASK WITHIN CELLS  Signif. ofF 29.61  1  29.61  10.75  .002  4.05  1  4.05  1.47  .23  13.34  2  6.67  2.42  .095  1.23  2  .62  0.22  .80  231.27  84  2.75  Simple Effects Analysis of CONDITION at Pre- and Post-Task Source of Variation  J1  Signif. ofF  Tests involving ‘TASK at Pre-Task’ Within-Subject Effect. COND atPRE-TASK  5.27  2  2.63  WITHIN+RESIDUAL  202.33  87  2.33  1.13  .33  6.82  .002  Tests involving ‘TASK at Post-Task’ Within-Subject Effect. COND at POST-TASK  49.76  2  24.88  WITHIN+RESLDUAL  317.23  87  3.65  —139—  Lenz  Response to Audience Agreement  Appendix B 9.  —  Number 9.  ANOVA Tables for the Appearance Esteem Subscale of the SSES  Tests of Between-Subjects Effects. Source of Variation  55  SEX  53.65  1  53.65  1.60  .21  COND  9.56  2  4.7  0.14  .87  SEXbyCOND  3.32  2  1.66  0.05  95  2788.72  83  33.60  WITHIN CELLS  Signif. ofF  M.  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation  SS  Signif. ofF  TASK  0.91  1  0.91  0.63  .43  SEXbyTASK  0.01  1  0.01  0.01  .94  COND by TASK  2.30  2  1.15  0.80  .45  SEX by COND by TASK  1.24  2  0.62  0.43  .65  118.99  83  1.43  WITHEN CELLS  —140—  Lenz  Response to Audience Agreement  Appendix B  —  Number 10.  10. ANOVA Tables for the Social Esteem Subscale of the SSES Tests of Between-Subjects Effects. Source of Variation  SS  SEX  25.33  1  25.33  0.40  .53  168.03  2  84.02  1.32  .27  24.43  2  12.22  0.19  .83  5299.40  83  63.85  COND SEX by COND WITHIN CELLS  Signif. ofF  Tests involving TASK’ Within-Subject Effect. Source of Variation  Signif. ofF  TASK  5.90  1  5.90  1.25  .27  SEXbyTASK  3.37  1  3.37  0.71  .40  COND by TASK  19.48  2  9.74  2.06  .14  SEX by COND by TASK  14.29  2  7.15  1.51  .23  393.38  83  4.74  WITHIN CELLS  —  141  —  Lenz  Response to Audience Agreement  Appendix B  —  Number 11.  11. ANOVA Tables for the Performance Esteem Subscale of the SSES Tests of Between-Subjects Effects. Source of Variation SEX COND SEXbyCOND  WITHIN CELLS  DF  MS  Signif ofF  9.36  1  9.36  0.19  .66  107.04  2  53.52  1.11  .34  0.26  2  13.00  0.00  .99  4004.03  83  48.24  Tests involving ‘TASK’ Within-Subject Effect. Source of Variation  Signif. ofF  TASK  0.21  1  0.21  0.06  .80  SEXbyTASK  5.57  1  5.57  1.66  .20  15.88  2  7.94  2.37  .10  1.02  2  0.51  0.15  .86  277.86  83  3.35  CONDbyTASK SEX by COND by TASK  WITHIN CELLS  —142—  Lenz  Response to Audience Agreement  Appendix B  —  Number 12.  12 Means and (Standard Deviations) of Change in Cardiovascular and Self-report Measures by Experimental Condition  SBP  DBP  HR  AFFECT GRID -Arousal  -Pleasantness  SSES  -Social  -Performance  -Appearance  *  n  AGREE  NEUTRAL  DISAGREE  22.29  23.91  26.72  (8.85)  (9.16)  (8.12)  24.69  22.09  24.49  (6.78)  (9.07)  (7.77)  15.11  19.49  28.97  (7.58)  (15.08)  (15.65)  1.77  1.60  2.10  (1.79)  (1.35)  (1.79)  —0.13  —0.83  -1.47  (2.49)  (2.29)  (2.21)  1.31*  0.03  —0.23  (2.69)  (3.36)  0.62*  0.00  (2.06)  (2.99)  (2.56)  0.17*  —0.37  —0.23  (1.47)  (1.61)  (1.91)  29; for all other scores, n  =  —  30.  143  —  (3.18) —0.83  

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.831.1-0088355/manifest

Comment

Related Items