
##4001166 ABSTRACT . Unionization votes of 133 members of an employee association at a large , midwestern , American university were examined . General beliefs , specific beliefs , normative pressures ( co-workers , family , and close friends ) , and dissatisfaction with the employee association influenced votes for unionization . Job satisfaction , age , race , and gender did not influence votes for unionization . <p> What are the attitudes of members of employee associations toward unionization ? Do some employee associations become unions ? A nationally representative survey of employees , conducted by Louis Harris and Associates in 1984 for the AFL-CIO Committee on the Evolution of Work ( 1985 ) , presented some interesting findings on the attitudes of nonunion workers in the United States . Although 56% of nonunion workers indicated that they would join an association , only 37% indicated that they would vote for union representation if an election were held at their workplace ( Ichniowski &; Zax , 1990 ) . <p> Prior research has suggested that some associations can become unions , and the AFL-CIO Committee on the Evolution @ @ @ @ @ @ @ @ @ @ Yet there is debate about associations and their linkage to unions . Some academics and union officials believe that associations can be substitutes for unions ; others see associations as complementary and as stepping stones to unions . <p> Although most private-sector association members are retirees and workers on lay-off , associations in the public sector are workplace specific , and the memberships consist of current employees ( Ichniowski &; Zax , 1990 ) . Since the 1970s , several employee associations in the public sector have merged with national labor unions , such as the American Federation of State , County and Municipal Employees ( AFSCME ) and the Service Employees International Union ( SEIU ; Cornfield , 1991 ) . In fact , in an analysis of census data on local government departments , Ichniowski and Zax ( 1990 ) found that the presence of an association in 1977 was a strong predictor of the formation of a union by 1982 . <p> Most of the research on employee associations has focused on association-union mergers ( Cornfield , 1991 ) . One area that has received special attention has @ @ @ @ @ @ @ @ @ @ of mergers , traits of merger partners , and barriers to mergers ) . However , only one published study has addressed the opinions of state employee association members toward merger with a union . In that study , Cornfield ( 1991 ) examined the impact of employees ' socioeconomic status and opportunity for upward mobility on their approval for merger with a labor union . Cornfield found a significant effect for both factors . But it is surprising that , although considerable research has been done to examine factors that influence union voting intent among nonunion employees , virtually no research has been undertaken to investigate the factors that influence an employee association member to vote for unionization . <p> Several causes of votes for unionization have been identified , and beliefs about unions and job satisfaction have been singled out as the more prominent ones ( Barling , Fullagar , &; Kelloway , 1992 ) . Although much has been written about the effect of union beliefs on unionization , there has been a wide diversity in the measures used to capture the effects of beliefs about unions in @ @ @ @ @ @ @ @ @ @ 1987 ) . Some researchers have used belief measures that simply refer to " unions " ( Kochan , 1979 ) ; others have referred to the effect of a union at the respondents ' workplace ( Youngblood , Denisi , Molleston , &; Mobley , 1984 ) . In previous studies , researchers have used either specific or general beliefs about unions as predictors of voting intentions ; however , Premack and Hunter ( 1988 ) , in a meta-analysis , treated both measures as the same . More recently ; in a meta-analysis of some determinants of union voting intent , Deshpande ( 1992 ) showed that the two constructs are distinct . <p> Recently , the effect of normative pressures on union voting intentions has been investigated . As Montgomery ( 1989 , p. 263 ) correctly noted , " The voting choice can not be adequately understood . . . without taking into account the influence that others have on the voter . " Youngblood et al . ( 1984 ) , in a sample of members of a consumer panel , found that a respondent 's @ @ @ @ @ @ @ @ @ @ had a significant effect on the respondent 's union voting intent . Zalesny ( 1985 ) found that university faculty were more likely to vote for unions when social forces were favoring unions . More recently , Montgomery ( 1989 ) , in a study of clerical employees of a public university , found that opinions of close friends , family members , and co-workers did influence respondents ' voting intentions , even though most of the respondents denied such an influence . <p> In the present study , demographic controls included measures for age , race , and gender . Additional control variables included a measure of ( a ) overall job satisfaction and ( b ) satisfaction with the existing employee association . The implications of age are mixed ( Chacko &; Greer , 1982 ) . Some researchers have found support for the proposition that younger workers are more militant , whereas others have found support for the proposition that younger workers prefer not to work under a rigid seniority system , typically found in unionized settings . There has consistently been a positive relationship between being non-White @ @ @ @ @ @ @ @ @ @ ) , most likely because union racial equality policies and protective policies inhibit possible employer discrimination . Hypotheses about gender effects have been generally similar to those about race , but studies with voting intent as the dependent variable have generally come to the conclusion that there is no gender effect ( Deshpande &; Fiorito , 1989 ) . <p> Satisfaction with the job has been shown in many studies to have a negative relationship with the desire for unionization ( Freeman &; Medoff , 1984 ) . Thus , I hypothesized that job satisfaction among association members would decrease pro-union voting intent . Finally , an overall measure of satisfaction with the employee association was used . I proposed that employees satisfied with their association would vote for unionization . This hypothesis is consistent with the findings of researchers who believe that employees are more likely to reject unionization if they are dissatisfied with the performance of their association ( Ichniowski &; Zax , 1990 ) . Examination of the impact of such factors as specific beliefs , general beliefs , and normative pressures as determinants of unionization votes among @ @ @ @ @ @ @ @ @ @ Participants <p> A survey of 330 members of an employee association of administrative and professional employees at a large , midwestern ( U.S. ) university was completed in 1992 . The faculty ( American Association of University Professors ) , hourly workers ( American Federation of State , County , and Municipal Employees ) , and the police at this university were already unionized . The goal of this survey was to collect information on attitudes of employee association members toward unions . The survey also included a wide range of individual and workplace-related variables . <p> The survey was conducted shortly after the members of the association participated in a secret ballot that would allow the association to petition for different unions and an election conducted by the state 's Employment Relations Commission . One month before the secret ballot , the association members met with representatives of three unions ( United Auto Workers , American Association of University Professors , and the Michigan Education Association ) . Concurrently , the university had released a four-page statement that answered many questions about what would happen if the association unionized . @ @ @ @ @ @ @ @ @ @ addressed the association members on some of the negative consequences of unionization . <p> A total of 243 employee association members responded to my survey . Of these , 63 association members were dropped from the study because they did not participate in the secret ballot . In addition , responses of 47 association members were not used because they did not respond to all items in the study . Thus , 133 respondents remained for the analysis . Measures <p> Self-reported voter behavior on the secret ballot was the dependent variable . The behavior was measured on a 2-point scale ( 1 =favor unionization , 0 = reject unionization ) . <p> On a 4-point Likert-type scale ranging from agree strongly ( 4 ) to disagree strongly ( 1 ) , the respondents were asked to react to 10 general statements about unions : ( a ) unions improve wages , ( b ) unions improve job security , ( c ) unions improve recognition you receive for the work you do , ( d ) unions improve fringe benefits ( e.g. , medical benefits , leave ) , ( @ @ @ @ @ @ @ @ @ @ unions improve participation in decisions that affect your job , ( g ) unions improve health and safety , ( h ) unions improve treatment by supervisors , ( i ) unions reduce discrimination against women by employer , and ( j ) unions reduce discrimination based on race or national origin . <p> On a 3-point Likert-type scale ranging from get better ( 3 ) to get worse ( 1 ) , the respondents were asked to react to the impact of unionization at the workplace on the following 10 items : ( a ) the pay you receive , ( b ) your job security , ( c ) the recognition you receive for the work you do , ( d ) your fringe benefits ( e.g. , medical benefits , leave ) , ( e ) your chances of job advancement , ( f ) your opportunity to participate in decisions that affect your job , ( g ) health and safety problems , ( h ) treatment by supervisors , ( i ) discrimination against women by employers , and ( j ) discrimination based on race @ @ @ @ @ @ @ @ @ @ union effects at respondents ' current jobs , and the general beliefs measures refer to generalizations . The specific beliefs measures of this study are consistent with those used by Deshpande and Fiorito ( 1989 ) . <p> The respondents were asked to indicate the perceptions of co-workers , relatives , and close friends about unions . These perceptions were measured by three items on a 3-point Likert-type scale ranging from yes ( 1 ) , to uncertain ( 2 ) , to no ( 3 ) : ( a ) Do you think your co-workers would favor a union ( reverse coded ) ? ( b ) Do you think your relatives would oppose your joining a union ? ( c ) Do you think your close friends would oppose your joining a union ? <p> A 4-point Likert-type scale ranging from very satisfied ( 4 ) to not at all satisfied ( 1 ) was used to measure overall job satisfaction , and a similar 4-point Liken-type scale was used to measure satisfaction with the association . For demographic variables , the respondents were asked to indicate their age @ @ @ @ @ @ @ @ @ @ choices was used to measure race . The responses were later compressed into Whites ( 1 ) and non-Whites ( 2 ) . The respondents were also requested to report their sex ( 1= male , 0 = female ) . <p> I conducted a regression analysis to examine the significance of the variables . Because the dependent variable was dichotomous , I used a probit procedure instead of the more conventional ordinary least squares regression procedure . Results <p> The correlation results show that votes for unionization were affected by general beliefs , specific beliefs , normative pressures , job satisfaction , and association satisfaction ( see Table 1 ) . But it would be premature to draw conclusions from these results , because there were intercorrelations among the predictors . <p> Results of the probit procedure show that general beliefs , specific beliefs , normative pressures , and satisfaction with association had a significant influence on votes for unionization ( see Table 2 ) . Age , sex , race , and job satisfaction did not significantly affect unionization votes . Discussion and Implications <p> The results of this @ @ @ @ @ @ @ @ @ @ unions play an important role in votes for unionization , the impact of a union at the workplace ( specific beliefs ) may be a more significant determinant of votes for unionization . This finding provides support for Gagala 's ( 1983 ) contention that union organizers need to put more emphasis on the employee 's own workplace than on general contributions of unions to society and to the work force . This can be done by providing potential members with information on labor contracts at comparable workplaces and on the significant impacts that have been made . Union organizers should also emphasize policies , practices , and unique characteristics of their union when trying to organize employees . <p> Unionization votes among association members are also influenced by the perceptions of co-workers , close friends , and family members . In fact , during an election campaign , consultants often advise employers to address their letters to the employees and their spouses ( Gagala , 1983 ) . A spouse is typically interested in activities concerning work ( especially union activity ) and can influence the employee 's final decision @ @ @ @ @ @ @ @ @ @ . <p> Job satisfaction did not influence unionization decisions of association members . It is possible that job dissatisfaction was a reason to join the association but not a factor in the transition from an association to a union . This theory is consistent with findings of previous research , which suggest that unions may not be the only means chosen by employees to redress dissatisfaction ( DeCotiis &; LeLouarn , 1981 ) . Thus , labor educators and researchers should examine the factors that influence the decision to join an association as opposed to joining a union . <p> There was a lack of support for the proposition that association satisfaction will have a positive impact on employees ' unionization votes . One could argue that dissatisfaction with association membership ( perhaps related to the association 's ability to influence overall employment conditions for the occupation or for an employer ) could lead to the choice of a representative that is , potentially , more militant . Anonymous notes sent by respondents with their completed surveys supported this contention . A major concern of the employees was that the association @ @ @ @ @ @ @ @ @ @ by any undesirable actions of the university , nor could it influence the university 's compensation policies . These comments imply that the desire for unionization could be driven by factors beyond the jurisdiction of an employee association . <p> The results of this study show the value of using micro-data rather than union organization-level data to address the association-to-union transition . Associations in the public sector are workplace specific , with current employees as members ; associations in the private . sector may not be workplace specific or may not even include working members . Thus , associations in the private sector are quite unlikely to be called on to be bargaining agents ( Ichniowski &; Zax , 1990 ) . Moreover , the sample in this study consisted of employee association members at a single , public university . The work environment of a public university is different from the work environment at a private-sector organization and from other public-sector organizations . Despite these limitations , this study has some important implications for labor educators , researchers , and union organizers . TABLE 1 Means , Standard Deviations , @ @ @ @ @ @ @ @ @ @ <p> Note . N = 133 . Coefficient alphas are shown on the diagonal in parentheses . TABLE 2 Results of Probit Analysis PREFORMATTED TABLE <p> Note . N = 133 . <p> * p < .10. **p < .05. ***p < .01 . <p> 
##4001167 ABSTRACT . Conservatism is often assumed to imply a continuity of values and positions between past and present . Previous research has demonstrated , however , that there are discontinuities between historical and contemporary political conservatism and thai modern political conservatives endorse programs and legislation once regarded as liberal . Testing whether the same pattern may hold for religious conservatism was the purpose of this research . Perceptions of conservative positions in Christian history were assessed among young adults ( N = 221 ) . Results showed that individuals who identified themselves as conservatives were less likely to recognize past conservative positions on religious issues than individuals who identified themselves as liberals . These findings were not accounted for by lack of knowledge about religious history on the part of conservatives . Furthermore , women were more likely than men to accurately identify conservative and liberal positions of the past . The findings suggest that religious conservatism , like political conservatism , does not necessarily involve a continuity of positions between past and present . <p> Conservatism is often assumed to imply a continuity of values and positions @ @ @ @ @ @ @ @ @ @ maintaining traditions and institutions ( Minogue , 1967 ) . Previous research on political conservatism , however , has demonstrated important discontinuities between historical and contemporary conservatism . For example , in a study of college-educated young adults , Sugar , Viney , and Rote ( 1992 ) found that individuals who were politically conservative in terms of contemporary issues ( e.g. , national health insurance ) supported historically liberal programs and legislation ( e.g. , child labor laws ) . <p> To date , no research has explored possible continuities and discontinuities between historical and contemporary religious conservatism . For example , do modem religious conservatives support historically liberal positions , such as vaccination to avert disease ? 1 An exploration of the perceived roots of modern religious conservatism seems timely in light of the current national debate on " traditional " values in education and social policy . A study of the relationship between historical and contemporary conservative religious views may also contribute to our understanding of some of the philosophical and psychological characteristics of conservative thought . <p> Another important question is whether perceptions of historical religious conservatism @ @ @ @ @ @ @ @ @ @ do not even report the gender composition of their sample ( e.g. , Kellstedt &; Smidt , 1991 ; Tamney , Johnson , &; Burton , 1992 ) . There is indirect evidence , however , that gender may play a role in views of religious positions . For example , a study of young adults found that women and liberals were less likely to justify male dominance than men and conservatives ( Lottes &; Kuriloff , 1992 ) . In another study , the relationship between fundamentalism and discriminatory attitudes toward Blacks , women , homosexual persons , and communists varied according to gender ( McFarland , 1989 ; but see Kirkpatrick , 1993 , for opposite findings ) . Because religious positions about women have evolved to a greater degree than positions about men , it is likely that women will be more accurate than men at recognizing conservative programs of the past . <p> This study is an adaptation and extension of a study of political conservatism by Sugar et al . ( 1992 ) . In recognition of the variability of beliefs and values across religions , @ @ @ @ @ @ @ @ @ @ contemporary Christian views , as perceived by Christian respondents . Based on the findings of Sugar et al. , we hypothesized that modem Christian conservatives would be less likely than modern Christian liberals to recognize conservative positions of the past . We anticipated that conservatives ' less accurate recognition of past religious positions , compared with liberals , might be a function of their limited awareness of historical religious events . Our second hypothesis was that women , especially women who identified themselves as liberal , would be more likely than men to accurately recognize religious positions of the past . Method Participants <p> Two hundred twenty-one undergraduate students ( 141 women and 80 men ) enrolled in introductory psychology classes participated in the study as one of several ways to fulfill a research requirement . Because the study was limited to exploring Christian views , we included only respondents who identified themselves as having a Christian background . In terms of religious orientation , participants identified themselves as Protestant ( n = 103 ) , Catholic ( n = 72 ) , nondenominational Christian ( n = 20 ) , @ @ @ @ @ @ @ @ @ @ other ( n = 8 ) . The " other " category included individuals who identified themselves as having no current Christian affiliation but whose parents were either Catholic or Protestant . Ages of participants ranged from 17 to 36 with a mean age of 18 years ( SD = 1.6 ) . The sample was composed of 155 first-year students , 41 sophomores , 19 juniors , and 6 seniors ; of these students , 187 were European Americans , 14 were Latinas/os , 8 were Asian Americans , 4 were African Americans , 3 were Native Americans , and 5 did not specify ethnicity . All participants signed a consent form and were treated in accordance with the " ethical principles of psychologists " ( American Psychological Association , 1992 ) . Materials <p> The Christian Religious History Quiz ( henceforth referred to as the CRH-Quiz ) was designed to measure participants ' knowledge of Christian religious history ( see Table 1 ) . Based on a review of literature on the history of Christianity ( Reid , Linder , Shelley , &; Stout , 1990 ) , we @ @ @ @ @ @ @ @ @ @ Christian ( true for both Catholics and Protestants ) statements . Three historically inaccurate statements were also included ( one presumably Catholic , one presumably Protestant , and one general Christian ) . <p> The final CRH-Q consisted of 19 statements ( 5 Protestant , 6 Catholic , and 8 general ) describing issues or ideological positions supported or opposed by Catholic and/or Protestant conservatives . These statements covered topics such as the separation between science and religion ( Items 3 , 8 , 15 ) , the status of women ( Items 1 , 5 , 6 ) , the regulation of sexual behavior ( Items 2 , 4 , 7 , 9 , 18 ) , the separation of church and state ( Items 10 , 11 , 14 , 17 ) , and the literal interpretation of the Bible ( Item 13 ) . Participants were asked to indicate whether the statements were historically true or false . A respondent 's CRH-Quiz score was obtained by averaging incorrect responses across the 19 items . Each incorrect response received a score of 1 . Averaged CRH-Quiz scores ranged from @ @ @ @ @ @ @ @ @ @ identification and lower scores indicating more accurate identification . <p> In addition to the CRH-Quiz , two scales were developed consistent with the method used by Sugar et al . ( 1992 ) . The first , entitled the Christian Religious Orientation Scale ( henceforth referred to as the CROS ) , was designed to assess whether respondents were conservative , moderate , or liberal in their current religious orientation ( see Table 2 ) . The second , Perception of Historical Christian Conservatism Scale ( henceforth referred to as the PHCCS ) , was designed to measure perceptions of conservative positions in Christian history ( see Table 3 ) . <p> The Christian Religious Orientation Scale ( CROS ) consisted of 19 statements describing positions that are endorsed by most Christian orthodoxies in the United States ( e.g. , creationism , sexual abstinence before marriage , male ministry , and the authority and infallibility of the Bible ; Reid et al. , 1990 ) . Several items were adapted from Fullerton and Hunsberger 's ( 1982 ) Unidimensional Scale of Christian Orthodoxy . The remaining items were based on a @ @ @ @ @ @ @ @ @ @ e.g. , Zodhiates , 1990 ) . Respondents were asked to indicate the extent of their agreement with these positions on a 7-point scale . Eleven of the scale items were worded so that a strongly agree response reflected a conservative religious orientation ( e.g. , " The Bible in every detail is the literal word of God " ) ; for the rest of the items a strongly agree response reflected a liberal religious orientation ( e.g. , " The professional ministry or priesthood should be open to women " ) . These latter items were scored on a 7-point scale with a strongly agree response receiving a score of 7 and a strongly disagree response receiving a score of 1 . The former items were reverse scored . <p> The final CROS score was computed for each participant by averaging across the 19 items . In addition , participants were asked to give a personal definition of religious conservatism and liberalism . These definitions were used to confirm knowledge of the meaning of conservatism and liberalism . Participants were also asked to rate on a 7-point scale how religiously @ @ @ @ @ @ @ @ @ @ as the religious self-rating item ) . <p> The Perception of Historical Christian Conservatism Scale ( PHCCS ) consisted of the same items as the CRH-Quiz with different instructions . Instead of judging whether a statement was true or false , participants were asked to assume that the statements were historically accurate and to judge whether each item was more likely to represent the position of a conservative or a liberal . Eight items reflected the position of a conservative . For example , one PHCCS conservative item stated that " Until 1961 , it was illegal in most states east of the Mississippi River to keep a business open on Sunday . Sunday was considered the appointed day of rest and worship . " Respondents were asked to indicate on a scale from 1 to 7 whether those who were opposed to opening a place of business on Sunday were more likely to be religiously liberal or conservative . A score of 1 indicated that the respondent judged a given position to be conservative and a score of 7 indicated that a respondent judged a given position to be liberal @ @ @ @ @ @ @ @ @ @ liberals and were reverse scored . For example , one item stated that " In the 18th century , the vaccine preventing smallpox was introduced . Some Christians denounced the practice of inoculation as interfering in God 's work . " 2 Respondents were asked to assume that the statement was historically accurate and to indicate on a 7-point scale whether those who were favorable to vaccination were more likely to be liberal or conservative . Respondents who answered liberal ( 7 ) received a score of 1 . <p> The three historically inaccurate items were included in the scale to disguise the purpose of the scale , but were not entered in the final score . The final PHCCS score was computed for each participant by averaging across the 16 historically accurate items . The final scores could range from I ( accurate judgment ) to 7 ( inaccurate judgment ) . Higher scores indicated more errors in the identification of the stance of conservatives on religious historical events . Procedure <p> Because some CROS items share themes with PHCCS items , we thought it possible that previous experience with @ @ @ @ @ @ @ @ @ @ , the order of presentation of these two scales was balanced . Half of the participants completed the three scales in the following order : CROS , CRH-Quiz , followed by the PHCCS . The other half completed the scales in the following order : CRH-Quiz , PHCCS , followed by the CROS . Upon completion of the three scales , participants were asked to provide information regarding their current religious affiliation , personal religious background , frequency of attendance at religious services , and their parents ' religious backgrounds . Participants were then debriefed on the goals of the study . Results <p> An alpha level of .05 was used for all statistical tests . No order effects were found for the CROS and PHCCS ; data were therefore collapsed across order of presentation . Internal consistency of the CROS scale was verified by Cronbach 's alpha ( .90 ) . The CROS score was found to be significantly correlated with the religious self-rating item , r ( 220 ) = .65 , p < .01 . Accordingly , the mean CROS score and the self-rating item were averaged to @ @ @ @ @ @ @ @ @ @ The TRO score could range from 1 to 7 ; the observed range was from 1.2 to 6.9 , with higher scores indicating a liberal orientation and lower scores indicating a conservative orientation . Consistent with the method used by Sugar et al . ( 1992 ) , participants were categorized as conservative if their mean TRO score was between 1.0 and 3.95 ( n = 69 ) , moderate if their mean score was between 3.96 and 4.95 ( n = 67 ) , and liberal if their mean score was between 4.96 and 7.0 ( n = 85 ) . The internal consistency of the PHCCS was confirmed by Cronbach 's alpha ( .78 ) . The overall mean score for the PHCCS was 2.6 ( SD = .77 ) ; the range was 1.1 to 6.3 . Higher scores indicate more errors in the identification of the stance of conservatives on religious historical events . <p> To assess whether knowledge of Christian history varied depending on current religious orientation , CRH-Quiz scores of participants classified as conservative , moderate , or liberal were compared . Based on the @ @ @ @ @ @ @ @ @ @ CRH-Quiz for conservatives was .29 , for moderates .29 , and for liberals .28 . A one-way analysis of variance ( ANOVA ) indicated no statistically significant difference in knowledge of Christian history across the three groups . <p> Finally , PHCCS scores were analyzed in a 2 x 3 ( Gender x Religious Orientation ) analysis of variance ( ANOVA ) to test the hypotheses that modern conservatives would be less likely to recognize past conservative positions than liberals , and that women would be more likely than men to identify historically conservative positions . A statistically significant effect for religious orientation was found , F(2,215) = 7.10 , p = .001 , indicating a difference between the three groups with respect to errors in identifying past conservative positions ( see Figure 1 ) . Religious orientation accounted for 6% of the variance in PHCCS scores ( etasup 2 = .06 ) . A priori comparisons using a Bonferroni adjustment ( p < .03 ) indicated that liberals ( M = 2.37 , SD = .65 ) were more accurate than conservatives ( M = 2.78 , SD = .75 @ @ @ @ @ @ @ @ @ @ .85 ) . Conservatives and moderates did not differ from each other in terms of recognition errors . With an alpha level of .05 , a statistically significant main effect for gender was also found , F(1,215) = 5.82 , p = .017 , indicating that the women identified conservative positions of the past more accurately than the men ( Ms = 2.52 and 2.78 , respectively ) . Gender accounted for 2% of the variance in PHCCS scores ( etasup 2 = .02 ) . There was no statistically significant interaction between gender and religious orientation . Discussion <p> The present data on religious conservatism are consistent with previous findings on political conservatism ( Sugar et al. , 1992 ) . We found evidence of important discontinuities between contemporary and historical conservatism . More specifically , the present data point to a failure of contemporary religious conservatives and moderates to identify religious conservative positions of the past , compared with liberals . These findings were not accounted for by differences across modern religious orientation groups in knowledge about the occurrence of these religious events . In other words , the @ @ @ @ @ @ @ @ @ @ be aware that certain positions ( e.g. , the perception that women who are menstruating are unclean , PHCCS Item 9 ) were endorsed , but were nonetheless less likely than the liberals to recognize these as conservative positions ( Ms for Item 9 were 2.41 for conservatives , 2.16 for moderates , and 1.4 for liberals ) , F(2,218) = 21.51 , p = .00001 . Although religious orientation accounted for only 6% of the variance in PHCCS , we believe that our findings are compelling because they were obtained under the most inauspicious of circumstances ( i.e. , with a sample of a college-educated individuals who were well informed about the historical occurrence of the religious events chosen for this study ) . As noted by Prentice and Miller ( 1992 ) , small effects can be impressive if they emerge in unlikely situations . <p> The failure on the part of the conservatives and moderates to recognize past conservative religious programs may involve both political and psychological processes . As we noted earlier , conservatives , by definition , aim at preserving the old order . In a @ @ @ @ @ @ @ @ @ @ , change is constantly creeping in . For conservatives , what is to be conserved may be constantly shifting ( Minogue , 1967 ) . When liberal advocates for change prevail , the proposed change ( e.g. , in the religious domain , marriage for priests ; in the political domain , an entitlement program , such as Social Security ) may be incorporated in a new social structure . Such a structure may result in changes in the community 's belief system and may become a candidate for conservation . The position of early conservatives who opposed the change may no longer be relevant to the contemporary conservative agenda . <p> This theory can be applied to critique the evolution of liberal thought as well . It is likely that modem liberals would fail to identify past failed liberal programs if such programs had become inconsistent with modem liberal beliefs and values . The difference is that , in a dynamic society , liberals are more likely to embrace novelty than conservatives , although conservatives may try to instigate changes that bring back past structures . Thus , psychologically and @ @ @ @ @ @ @ @ @ @ find themselves in the position of wanting to abandon and even forget past agendas . <p> The second main finding of this study was that women were significantly more likely than men to identify conservative positions of the past , as we predicted . Again , this finding is statistically significant ( p < .05 ) , although the amount of variance accounted for by gender is small ( 2% ) . In accordance with the arguments set forth earlier , it makes sense that women would have a more accurate recollection of historically conservative positions . Many of the changes in Christian religious doctrine initiated by liberals involved an expansion in the role of women . Some liberal positions ( e.g. , the rejection of the menstruation taboos ) have been completely assimilated by mainstream Christianity . As a result , women may be more motivated to remember past religious positions because many of these positions had far-reaching consequences for the quality of their lives . It is of interest that , in each group ( conservatives , moderates , and liberals ) , women made fewer errors in identifying @ @ @ @ @ @ @ @ @ @ Figure 1 ) . <p> Yet , it is somewhat surprising that a gender effect would be found among well-educated respondents from a large secular university with an established Women 's Studies program . One would think that these college men would be as likely as the women to have been exposed to information about conservative positions in religious history . It is possible , however , that these men processed the information differently than the women . For men , information about past conservative positions may have conflicted with well-established beliefs and strong positive feelings about Christianity . The historical data might have been unconsciously miscoded to preserve the familiar beliefs and feelings , as suggested by a theory of conceptual conservatism recently proposed by Nissani ( 1994 ) . " The most difficult mental act , " writes Nissani , " is to rearrange a familiar bundle of data , to look at it differently " ( p. 310 ) . <p> Although small effects are typical in religion research , future research on religious conservatism should experiment with different operationalizations of the dependent and independent variables to test @ @ @ @ @ @ @ @ @ @ present data on religious conservatism and those on political conservatism ( Sugar et al. , 1992 ) accurately reflect major discontinuities between historical and contemporary conservatism , they suggest a curious irony . In a dynamic , diverse world , conservatism is changing so fast that it is no longer consistent with its earlier expressions . The conservative nature of the human mind may be all that remains . <p> The authors are grateful to Alicia S. Cook and David B. Wohl for their comments and suggestions for revision . Religious conservatives opposed vaccination on the grounds that it flies " in the face of Providence " and encroaches " on the prerogatives of Jehovah , whose right it is to wound and smite " ( White , 1896/1978 , p. 56 ) . The original version of the scale used for this study inadvertently noted that the time and introduction of the smallpox vaccine was the 17th century . Re-analysis of the data without this item did not affect the outcome of this study . TABLE 1 Items , Means , and Standard Deviations for the Christian Religious History Quiz @ @ @ @ @ @ @ @ @ @ CRH-Quiz , participants were asked to indicate whether the statements were historically true or false . b Letters in the parentheses to the left of each item indicate the religious position ; C = conservative , L = liberal , and HI = historically inaccurate . c Letters in the parentheses to the right of each item indicate ideological positions supported or opposed by Catholic and/or Protestant conservatives ; C = Catholic , P = Protestant , and G = general . dA participant 's CRH-Quiz score was obtained by averaging incorrect responses across the 19 items . The range of possible scores was from 0 to 1 . TABLE 2 Items , Means , and Standard Deviations for the Christian Religious Orientation Scale ( CROS ) a PREFORMATTED TABLE <p> Note . Items 2 , 5 , 10 , 11 , 14 , and 17 were adapted from Fullerton and Hunsberger 's ( 1982 ) Unidimensional Scale of Christian Orthodoxy with permission from the Society for Scientific Study of Religion . a In the CROS , participants were asked to indicate their opinion on each statement by writing the number @ @ @ @ @ @ @ @ @ @ statement . b Letters in the parentheses to the left of each item indicate direction of agreement ; agreement with C = conservative , and agreement with L = liberal . c Mean scores for each item can range from 1 to 7. d The statistic r represents the correct item -- total correlation . TABLE 3 Items , Means , and Standard Deviations for the Perception of Historical Christian Conservatism Scale ( PHCCS ) PREFORMATTED TABLE <p> Note . The thee historically inaccurate items ( 4,11 , and 16 ) were included in the scale to disguise the purpose of the scale , but did not enter into the scoring . <p> a The PHCCS consisted of the same items as the CRH-Quiz with different instructions . Instead of asking participants to judge whether a statement was true or false , participants were asked to assume that the statements were historically accurate and to judge whether each item described a conservative or liberal position by circling a number on a 7-point scale . b Mean scores for each item can range from 1 to 7. c The statistic r represents @ @ @ @ @ @ @ @ @ @ <p> 
##4001168 ABSTRACT . The relationships among a cluster of attitudes toward work and job were investigated using a sample of 171 nurses . The hypothesis was that involvement in work and job , commitment to the employing organization , and satisfaction with the job ( overall , and with specific facets of the job ) would be significantly correlated . The results showed that involvement was not related to overall satisfaction but only to two specific facets , satisfaction with work and promotion opportunities . In contrast , the degree of relationship between overall and various facets of satisfaction and commitment and between involvement and commitment was moderately high . <p> Employee attitudes toward involvement in and satisfaction with the job and commitment to the employing organization have become of compelling interest to industrial psychologists because of their impact on behavior at work ( Robbins , 1993 ) . Job involvement frequently includes identifying with the job , actively participating in the job , and perceiving job performance to be important to self-worth ( Blau , 1985 ; Rabinowitz &; Hall , 1977 ) . Job satisfaction in the @ @ @ @ @ @ @ @ @ @ toward the job or toward specific dimensions of the job ( Hodson , 1991 ) . Organizational commitment refers to identification with and loyalty to the organization and its goals ( Blau &; Boal , 1987 ) . Employee attitudes are reflected in tendencies to respond to the job and the organization and its people and situations either positively or negatively . <p> Attitudes tend to cluster and categorize themselves . A person who has developed a favorable attitude toward one aspect of the job based on unique experiences , is likely to react favorably to other related job aspects . Thus , if one is involved in a job , one is likely to be satisfied with the job and committed to the organization . A person who is dissatisfied with a job may become less involved in the work and less committed to the employer . Although attempts have been made to introduce causality -- for instance , that satisfaction determines involvement ( Mortimer &; Lorerice , 1989 ) -- studies of discriminant validity suggest that these attitudes are relatively unrelated ( Mathieu &; Farr , 1991 ) . @ @ @ @ @ @ @ @ @ @ of these attitudes in conjunction with organizational outcomes . For example Farris ( 1971 ) , Horn , Katerberg , and Hulin ( 1979 ) , Mowday , Porter , and Steers ( 1982 ) , Blau and Boal ( 1989 ) , and Mathieu and Kohler ( 1990 ) have variously used involvement and commitment separately and interactively to predict tardiness , turnover , and absenteeism . Other researchers have linked job involvement to organizational factors ( Jans , 1985 ) and organizational commitment to influence ( Angle &; Perry , 1983 ) or work experiences ( Pierce &; Dunham , 1987 ) . However , the interrelationship of these attitudes has not been studied separately and exclusively . Only Moser and Schuler ( 1993 ) , in validating Lodahl and Kejner 's ( 1965 ) job involvement scale , predicted and found a high correlation among them . <p> In this research , I examined to what degree these attitudes were related . Generally , no causality was assumed : People may become involved in their jobs because they are satisfied with their jobs , or satisfaction may lead @ @ @ @ @ @ @ @ @ @ , or commitment may lead to involvement and satisfaction . However , it is also possible that " chunks " of one attitude may be related to chunks of other attitudes in many ways -- causally , acausally , or spuriously . This contention can be derived from contingency theories that predict that the strength and direction of any relationship among a set of variables is likely to vary with the individual involved and with the context within which the individual works . I hypothesized that these three job-related attitudes would be significantly related : involvement with satisfaction , satisfaction with commitment , and involvement with commitment . Method Participants <p> The participants were 171 nurse educators and registered nurses employed by 11 hospitals and 3 community colleges in southern Ontario . Two thirds of the nurses were between 35 and 45 years old . Most of them were married ( 72% ) . They had worked for their organization an average of 8 years . Instruments <p> Organizational commitment was measured with Mowday , Steers , and Porter 's ( 1979 ) Organizational Commitment Questionnaire . Two scales were used @ @ @ @ @ @ @ @ @ @ by Kanungo ( 1982 ) , which asks respondents to indicate the degree to which they are involved in their present job . The other asks respondents , on a 5-point scale , to what degree they are involved in their work ( as opposed to involvement with their families or with leisure activities ) . A short form of the Job Descriptive Index , developed by Hatfield , Robinson , and Huseman ( 1985 ) , was used to measure the various facets of job satisfaction . Finally , instead of using an additive version of the five facets of job satisfaction , a separate , one-item measure of overall job satisfaction was also used ( Iris &; Barrett , 1972 ) . Procedure <p> The data were collected by seven nurse educators enrolled in a graduate program in education . They distributed 245 instruments to all of the nurses employed in the hospitals and colleges surveyed . To show the relationships among the attitudes , simple Pearson correlations were calculated for all variables . The required level of significance ( p ) was set at < .01 . The @ @ @ @ @ @ @ @ @ @ analyses to assess their respective contributions to explained variance in commitment and involvement . F values were considered significant at p < .01 . Results <p> The means and standard deviations for the variables investigated were as follows : Organizational commitment , M = 24. 1 , SD = 5.0 : overall job satisfaction , M = 3.5 , SD = 1.0 ; satisfaction with work , M = 3.9 , SD = .7 ; satisfaction with pay , M = 3.1 , SD = .9 ; satisfaction with promotion , M = 3.3 , SD = .9 ; satisfaction with supervision , M = 3.6 , SD = .9 ; satisfaction with co-workers , M = 3.9 , SD = .7 ; involvement in work , M = 4.1 , SD = .9 ; and job involvement , M = 3.7 , SD = 1.0 . Involvement and Satisfaction <p> Simple correlations showed that neither job involvement nor involvement in work were statistically related to overall job satisfaction ( r = . 19 and .16 , respectively , ns ) . However , some facets of job satisfaction were significantly @ @ @ @ @ @ @ @ @ @ involvement . Job involvement was related to satisfaction with the work itself ( r = .33 , p < .001 ) and with opportunities for promotion ( r = .24 , p < .001 ) , but not to satisfaction with pay , supervision , or co-workers . Involvement with work was related to three of the five facets of satisfaction : work ( r = .29 , p < .001 ) , opportunities for promotion ( r = .33 , p < .001 ) , and supervision ( r = .25 , p < .001 ) . <p> Regression analyses indicated that facets of satisfaction are not powerful predictors of involvement . Concerning job involvement , only satisfaction with work entered the regression equation , explaining 11% of the variance . For involvement with work , two dimensions explained a total of 13% of the variance -- satisfaction with promotions ( 11% ) and satisfaction with work ( an additional 2% ) . Satisfaction and Commitment <p> In contrast to involvement and satisfaction , satisfaction and commitment were both statistically and meaningfully related . Commitment was related to overall job @ @ @ @ @ @ @ @ @ @ , to satisfaction with work ( r = .58 , p < .001 ) , pay ( r =.22 , p < .01 ) ; opportunities for promotion ( r = .55 , p < .001 ) , supervision ( r = .54 , p < .001 ) , and co-workers ( r = .30 , p < .001 ) . <p> When the facets of job satisfaction were subjected to a regression analysis , three facets contributed significantly to the variance in commitment . Satisfaction with the work itself explained the highest portion of the variance , 33% , followed by supervision , 13% , and promotion , 4% . Altogether , the three facets explained 50% of the variance in organizational commitment . Involvement and Commitment <p> Although involvement was not related to satisfaction , the concept was moderately related to commitment . There was a correlation between commitment and job involvement ( r = .32 , p = .001 ) . Commitment was correlated with involvement in work ( r = .38 , p = .001 ) . Involvement , Satisfaction , and Commitment <p> Each of the @ @ @ @ @ @ @ @ @ @ , with the two remaining attitudes as independent variables . The following pattern of results emerged . <p> When I treated commitment as the dependent variable and the various facets of job satisfaction and involvement as independent variables , only satisfaction but not involvement contributed to the variance in commitment . With commitment as the dependent variable and satisfaction and involvement in work as independent variables , the latter contributed only 1% additional variance to commitment in addition to the 50% contributed by satisfaction . <p> When commitment was predicted from overall job satisfaction and involvement in work , overall satisfaction contributed 41% and involvement in work an additional 7% , resulting in 48% of the overall explained variance . When overall job involvement was substituted for involvement with work , overall job satisfaction again explained 41% of the variance in commitment ; job involvement contributed an additional 4% , to result in 45% overall explained variance in commitment . <p> When I treated job satisfaction as the dependent variable and involvement and commitment as independent variables , commitment explained 41% of the variance in satisfaction , and job involvement did @ @ @ @ @ @ @ @ @ @ and commitment were regressed against the satisfaction variable , commitment also explained 41% of the variance , whereas involvement in work did not contribute to explained variance . <p> When job involvement was treated as the dependent variable , commitment explained 10% of the variance in involvement , whereas overall satisfaction did not contribute to the variance . In a separate analysis , when the facets of satisfaction and commitment were treated as independent variables , satisfaction with the work itself contributed 11% and commitment an additional 2% to the variance in involvement . <p> When the second measure , involvement with work , was substituted for overall job involvement , the pattern of variances did not change much . Commitment explained 14% of variance , whereas overall job satisfaction did not contribute to involvement with work . In the analysis of facets of satisfaction and commitment as independent variables , commitment again contributed 14% , whereas satisfaction with promotion added 2% , for a total of 16% of explained variance in involvement in work . Discussion <p> For this sample of nurses , three conclusions can be drawn from the @ @ @ @ @ @ @ @ @ @ degree of relationship between satisfaction and commitment was moderately high , between satisfaction and involvement was negligible , and between involvement and commitment was moderate . <p> Clearly , involvement in work or job was related to commitment to , and identification with , the employing organization , but not to satisfaction . The findings also suggest that involvement does not necessarily lead to satisfaction . Why is this the case ? The answer may have to do with the nature of the work . People may become enveloped and drawn into work that is difficult , troublesome , or dangerous , leading to a discharge of obligations rather than pleasure . Alternatively , as Hackman and Oldham ( 1980 ) suggested , the work may be perceived as repetitive and routine , lacking in variety or significance . In both cases , the nature of the work may still demand involvement but may not lead to satisfaction . <p> Another set of explanations pertains more to the personality of the person doing the work than to the work itself . People may not be satisfied with the job because the @ @ @ @ @ @ @ @ @ @ their needs , especially the higher-order needs , for instance , for autonomy , achievement , or competence . Or perhaps satisfaction of these needs does occur but leads to overinvolvement , which in turn results in low job satisfaction . Also , work may merely lead to the satisfaction of lower-order needs , like pay and security . " I am involved in my work because I have to feed the children and pay the rent , not because I like the work " may be a typical response . <p> Furthermore , different types of personalities ( Jung , 1971 ) may become involved and derive satisfaction in different ways . In short , it is equally likely that ( a ) people become involved and do derive satisfaction , ( b ) people become involved but do not derive satisfaction , ( 3 ) people do not become involved and do not derive satisfaction , or ( 4 ) people do not become involved yet do derive satisfaction ( with pay and/or co-workers , for example ) . <p> These reflections on aspects of job involvement are only @ @ @ @ @ @ @ @ @ @ , Rabinowitz and Hall 's ( 1977 ) , and Blau 's ( 1985 ) explanations of involvement . It is possible that the absence of these aspects in the definition and measurement of job involvement accounts for the elusiveness ( Knoop , 1986 ) or entanglement ( Mortimer &; Lorence , 1989 ) of the concept . It seems likely that people get involved in work or job for reasons other than identification , participation , or self-worth , the accepted ingredients of the definition of involvement . <p> Yet the necessity of becoming involved , in order to keep a job , for example , could also lead to commitment , especially if commitment was perceived differently . Instead of identification or loyalty , as Blau and Boal ( 1987 ) advocated , commitment could mean making a pledge or promise to the employer . An individual could well pledge to stay with an organization and to get involved in difficult work that brings little gratification to discharge an obligation . <p> The relationship between satisfaction and commitment is easier to explain . Satisfaction with the work itself in @ @ @ @ @ @ @ @ @ @ be related to commitment . These findings seem logical . When one likes one 's job or work and derives contentment from it , one can see oneself staying with it and with the organization offering this work . This finding implies that commitment can decrease if the job ceases to be challenging or interesting , either because the job gets changed or because the job holder gets bored with it . <p> The present study needs to be replicated in other work settings . Nurses may derive satisfaction and gratification from caring for patients , but they may not want to get too involved with their patients ' illnesses . Nurses may also be committed to their organization because they chose nursing as a profusion ; the particular hospital they are employed in may not mean as much as the profession itself . <p> 
##4001169 ABSTRACT . Conceptual level ( CL ) has been proposed as an important individual difference variable related to counselor training and effectiveness . However , relatively little attention has been paid to CL measurement issues in relation to counseling . Previous researchers have concluded that counselor CL is related to counselor training and counseling outcomes . A meta-analysis of 10 studies using counselors or counselor trainees as participants and outcomes based on counseling interviews showed the relationship between counselor CL and outcome to be inconsistent and weak . Although further research into CL and counseling appears warranted , such research will require valid domain-specific procedures for measuring counselor CL . <p> It has been accepted generally for some time that personality dispositions measured by conventional self-report personality inventories are not consistently related to counselor effectiveness or counselor training outcome ( Loesch , Crane , &; Rucker , 1978 ; Rowe , Murphy , &; DeCsipkes , 1975 ) . Counseling psychology researchers have investigated possible relationships between counselors ' performance and their characteristic global ways of conceptualizing . Harvey , Hunt , and Schroder ( 1961 ) defined @ @ @ @ @ @ @ @ @ @ provides the basis by which the individual relates to the environmental events he experiences " ( pp.244-245 ) . <p> Hunt ( 1970a , 1970b , 1971 ) proposed conceptual level ( CL ) as the major person variable in a person-environment interaction model of interpersonal functioning . He defined CL as " a personality characteristic that describes persons on a developmental hierarchy of increasing conceptual complexity , self-responsibility and independence " ( Hunt , 1978 , p. 78 ) . The most generally accepted implication of Hunt 's work is that high-CL individuals are more likely to benefit from those learning environments that encourage discovery learning , whereas low-CL individuals are more likely to benefit from learning environments that are relatively more structured and emphasize following the rules ( Miller , 1978 , 1981 ; Miller &; Wilson , 1979 ) . Several authors have discussed possible implications of Hunt 's proposals concerning CL in relation to counseling ( Berg &; Stone , 1980 ; Holloway &; Wolleat , 1980 ) . <p> Two general hypotheses related to counseling have been derived from CL theory . The first ( represented @ @ @ @ @ @ @ @ @ @ or counselor training setting is inherently ambiguous and poorly structured , high-CL counselors or counselor trainees will tend to perform better than their low-CL counterparts on average . The second general hypothesis ( represented by Type B studies ) is that high-CL counselors or trainees will perform relatively better in more ambiguous and unstructured situations , whereas low-CL counselors or trainees will perform relatively better in more structured situations . A corollary is that counseling will be more effective when counselor and client are matched on the basis of their CL ( high with high , low with low ) . Conceptualizing and Measuring Counselor CL <p> Miller ( 1981 ) described conceptual systems theory as " a personality theory that focuses on individual differences in social cognition within a developmental framework " ( p. 34 ) and proposed that the theory characterizes persons in terms of the structure and content of their conceptual systems . In most accounts , conceptual system complexity is identified as the key element , where high levels of complexity imply high levels of both cognitive differentiation and cognitive integration ( Miller &; Wilson , 1979 @ @ @ @ @ @ @ @ @ @ and cognitive integration go together , others ( notably Miller , 1981 ) have commented that the relationship between these two cognitive structure qualities is far from clear in that " high differentiation may actually hinder the development of integrative skills " ( p. 35 ) . <p> In a counseling interview , a counselor must not only be able to make differentiations and discriminations in relation to client behavior and experiences in the course of a counseling interview or series of interviews , but must also be selective so as to use the most relevant information and be able to integrate this information so as to respond appropriately to the client 's real needs . It is suggested that the methods of measuring counselor CL used most frequently emphasize the differentiation and discrimination aspect of CL but fail to address adequately the selection and integration of information aspect . <p> The most widely used procedure for estimating CL involves participants undertaking a paragraph- or sentence-completion task by writing personal responses to a series of topics ( Harvey , 1966 ; Hunt , Butler , Noy , &; Rosser , 1978 @ @ @ @ @ @ @ @ @ @ . Responses are rated for their level of conceptual abstractness , in which higher ratings are given for expression of several alternatives and the ability to hold multiple viewpoints ( Hunt , 1970a ) . This assessment procedure assumes that structural complexity necessarily implies structural integration . However , apparent complexity can be associated with disintegration , and the standard rating systems appear not to take this into account in estimating an individual 's apparent CL . <p> The second procedure for assessing CL is to use a role construct repertory grid technique , based on proposals by Bieri ( 1955 ) , to measure a participant 's level of cognitive complexity . Carr 's ( 1965 ) Interpersonal Discrimination Test requires participants to rate a set of significant other individuals using a list of bipolar adjective pairs ( or constructs ) . A ( relatively ) cognitively simple participant will apply the bipolar adjectives in much the same way to most of the individuals in the set , whereas a cognitively complex participant will apply the adjectives differently to individuals in the set . <p> Several quantitative procedures can be @ @ @ @ @ @ @ @ @ @ ; such as the percentage of variance in the individual x Adjectives rating grid accounted for by the first principal component of the grid : the smaller the percentage , the more cognitively complex is the individual . The major shortcoming with all such grid rating analysis procedures is that they fail to distinguish true cognitive complexity from cognitive disorganization -- or even random responding -- because all three processes will result in low correlations among the bipolar adjective pairs . <p> Both of these most frequently used approaches to the measurement of CL are therefore potentially inadequate . Each may be sensitive to the differentiation and discrimination components of CL but fail to tap the integration , or quality of organization , component implied in the CL construct . <p> A further shortcoming of the standard paragraph- or sentence-completion method for measuring CL in counseling research is that the topics involved are very general and have little overlap with the domain of interpersonal helping : for example , " When I am in doubt ... " or " What I think about rules ... " ( Harvey et al. , @ @ @ @ @ @ @ @ @ @ issue when he reported non-significant correlations between scores on the standard paragraph-completion measure of CL and scores on his Teacher Adaptability Index , a CL measure with teaching-specific content : <p> The most likely explanation for these low correlations is the very generalized , non-specific nature of the Conceptual Level ( CL ) measure . When a CL-type Index is put into a more domain-specific content , as the " Attitude to Teaching " measure , the relationship is considerably higher . ( p. 333 ) <p> This explanation implies that the paragraph- and sentence-completion methods typically used to measure CL as a global quality of counselors ' conceptual systems may be unable to show that CL is related strongly to performance in actual counseling interviews because the item content is related only tenuously to the specific interpersonal demands of the counseling situation . <p> In addition to these potential shortcomings in methods of measuring counselor CL , there is little evidence that the different approaches tap the same psychological processes . Whereas the Paragraph Completion Test ( Schroder et al. , 1967 ) and the Paragraph Completion Method ( Hunt @ @ @ @ @ @ @ @ @ @ gives results correlated with those of Harvey 's ( 1966 ) This I Believe Test ( Miller , 1978 ) . Vannoy ( 1965 ) found that the Paragraph Completion Test was unrelated to cognitive complexity scores derived from interpersonal bipolar adjective rating grids . This apparent lack of convergent validity among CL measures casts doubt on the comparability of research findings based on different approaches to CL measurement . <p> This discussion of the shortcomings in methods currently used to measure counselor CL suggests that , regardless of any potential merits of CL theory in relation to counseling and counselor training , problematic aspects of current methods of measuring counselor CL make it difficult for researchers to actually demonstrate a strong relationship between counselor CL and performance . Reviews of Research on CL and Counseling <p> Two reviews of research on counseling and CL have been reported ( Holloway &; Wampold , 1986 ; Stoppard &; Miller , 1985 ) , covering 28 published studies and revealing considerable diversity , making it difficult to draw general conclusions : Two characteristics on which studies differed seem particularly important . The first @ @ @ @ @ @ @ @ @ @ or counselor trainees ) , or whether it focused on other kinds of individuals , such as clients or research volunteers . Clearly , conclusions about counselor CL and performance based on data generated by individuals other than counselors are somewhat uncertain , given that differing motivational sets would probably be brought to the research task by participants other than counselors . <p> The second characteristic is whether the dependent , or outcome , variables used in a particular study were based on counseling interviews , or whether they involved some other analog procedure , such as responding to written or taped client statement vignettes . Several authors have drawn attention to the fact that noninterview-based measures of counselor performance are related tenuously , at best , to counseling interview effectiveness ( Ford , 1979 ; Matarazzo , 1978 ) . In addition , as Miller ( 1981 ) noted , CL differences are most likely to find expression in complex interpersonal behaviors rather than in relatively simple tasks such as responding to multiple choice counseling response tests or writing responses to written client statements . <p> Stoppard and Miller ( @ @ @ @ @ @ @ @ @ @ the question of matching individuals ' CL with the level of environmental structure ( Type g research studies ) . This was a traditional voting or box score review ; Stoppard and Miller concluded that results in 11 of the studies supported Hunt 's matching hypothesis , and they suggested that the findings warranted further research into matching clients and counselors on CL . However , of the 15 studies they examined , only 2 used counselors or trainee counselors as participants and relied on interviews as a basis for determining outcome . Of these 2 studies , I reported a significant positive outcome ( McLachlan , 1972 ) ; the other ( Rosenthal , 1977 ) did not . The other 13 studies used student research volunteers , or clients , or trainee teachers as participants and/or did not rely on outcomes involving counseling interviews . <p> Holloway and Wampold ( 1986 ) carried out a quantitative meta-analysis of 23 reports ( published prior to September 1983 ) that examined either the direct relationship between CL and counseling effectiveness ( Type A studies ) or the effect of matching participants @ @ @ @ @ @ @ @ @ @ . It should be noted that a Type B study will also typically provide Type A data in the form of a main effect for counselor CL in addition to the Type B result in the form of a counselor/client CL interaction effect . <p> Holloway and Wampold ( 1986 ) used standard meta-analytic procedures ( Glass , McGaw , &; Smith , 1981 ) to convert group mean differences or correlations into standard effect sizes ( ES ) and regression analyses to identify any relationships between the magnitude of effect sizes and other moderating variables such as type of participant and rated design adequacy . Holloway and Wampold reported a mean ES for Type A studies of 1.07 : that is , the outcome measure score of the average high-CL participant exceeded that of 86% of low-CL participants . However , there was no evidence of a corresponding main ( or Type A ) effect for CL in their Type B ( interaction effect ) studies : ES = 0.02 . <p> Holloway and Wampold ( 1986 ) explained these discrepant findings as perhaps stemming from the relatively poorer methodologies @ @ @ @ @ @ @ @ @ @ a mean interaction ES for matching participants ( high CL or low CL ) with environments ( low structure or high structure , respectively ) of 0.15 ; that is , the outcome measure score of the average participant in a matched condition exceeded that of 56% of participants in an unmatched condition . Holloway and Wampold concluded that <p> The findings of Type A studies imply that high-CL rather than low-CL individuals may be preferable as applicants to counselor training programs due to their superior performance on perceptual tasks in complex , ambiguous situations . Type B studies , by testing the effectiveness of using a matching model to increase the effectiveness of a person 's performance , provide support for the more prescriptive technological approach . It appears that low-CL persons only need to be in more structured conditions to benefit from training or counseling environment . ( p. 318 ) <p> In their discussion of their positive conclusions concerning the importance of counselor CL in relation to training and performance , Holloway and Wampold acknowledged that studies in their review were disparate in many ways and should best @ @ @ @ @ @ @ @ @ @ rather than with counseling solely . <p> I decided to re-examine the studies reviewed by Holloway and Wampold ( 1986 ) , placing my attention only on those that focused on counselor CL and used counseling interviews as a basis for determining counseling or training outcome . Reappraising Counselor CL and Performance <p> Of the 23 reports reviewed by Holloway and Wampold ( 1986 ) , 7 were not related directly to the outcome of training in counseling or to the effectiveness of counselors . Of the remaining 16 studies , 10 did not use performance in an interview situation to provide an outcome or performance measure . Thus , only 6 of 23 studies were concerned directly with the performance of counselors or counselor trainees in an interview situation . These studies are summarized in Table 1 . They include the two studies described previously by Stoppard and Miller ( 1985 ) : McLachlan ( 1972 ) and Rosenthal ( 1977 ) . <p> In view of the small number of studies that met the inclusion criteria , I decided to see whether more studies could be located . I @ @ @ @ @ @ @ @ @ @ supplemented by a hand search , to locate eight additional reports concerning counselor or counselor trainee CL and performance published up to December 1992 . Of these , four did not involve individuals conducting an interview . The remaining four studies ( using interviews ) are summarized in Table 1 . <p> Restricting consideration only to studies that focused on counselors or counselor trainees involved in some form of interview as the basis for assessing counselor performance or counselor training outcome resulted in inconsistent and equivocal evidence for a relationship between CL and counselor performance . Of the 10 studies listed in Table 1 , 5 showed that CL was related to counseling performance or counselor training outcome . Lutwak and Hennessy ( 1982 ) found that empathy level following training was related to trainees ' CL . MacLachlan ( 1976 ) claimed indirect evidence that clients may perceive high-CL counselors more positively than low-CL counselors : clients of low-CL counselors were more likely to discontinue prematurely . Mayadas and Duehn ( 1978 ) found that high-CL counselors performed better than low-CL counselors , and that CL was related to the @ @ @ @ @ @ @ @ @ @ found that group psychotherapy patients who were matched for CL level with therapist CL level showed greater average improvement than patients not matched with their therapist on CL level . O'Brien ( 1984 ) found evidence that counselor CL was related to client self-rated outcome . However , the other five studies reported no significant relationship between CL and counselor performance or training outcome . On a simple voting basis , these findings suggest that counselor CL is probably unrelated to performance , because the number of studies reporting significant positive findings equaled the number of studies reporting nonsignificant findings . <p> Reviews based on simply counting the proportion of studies returning significant results have been criticized on the grounds that such a narrow approach is insensitive to overall trends in findings ( Wolf , 1986 ) . Accordingly , I conducted a quantitative meta-analysis of the findings in the 10 studies contained in Table 1 . The procedures described by Holloway and Wampold ( 1986 ) were used to calculate effect sizes for both main effects due to counselor CL ( Type A ) and for interactions ( or matchings @ @ @ @ @ @ @ @ @ @ . In all cases , the Hedges ( 1982 ) Formula 4 correction was applied to produce unbiased effect sizes , correcting for small sample size , using the program EFFECT ( Omodei , 1993 ) to make the calculations . Following the recommendations of Berman , Miller , and Massman ( 1985 ) , in those studies that used multiple outcome measures , I derived an overall effect size by averaging the effect sizes calculated for each outcome measure . Failure to do so would have arbitrarily weighted the importance of studies according to the number of outcome measures used . <p> The overall average Type A ES ( for the two Type A and eight Type B studies combined ) was 0.59 . That is , the client outcome measure score of the average high-CL counselor exceeded that of 72% of low-CL counselors . The overall average Type B ( or matching ) ES was 0.31 . That is , the outcome measure score for the average client in a matched condition exceeded that of 62% of the clients in an unmatched condition . However , the overall average @ @ @ @ @ @ @ @ @ @ Lutwak and Hennessy ( 1982 ) and Mayadas and Duehn ( 1978 ) reported quite large effect sizes . <p> Lutwak and Hennessy ( 1982 ; Type A ES = 2.22 ) exercised no experimental control over the counseling interviews conducted by their participants , who simply recruited volunteers from among their acquaintances to serve as analog clients in informally arranged interviews . This may well have allowed uncontrolled motivational factors to inflate the relationship between CL and counseling performance . Mayadas and Duehn ( 1978 ; Type A ES = 2.9 , Type B ES = 1.7 ) relied simply on changes in the frequency of use of six counseling microskills by their counselor trainees as the outcome variables . <p> As Hill ( 1990 ) noted , there is no evidence that use of particular forms of counselor responses is necessarily associated with greater counselor effectiveness ; thus , using this type of outcome variable as a measure of counselor performance is questionable . If these two studies are dropped from the analysis , then the overall average Type A ( simple main effect ) ES falls to 0.09 @ @ @ @ @ @ @ @ @ @ matching effect ) ES falls to 0.19 . That is , using these more conservative estimates of overall effect size , the client outcome measure score of the average high-CL counselor exceeded that of 54% of low-CL counselors , and the outcome measure score for the average client in a matched condition exceeded that of 58% of clients in an unmatched condition . These are very small effect sizes ( Wolf , 1986 ) , and at this juncture it must be concluded that the proposed relationship between counselor CL and performance in counseling interviews is inconsistent and weak . <p> With only eight studies judged to be methodologically sound , it was difficult to explore what factors might be associated with relatively larger effects for CL . Six studies used actual counseling clients . The average main effect ( Type A ) ES for these was 0.09 , the average interaction effect ( Type B ) ES was 0.16 . Two studies used role-playing clients and the corresponding values were 0.11 and 0.29 , respectively . The overall average ES for the five studies that used the Paragraph Completion Test @ @ @ @ @ @ @ @ @ @ Completion Method ( Hunt et al , 1978 ) was greater ( Type A = 0.17 , Type B = 0.35 ) than that of the one study that used Harvey 's ( 1966 ) This I Believe Test ( MacLachlan , 1972 ) or the two studies that used combined measures ( Blaas &; Heck , 1978 ; Larimer , 1978 ) . Discussion <p> The merits and limitations of quantitative meta-analysis procedures have been discussed by several writers ( e.g. , Wolf , 1986 ) , and there is substantial agreement that such approaches are preferred generally to less sophisticated alternative review methods , such as simple enumeration of outcomes . There seems to be little doubt that the general quantitative meta-analytic strategy of including all studies relevant to the topic and relying on multiple regression techniques to establish the relative importance of various factors ( including methodological differences ) should be the method of choice when a large number of studies have been published that clearly involve the variables of central interest . <p> In the present case , however , inspection of the content of the 36 @ @ @ @ @ @ @ @ @ @ showed that only iO ( 28% ) actually focused on counselors or counselor trainees and used performance in an interview as the basis for assessing outcome . I judged that , with such a small number of studies fulfilling these important criteria , it was potentially misleading to aggregate them with the much greater number of studies that did not meet the criteria . <p> The 10 studies contained in Table 1 are quite disparate in detail . Apart from methodological weaknesses in the studies of Mayadas and Duehn ( 1978 ) and Lutwak and Hennessy ( 1982 ) , no feature could be identified confidently that distinguished those studies with larger effect sizes from those reporting smaller effect sizes . In light of the earlier discussion of shortcomings in existing CL measures in relation to the counseling setting , the finding of even a weak relationship between counselor CL and counselor performance , using a set of studies selected on the basis of very stringent inclusion criteria , suggests that further research into counselor CL and counseling is warranted . However , such research is unlikely to prove fruitful until @ @ @ @ @ @ @ @ @ @ In the absence of counseling-specific measures , counselor CL takes on the status of a static global personality disposition related only indirectly to the specific cognitive structures underlying those processes that mediate a counselor 's behaviors during a counseling interview . <p> There thus appears to be a need for the development of a measure of counselor cognitive structure that is content-specific to the counseling interview domain . This need could be addressed in several ways . First , it may be sufficient to modify existing approaches : that is , ( a ) to construct a paragraph- or sentence-completion task involving items whose content relates more directly to beliefs about counseling , or ( b ) to develop an interpersonal constructs rating grid with elements describing client characteristics and to devise scoring systems that address both the complexity and the integration components of CL . <p> Alternatively , it may be that new approaches to measuring counselor CL are required in order to determine empirically whether CL is an individual difference variable related strongly to counselor training outcome and counselor performance . In this regard , results of a study @ @ @ @ @ @ @ @ @ @ suggest that counselor cognitive complexity may not be a fixed traitlike dispositional characteristic that a counselor brings with him or her into every interview but may rather be a dynamic quality that depends to some extent on the nature of a particular counseling interview . Developing such a measure would entail using procedures that assess counselor CL in action rather than as a fixed personality characteristic . <p> In the meantime , there appears to be little empirical justification for using existing procedures as a basis for selecting counselor trainees , assigning trainees to more versus less structured training conditions , or matching clients with counselors , regardless of the attractiveness of CL theory for some counselor educators . TABLE 1 Summary of Studies of Counselor and Counseling Trainee Performance and Conceptual Level Using Interview-Based Outcomes PREFORMATTED TABLE <p> Note . Explanation of numbered entries follows : 1 . Number and type of clients interviewed . 2 . Number and type of counselors or counselor trainees . 3 . Focus of the study . 4 . Method of assessing CL. 5 . Outcome measure . BLRI = Barrett-Lennard Relationship Inventory ; @ @ @ @ @ @ @ @ @ @ : SCL = Symptom Check List . <p> a Type B studies arranged in descending order of interaction ES magnitude . b Averaged across all outcomes in the study ; the figure in parentheses shows the number of outcomes in the study . c Examined originally by Holloway and Wampold ( 1986 ) . <p> 
##4001170 ABSTRACT . Effects of adult attachment style on the perception of and search for social support were examined . One hundred and fifty undergraduate students completed self-report scales tapping their attachment styles , the extent to which they perceive the availability of emotional and instrumental support from significant figures ( father , mother , same-sex friend , opposite-sex friend , romantic partner ) , and the extent to which they look for instrumental and emotional assistance in times of need : Findings showed that secure persons perceived higher levels of emotional and instrumental support from the assessed figures and reported seeking more emotional and instrumental support than avoidant and ambivalent persons did . Results are discussed in the context of an attachment perspective of social support . <p> The contribution of social support to well-being has been one of the major areas of psychological research during the last decade . Social support is a multidimensional construct that may be defined as the comfort , assistance , and/or information one receives through formal or informal social contacts ( Wallston , Alagna , DeVellis , &; DeVellis , 1983 ) @ @ @ @ @ @ @ @ @ @ social support -- the generalized appraisal that one is cared for and valued -- is related to particular personality traits ( Lakey &; Heller , 1988 ; Sarason , Shearin , Pierce , &; Sarason , 1987 ) . The present study followed this idea and examined the relationship between one basic personality characteristic , attachment style , and a sense of social support . <p> Attachment theory ( Bowlby , 1969 , 1973 , 1980 ) can serve as a theoretical framework for studying how the sense of social support originates . In our terms , a sense of social support can be related to the concept of secure attachment . Bowlby defined secure relationships in infancy as those interactions in which parents are responsive to infants ' distress , assist infants in regulating tension , and bring relief and comfort . The responsiveness of parents to infants ' distress signals and their availability in stressful situations provide infants with a " secure base " and foster the sense of a " good-supportive world . " This relationship between attachment and social support has been also emphasized by Sarason , @ @ @ @ @ @ @ @ @ @ , the formation of attachment styles may be related to the development of the sense of social support . Securely attached children , who experience supportive relationships with attachment figures and use these figures as a " secure base " ( Ainsworth , Blehar , Waters , &; Wall , 1978 ) , may develop a strong sense of social support . In contrast , insecurely attached children ( either avoidant or ambivalent ) , who grow up with doubts about the extent to which attachment figures can comfort them in times of stress ( Ainsworth et al. , 1978 ) , may develop a generalized belief in a " non-supportive world . " <p> Early attachment experiences and the resulting sense of social support may have important implications for personality and social development . Bowlby ( 1973 ) claimed that the individual 's attachment style is a stable personality pattern and that it organizes cognition , emotion , and behavior throughout life . Several studies have provided evidence on the stability of attachment style ( Arend , Gove , &; Sroufe , 1979 ; Sroufe , 1983 ) and its @ @ @ @ @ @ @ @ @ @ &; Noller , 1990 ; Hazan &; Shaver , 1987 ; Mikulincer &; Erev , 1991 ) and cope with stress ( Collins &; Read , 1990 ; Kobak &; Sceery , 1988 ; Mikulincer , Florlan , &; Tolmacz , 1990 ) . <p> There is also evidence indicating an association between adult attachment style and social support . Kobak and Sceery ( 1988 ) found that secure persons reported having more available support from their families than avoidant persons did . Simpson , Rholes , and Nelligan ( 1992 ) found that secure women were more likely than avoidant women to seek emotional support from their partners before undergoing an anxiety-provoking experience in the laboratory . Mikulincer , Florian , and Weller ( 1993 ) found that secure Israeli adults reported seeking more social support than either avoidant or ambivalent adults did as a way of coping with the Iraqi SCUD missile attacks on Israeli cities . <p> The problem with those studies is that they did not address the multifaceted nature of social support . Specifically , they did not differentiate between different components of social support ( @ @ @ @ @ @ @ @ @ @ e.g. , parents , friends , lovers ) . <p> The perception of and search for social support are far from being unitary concepts ; rather , they appear to be differentiated along several dimensions . Cutrona ( 1990 ) differentiated between emotional support ( expression of comfort and caring ) and instrumental support ( provision of services , resources , and problem-solving means ) . Although high correlations have been found between these two components ( Brown , 1986 ; House , Kahn , McLeod , &; Williams , 1985 ) , some recent studies have demonstrated their existence via confirmatory factor analysis ( Brookings &; Bolton , 1988 ; Cutrona &; Russell , 1987 ; Vaux , Riedel , &; Stewart , 1987 ) . In addition , another differentiation has been made according to the identity of the supportive figure -- family members , friends , or romantic partners ( Cohen , 1988 ; Sarason et al. , 1987 ) . <p> In the current study , we examined the association of attachment style and self-reports on the perception of and search for social support . Undergraduate students @ @ @ @ @ @ @ @ @ @ to which they have and search for emotional and instrumental support from their mothers , fathers , same-sex friends , opposite-sex friends , and romantic partners . <p> We hypothesized that secure adults , who had experienced supportive relationships ( Shaver &; Hazan , 1993 ) , would be more likely to perceive available social support in times of need than avoidant and ambivalent adults , who had experienced rejecting or inconsistent relationships with attachment figures . In addition , both avoidant and ambivalent persons would be less prone to search for support than secure persons would , but their tendency would reflect different psychological reasons . On the one hand , avoidant persons would not search for support because they rely on themselves and maintain distance from others ( Bowlby , 1982 ) . On the other hand , ambivalent persons would not seek support because of their strong fear of rejection ( Mikulincer et al , 1990 ) . <p> We also hypothesized that the tendency of secure persons to perceive and seek more support than avoidant and ambivalent persons may be found mainly when assessing emotional components and @ @ @ @ @ @ @ @ @ @ support concerns experiences with parents who have been successful or unsuccessful in bringing emotional relief . However , if attachment style is generalized across interactions and domains ( Bowlby , 1973 ) , the effects of attachment styles may also be found with regard to instrumental support and extrafamilial interactions . Method Participants <p> One hundred and fifty undergraduate students ( 85 women and 65 men ranging in age from 21 to 28 years ) from the social science faculty at Bar-Ilan University volunteered to participate in the study without monetary reward . Of these participants , 85% were single . Materials and Procedure <p> The volunteers were asked to participate in a study on social relations , and they filled out the questionnaire during regular class time . They answered questions on scales tapping attachment and social support in random order . The procedure took approximately 30 min . <p> Attachment styles were assessed via two instruments previously used by Mikulincer et al . ( 1990 ) . First , participants read the three Hazan and Shaver descriptions of feelings and cognitions regarding attachment styles and endorsed the description that @ @ @ @ @ @ @ @ @ @ 15 statements ( five items per attachment style ) , constructed by decomposing Hazan and Shaver 's descriptions ( for more details , see Mikulincer et al. , 1990 ) . Cronbach alphas for the three attachment factors were acceptable ( from .75 to .84 ) . On that basis , we averaged items corresponding to each factor , compared those scores , and assigned each participant to the attachment style that had the maximal value of the three alternatives . <p> Only five mismatches resulted from comparing the results of the two techniques . No clear pattern was detected in these failures to coincide . To avoid classification ambiguities , we decided to drop those cases from the analyses . Frequencies of attachment styles in the current sample were similar to those in previous studies with American and Israeli samples ( Hazan &; Shaver , 1987 ; Mikulincer et al. , 1990 ) . Fifty-nine percent of the participants ( n = 86 ) were classified as secure , 30% as avoidant ( n = 43 ) , and 11% as ambivalent ( n = 16 ) . <p> Social @ @ @ @ @ @ @ @ @ @ study . The Perceived Available Instrumental and Emotional Support Scale ( PAIESS ) tapped the extent to which five figures ( mother , father , same-sex friend , opposite-sex friend , and romantic partner ) were perceived as providing instrumental and emotional support . The Seeking of Instrumental and Emotional Support Scale ( SIESS ) tapped the extent to which the respondent looked for instrumental and emotional support from those figures during difficult times . <p> The PAIESS , a 14-item , self-report questionnaire , was constructed on the basis of items selected and translated into Hebrew ( using a back-translation technique ) from the Social Support Questionnaire ( Sarason , Levine , Basham , &; Sarason , 1983 ) , the Social Support Perception Scale ( Vaux , 1985 ) , and the Social Support Behavior Scale ( Vaux et al. , 1987 ) . The PAIESS included seven items tapping instrumental support ( e.g. , " My father is ready to assist me financially when I need it " ) and seven items tapping emotional support ( e.g. , " My father is ready to listen to my innermost @ @ @ @ @ @ @ @ @ @ to which a particular person provided support on a 6-point scale ranging from 1 , not at all , to 6 , very much . They completed five versions of the PAIESS , each for a different figure . The order of the versions was randomized across participants . <p> Factor analyses with varimax rotation conducted separately for each PAIESS version yielded two main factors ( eigenvalue > 1 ) , which explained 65% to 73% of the variance . The first factor ( between 42% and 51% of explained variance ) included the seven instrumental support items ( loading > .40 ) . The second factor ( between 13% and 22% of the variance ) included the seven emotional support items . Cronbach 's alpha coefficients were appropriate ( between .83 and .89 for instrumental support items and between .79 and .92 for emotional support items ) . On that basis , we computed two scores of perceived instrumental support and perceived emotional support for each figure by averaging the items loading high on each factor . Higher score reflected more perceived support . <p> The SIESS , a 10-item @ @ @ @ @ @ @ @ @ @ items selected from the Hebrew version of the Ways of Coping Checklist ( Folkman &; Lazarus , 1980 ) . The SIESS included five items tapping instrumental support ( e.g. , " I ask for advice from my father about solving the problem " ) and 5 items tapping emotional support ( e.g. , " I seek sympathy and understanding from my father " ) . Participants rated the degree to which they seek support in difficult times from a particular person , on a 6-point scale ranging from 1 , not at all , to 6 , very much . They completed five versions of the SIESS , each for a different figure , as mentioned earlier . The order of the versions was randomized across participants . <p> Factor analyses with varimax rotation conducted separately for each SIESS version yielded two main factors that explained 68% to 78% of the variance . The first factor explained 57% and 71% of the variance , and it was composed of the five instrumental support items . The second factor explained between 7% and 11% of the variance , and it was @ @ @ @ @ @ @ @ @ @ alpha coefficients were appropriate ( between .76 and .85 for instrumental support items and between .84 and .90 for emotional support items ) . On this basis , we computed two scores for the search for instrumental and emotional support for each figure by averaging the items loading high on each factor . Higher scores reflected a higher tendency to seek support . Results <p> We analyzed the data by three-way analyses of variance ( ANOVAs ) for attachment style ( secure , avoidant , ambivalent ) , component of support ( instrumental , emotional ) , and identity of the figure ( father , mother , same-sex friend , opposite sex-friend , and partner ) . The two last factors were treated as within-subject repeated measures. 1 <p> The three-way ANOVA on the perception of available support yielded a significant main effect for attachment style , F(2,142) = 9.27 , p < .01 . Duncan post hoc tests ( alpha = .05 ) indicated that secure persons perceived more available support ( M = 5.01 ) than avoidant ( M = 4.57 ) and ambivalent persons ( M = 4.40 @ @ @ @ @ @ @ @ @ @ results for avoidant and ambivalent persons . The interactions of attachment style with the two other factors were not significant , implying that attachment groups differed in their perception of support availability regardless of the component of support endorsed and the identity of the support provider . Secure persons reported having more available support in all the categories assessed than insecure persons did ( see Table 1 for means and univariate F ratios ) . <p> The three-way ANOVA also yielded significant main effects for component of support , F(1,142) = 8.18 , p < .01 , and identity of the figure , F(4,559) = 13.82 , p < .01 . Participants perceived more emotional support ( M = 4.87 ) than instrumental support ( M = 4.76 ) . In addition , Duncan post hoc tests indicated that they reported that romantic partners provided more support ( M = 5.08 ) than did mothers ( M = 4.88 ) and same-sex friends ( M = 4.87 ) , who , in turn , were perceived as providing more support than opposite-sex friends ( M = 4.66 ) and fathers ( @ @ @ @ @ @ @ @ @ @ effects were qualified by a significant two-way interaction for component of support and identity of the figure , F(4,559) = 52.05 , p < .01 . Tests for simple main effects ( Winer , 1971 ) indicated that participants perceived their romantic partners and friends to provide more emotional than instrumental support . In contrast , they perceived their fathers to provide more instrumental than emotional support . Mothers were perceived as providing equal amounts of emotional and instrumental support ( see Table 1 ) . <p> The three-way ANOVA on the search for social support yielded a significant main effect for attachment style , F(9,142) = 8.57 , p < .01 . Duncan tests indicated that secure persons reported seeking more social support in dealing with life problems ( M = 4.15 ) than avoidant ( M = 3.66 ) and ambivalent persons ( M = 3.42 ) did . No significant difference was found between avoidant and ambivalent persons . Secure persons reported seeking more social support in most of the categories assessed than insecure persons ( see Table 2 , means and F ratios ) . <p> The @ @ @ @ @ @ @ @ @ @ , F(1,142) = 13.46 , p < .01 , and identity of the supportive figure , F(4,559) = 45.96 , p < .01 . Participants reported seeking more instrumental ( M = 4.01 ) than emotional support ( M = 3.85 ) . Duncan tests indicated that romantic partners were preferred as supportive figures ( M = 4.67 ) over same-sex friends ( M = 4.19 ) , who , in turn , were preferred over mothers and opposite-sex friends ( M = 3.75 , M = 3.71 , respectively ) . Fathers were the least preferred supportive figures ( M = 3.36 ) . <p> Those two main effects were qualified by a significant two-way interaction for component of support and identity of supportive figure , F(4,559) = 23.18 , p < .01 . Tests for simple main effects indicated that fathers were sought out for more instrumental than emotional support . Romantic partners , mothers , and friends were equally sought out for emotional and instrumental support ( see Table 2 ) . <p> To examine the association between the perception and seeking of social support , we computed @ @ @ @ @ @ @ @ @ @ Most of the correlations between the perception and seeking of support were relatively high -- between .50 and .69 -- in the total sample ( see Table 3 ) : A similar pattern was identified among both secure and avoidant persons . With regard to ambivalent persons , however , the pattern of correlations differed . We found significant associations between the perception and seeking of social support only for the emotional component of support . This pattern may imply that ambivalent persons did not perceive the availability of instrumental support as related to the extent to which they search for it . Discussion <p> In the present study we examined the association between adult attachment styles and sense of social support . In general , the findings support the hypothesis that attachment groups differ in the extent to which they perceive available emotional and instrumental support from significant others . In addition , attachment groups were found to differ in their tendency to seek support in times of need . <p> The pattern of social support shown by secure persons was in line with the predictions . Those persons have @ @ @ @ @ @ @ @ @ @ in times of need and were responsive to signals of distress ( Hazan &; Shaver , 1987 ) . This attachment history creates expectations that significant others will be available when needed and will be able to bring comfort and relief ( Hazan &; Shaver , 1987 ) . In the present study , these expectations are reflected in secure persons ' tendency to see significant others as providing relatively high levels of support . This trust in significant others may also underlie secure persons ' tendency to seek support . Because secure persons perceive others as efficient means for affect regulation , they would be prone to seek their company in times of need . <p> The pattern of social support shown by insecure persons , either avoidant or ambivalent , is also in line with our predictions . Insecure persons have a history of relationships with figures who were not responsive and available in times of need ( Hazan &; Shaver , 1987 ) . They grow up with worries about the intentions and responses of significant others . They may perceive the surrounding social world as a threatening @ @ @ @ @ @ @ @ @ @ social interactions for help coping with life 's adversities ( see , e.g. , Hazan &; Shaver , 1987 ) . This basic mistrust is reflected in insecure persons ' tendency to perceive a relatively low level of available instrumental and emotional support from others and their relatively low tendency to seek social support in times of need . <p> Ambivalent persons differed from secure persons not only in the degree of their perception and seeking of social support but also in the association between those two support-related constructs . Whereas secure persons showed a positive correlation between the perception and seeking of support , ambivalent persons showed low correlations for instrumental support . Possibly the search for instrumental support among ambivalent persons is guided more by their own basic anxieties than by a rational decision based on the availability of support . This interpretation is consistent with previous findings that ambivalent persons are not highly attuned to relevant information available in their social surroundings ( Mikulincer &; Nachshon , 1991 ) . <p> With regard to support seeking , our findings are in line with an earlier study ( Mikulincer @ @ @ @ @ @ @ @ @ @ have coped with the Iraqi missile attacks on Israeli cities . Our findings also provide partial support for the findings of Simpson et al . ( 1992 ) on support seeking within couples in an anxiety-provoking situation . Although both of those studies indicated that avoidant persons seek less support in times of need than secure persons do , only in the present study were significant differences found between ambivalent and secure attachment groups . A clear interpretation of this empirical discrepancy is problematic , because the two studies used entirely different methodologies . <p> The present findings were in line with the suggestion by Sarason et al . ( 1990 ) that the sense of social support has a personality substrata . Even though the objective presence or absence of supportive relationships may contribute to a sense of social support , the way a person appraises and reacts to them may also be determined by his or her personality . In this theoretical context , attachment style may be an important personality contributor to the sense of social support , but it can not be considered the exclusive or most @ @ @ @ @ @ @ @ @ @ related to attachment style ( e.g. , hardiness , social orientation ) , may also shape a person 's appraisal of , attitudes toward , and reactions to social support . <p> Our findings also show that attachment styles were similarly and significantly related to the emotional and instrumental components of social support that were provided by intra- and extrafamilial figures . These findings can be explained on the basis of the self-confirming property of cognitive schemata such as attachment styles , which generalize expectations beyond the original intrafamilial interaction . Through such a generalization , secure persons come to expect the receipt of support from the social word , whereas insecure persons have doubts about the responsiveness of that world . This reasoning suggests that differences in attachment style reflect general positive or negative orientations toward the social world . <p> The present study also delineates the particular structure of the sense of social support among young adults . It seems that the support provided by romantic partners and friends may be more appropriate to the needs of young adults than that provided by parents . Moreover , our sample @ @ @ @ @ @ @ @ @ @ the kind of support they could provide : Romantic partners and friends were perceived mainly as sources of emotional support , fathers were perceived as potential sources of instrumental support , and mothers were perceived as the providers of both instrumental and emotional types of support . <p> Our findings and their interpretation should be viewed as an initial step in the exploration of the complex association between personality and social support . Several possible limitations of the present study may reduce the theoretical generalizability of the findings . First , we did not examine differences among attachment groups in actual behaviors of seeking support in real-life situations . We also have no information about whether the relatively low perception of available social support among insecure persons reflects their subjective experiences or the presence of objective , problematic social interactions . We also did not examine what it is that the three attachment types seek from supportive figures and how they feel about the types of support they receive . However , the data provided here may further contribute to the integration of two theoretical frameworks that emphasize the importance of @ @ @ @ @ @ @ @ @ @ conducted with gender as another independent variable . No significant association was found between the distribution of attachment style and gender , and no significant interaction was found between them in their effects on the perception and seeking of support . However , the analyses revealed the frequently found main effect of gender on social support ( Vaux , 1985 ) : Women perceived and searched for more emotional and instrumental support than men did . TABLE 1 Means and Standard Deviations for Available Social Support According to Attachment Style PREFORMATTED TABLE <p> * p < .05. ** p < .01 . TABLE 2 Means and Standard Deviations for Search for Social Support According to Attachment Style PREFORMATTED TABLE <p> * p < .05. ** p < .01 . TABLE 3 Pearson Correlations Between the Perception of and Seeking of Support PREFORMATTED TABLE <p> ** p < .01 . <p> 