
##4001150 ABSTRACT . Sixty undergraduate subjects were exposed to one of three depressive mood inductions : cognitive , behavioral , or control . Subjects in the cognitive induction group read negative self-statements consistent with Beck 's ( 1974 ) cognitive theory of depression . Subjects in the behavioral induction group were exposed to insoluble discrimination problems consistent with Lewinsohn 's ( 1974 ) behavioral theory that lack of reinforcement produces depression . The control group read a neutral passage . Following the mood induction , subjects rated their current moods using the Multiple Affect Adjective Checklist ( MAACL ; Zuckerman &; Lubin , 1965 ) . The cognitive and behavioral inductions produced significantly higher MAACL total and subscale scores than did the control condition ; however , the cognitive and behavioral inductions did not differ from one another . Because the cognitive and behavioral mood inductions appear equally powerful , they may be used by researchers who are attempting to identify variables that are related to subjects ' differing responses to behavioral and cognitive influences on depression . <p> COGNITIVE AND BEHAVIORAL MODELS of depression have become increasingly influential @ @ @ @ @ @ @ @ @ @ ) cognitive theory and Lewinsohn 's ( 1974 ) behavioral theory have both generated substantial research on the cause and treatment of depression . To date , evaluations of cognitive and behavioral therapy for depression have shown that both treatments are effective but that there is no clear support for the greater efficacy of one over the other ( Sacco &; Beck , 1985 ) . Rehm , Kaslow , and Rabin ( 1987 ) attempted to identify variables that predict responses to cognitive versus behavior therapy . To date , however , there are no clear indicators as to which approach will be more effective for a client . <p> Rexford and Wierzbicki ( 1989 ) suggested that laboratory mood induction studies might help identify variables that predispose one to respond to cognitive versus behavioral therapy . They developed cognitive and behavioral mood inductions and found that both were effective in inducing negative mood but were unable to identify the variables that predicted different responses to cognitive and behavioral inductions . <p> One problem with Rexford and Wierzbicki 's study was that its behavioral induction had subjects read about , @ @ @ @ @ @ @ @ @ @ the behavioral induction could be regarded as cognitive . Rexford and Wierzbicki suggested that future researchers use behavioral inductions in which subjects actually experience negative events . <p> Hiroto and Seligman ( 1975 ) administered insoluble problems to subjects and observed that they exhibited decreased learning . Others found that exposure to insoluble problems increased depressed affect ( Sedek &; Kofta , 1990 ; Trice , 1982 , 1985 ) . <p> The purpose of this study was to compare the intensities of a cognitive and a behavioral mood induction . If the two types proved equally intense , researchers might then use them in efforts to identify variables that predict different responses to cognitive and behavioral therapy . Method <p> Subjects and Procedure <p> Subjects were 60 college students ( 49 women , 11 men ) in undergraduate psychology courses at Marquette University ; the students received extra course credit for volunteering to participate in research . <p> Subjects met individually with an experimenter , who assigned twenty subjects randomly to each of three mood induction groups : cognitive , behavioral , and control . The cognitive and behavioral induction @ @ @ @ @ @ @ @ @ @ 10.6 min ) . Subjects in the control condition read a passage of text for 12 min . After completing this procedure , subjects evaluated their moods , using the Multiple Affect Adjective Check List ( MAACL ; Zuckerman &; Lubin , 1965 ) . Subjects were then debriefed . Those who had been exposed to the depressive mood inductions were exposed to an elating induction designed to remove any remaining negative affect . <p> Materials <p> The behavioral mood induction consisted of a set of insoluble discrimination problems described by Hiroto and Seligman ( 1975 ) . Subjects were first administered one sample problem consisting of a series of five cards , each with two stimulus patterns . Each pattern contained five dichotomous dimensions : a letter ( X or Y ) , a color ( red or blue ) , a size ( large or small ) , a border ( square or circle ) , and a border line ( solid or dashed ) . <p> The subjects were told that each pattern contained only one value for each of the five dimensions and that 1 of the @ @ @ @ @ @ @ @ @ @ to guess which of the two patterns on each card contained the target value , after which they were told whether they were correct or incorrect . Using this feedback , the subjects tried to identify the target value for the problem . Subjects viewed each card for a maximum of 15 s ; after 10 s , subjects were told that they must make a selection in 5 s . <p> The subjects were then administered the sample problem , after which they were given the opportunity to ask questions about the procedure . We then administered three discrimination problems similar to the sample problem except that they consisted of 10 cards each and only four stimulus dimensions ( border line was omitted ) . Two other dimensions differed slightly from the sample problem : letter ( A or T ) and color ( black or white ) . <p> Feedback ( whether selections were correct or incorrect ) was provided on a predetermined schedule , independent of the subjects ' actual selections . After the last card in a problem was viewed , the subjects were asked to identify @ @ @ @ @ @ @ @ @ @ subjects were told that their answers were incorrect . <p> After the first two problems , the subjects were told that a new 10-card problem would be presented and that a new target value may or may not have been chosen . The same procedures were used for each of the three problems . Insoluble discrimination problems have been shown to decrease ability to solve later problems ( Hiroto &; Seligman , 1975 ) and to increase depressed mood ( Sedek &; Kofta , 1990 ; Trice , 1982 , 1985 ) . <p> The cognitive mood induction was modeled after Velten 's ( 1968 ) procedure in which subjects read 59 depressing self-statements . Rexford and Wierzbicki ( 1989 ) identified 59 statements of negative thoughts and thinking errors , consistent with Beck 's theory of depression , that were matched for severity with Velten 's statements . Velten 's instructions were used , with each step printed on an index card . After the instructions , subjects read a series of statements printed on index cards . Velten 's initial statement appeared first : " Today is neither better @ @ @ @ @ @ @ @ @ @ of the 59 statements used by Rexford and Wierzbicki were presented . The other 18 statements were omitted so that the two induction procedures would take the same time . An analysis of variance ( ANOVA ) showed that this subset of 41 and the original set of 59 statements did not differ in mean severity , F(1,98) = 1.77 , ns . <p> Subjects in the control condition read a passage on psychotherapy from an introductory psychology textbook ( Kalat , 1990 ) . We selected this passage because it had not been assigned to subjects during the semester and because we judged it to be neutral with respect to mood . Subjects were told that they would not be tested on this material , and they read for 12 min . <p> The MAACL ( Zuckerman &; Lubin , 1965 ) was administered after the mood induction or control condition . The MAACL consists of 132 adjectives that are endorsed if they describe how subjects currently feel ; it provides measures of depression , anxiety , and hostility and has been shown to be both a reliable ( Pankratz @ @ @ @ @ @ @ @ @ @ valid index of negative mood ( Nagata &; Trierweiler , 1988 ) . It has commonly been used to assess the effects of laboratory mood inductions ( e.g. , Nagata &; Trierweiler , 1988 ; Rexford &; Wierzbicki , 1989 ; Schare &; Lisman , 1984 ) . <p> Following debriefing , the subjects who had been exposed to the mood inductions were exposed to an elating mood induction . This consisted of reading 30 positive self-statements developed by Velten ( 1968 ) . These statements have been used to remove any remaining effects of depressive mood inductions ( Frost &; Green , 1982 ; Rexford &; Wierzbicki , 1989 ) . Results <p> We summed the MAACL subscale scores to produce the MAACL total score , which has been interpreted as an index of general negative affect ( Pankratz et al. , 1972 ; Turzo &; Range , 1991 ) . Using ANOVA , we determined that the MAACL total scores differed significantly across groups , F(2,57) = 7.30 , p < .01 . Scheffe tests showed that the cognitive and behavioral groups both differed significantly ( p < .05 @ @ @ @ @ @ @ @ @ @ and behavioral groups did not differ ( Table 1 ) . <p> A multivariate analysis of variance ( MANOVA ) using Wilks 's lambda criterion determined that MAACL subscale scores did not differ by gender , F(3,52) = 2.27 , ns , or in the interaction between gender and group , F(6,104) = 1.03 , ns . However , MAACL subscale scores did differ across groups , F(6,104) = 2.35 , p < .05 . <p> Univariate ANOVAs showed that the MAACL Anxiety , F(2,54) = 3.41 , p < .05 , and Hostility , F(2,54) = 5.05 , p < .01 , subscales differed across groups , although the MAACL Depression subscale did not , F(2,54) = 2.03 , ns . Planned comparisons showed that the behavioral and cognitive inductions did not differ from one another on the MAACL Depression ( t = 0.46 , ns ) , Anxiety ( t = 0.32 , ns ) , or Hostility ( t = 1.85 , ns ) subscales . However , the combined behavioral and cognitive inductions differed from the control condition on the Anxiety ( t = -2.44 , p @ @ @ @ @ @ @ @ @ @ p < .05 ) subscales , and almost differed significantly on the Depression subscale ( t = -1.81 , p < .10 ) . Table 1 contains the group means on the MAACL subscale scores . Discussion <p> This study compared the effects of a cognitive and a behavioral depressive mood induction . The two procedures did not differ in severity , but both produced significantly more negative mood than a control condition . <p> Exposure to a set of insoluble problems produced a significant increase in negative affect consistent with Lewinsohn 's ( 1974 ) behavioral model , which posits that low reinforcement leads to depression , and with previous studies ( Hiroto &; Seligman , 1975 ; Trice , 1982 , 1985 ) showing that exposure to insoluble problems increases negative affect . <p> The cognitive procedure also induced negative mood . This is consistent with Beck 's ( 1974 ) theory that depression is caused by cognitive factors , such as automatic negative thoughts and thinking errors . Many studies have now shown that Velten 's ( 1968 ) procedure induces depressed mood ( Larsen &; Sinnett , @ @ @ @ @ @ @ @ @ @ 's ( 1989 ) finding that a Velten-like induction consisting solely of statements consistent with Beck 's cognitive theory induces negative affect . <p> Previous research addressed how psychological variables moderate the effects of laboratory mood inductions . For example , responding to Velten-like inductions of depressed mood has been found to be related to personality characteristics , such as self-consciousness ( Scheier &; Carver , 1977 ) and neuroticism ( Hill , 1985 ) and to cognitive processes such as frequent automatic negative thoughts ( Blackburn , Cameron , &; Deary , 1990 ) . <p> Because the cognitive and behavioral inductions in this study were equally powerful , future studies may use both these procedures in attempts to identify variables that predict different responses to cognitive and behavioral therapy . In this way , laboratory mood induction studies may help to identify people who have different susceptibilities to cognitive and behavioral influences on depression and who may respond differently to cognitive and behavioral treatments of depression . <p> TABLE 1 <p> Mean MAACL Total and Subscale Scores Across Induction Groups PREFORMATTED TABLE <p> <p> 
##4001151 ABSTRACT . Ethical values of 171 college students at California State University , Chico , were measured , using a subset of the Rokeach ( 1968,1971 ) Value Survey . Non-parametric statistical analysis , four value measures , and four different consistent tests of significance and probability showed , surprisingly , that the younger students were more ethical than the older students . College students under 21 scored significantly higher ethically on three out of the four measures . Younger college students valued equality , freedom , and honesty more than their older classmates did . Surprisingly also , the younger students were significantly more concerned with being helpful and intellectual and were less involved in pursuing an exciting life and in social recognition than were the older students . <p> AGE AND ETHICS are generally believed to be related ( Kohlberg , 1984 ) . The common understanding that " older is wiser " means that age advances mental , emotional , social , and moral maturity ( Glenn , 1992 ) . Several recent studies report relationships between age and ethics . Serwinek ( 1992 ) @ @ @ @ @ @ @ @ @ @ did indeed explain variance in ethical viewpoints .. Burnett and Karson ( 1987 ) , Posner and Schmidt ( 1984 ) and Arlow ( 1991 ) found that older people become more conservative in their ethical viewpoints . Younger employees tend to have a more liberal view of potentially unethical situations ( Grant &; Broom , 1988 ) , whereas older workers tend to have somewhat more adamant opinions about what should be considered acceptable behavior ( Brenner , 1988 ) . <p> Older people have had more opportunity to see the consequences of unethical behavior ( Wood , Longenecker , McKinney , &; Moore , 1988 ) and are more likely to realize from experience that the unethical eventually suffer the consequences of their actions ( Miesling &; Preble , 1985 ) . Also , a longer , continuous exposure to tradition and custom appears to be a major explanation of why a person 's age is a significant predictor of ethical values ( Mudrack , 1989 ) . In short , age has been the most influential demographic predictor of ethical standards . As age increases , subjects have @ @ @ @ @ @ @ @ @ @ compromising interpretations of what is to be judged ethical . <p> For more than 25 years , the Rokeach Value Survey has been a commonly used research instrument to measure human beliefs , attitudes , and values ( Rokeach , 1968a , 1968b ) . Personal and organizational change have also frequently been measured using this instrument , and human values and value systems have been studied for more than 20 years using this methodology ( Rokeach , 1971 ; Sikula , 1972 ) . Some early value research has indirectly used the Rokeach Value Survey to measure religious values , but to the best of our knowledge , this study is one of the first attempts to define ethical values in terms of this survey . <p> The Rokeach Value Survey consists of two parts : 18 terminal ends and 18 instrumental means . Each set of 18 values is arranged in alphabetical order . Respondents are asked to arrange the 18 values in order of their importance as guiding principles of life from 1 ( most important ) to 18 ( least important ) . <p> Our purposes in @ @ @ @ @ @ @ @ @ @ and , using student subjects , to determine whether younger ( 21 or under ) and older ( 22 or over ) students have different ethical values . We defined ethical values in terms of the Rokeach Value Survey as a subset of the terminal ends and the instrumental means . Two terminal ends and two instrumental means constitute ethical values . Thus , the ethical values are equality , freedom , honesty , and responsibility . Method <p> Subjects <p> Subjects were 171 students in three different sections of the introductory , junior , and senior level Management and Organization course , all taught by the same professor at California State University , Chico , during the spring semester of 1992 . Each student filled out a Rokeach Value Survey about 1 month into the semester . Actually , 211 students filled out surveys , but 40 were not used for a variety of reasons pertaining to incomplete information . <p> The 171 students were mostly college juniors and seniors , although a few were sophomores or graduate students . The subjects were divided into two groups , 21 or @ @ @ @ @ @ @ @ @ @ because it is the legal age for voting ( and drinking ) in many states and because it split the subjects almost evenly in half ( 80 vs. 91 , respectively ) . California State University at Chico is a residential campus in a rural town and attracts largely traditional college-aged students . Although some students take more than 4 years to complete their degrees , most juniors and seniors are relatively young compared with students in urban communities where working people take part-time loads and evening classes . We ran data on age splits at 25 and 30 in addition to 21 , but the older samples were so small that we were not confident in the numbers . Of the total 171 students only 22 were over 25 and only 8 were over 30 . <p> The professor designed the course so that it contained a heavy dose of ethical and moral information , although management processes and functions remained the basic subject matter and format of the course . Topics included planning , organizing , controlling , leading , staffing , motivating , communicating , and decision making @ @ @ @ @ @ @ @ @ @ and international subject matter . The professor focused on each topic for about 1 week , and each topic contained at least some discussion of its ethical aspects and moral implications . <p> Statistical Analysis <p> Because nominal ranking scales instead of ordinal intervals are involved when using the Rokeach Value Survey , nonparametric statistics rather than parametric statistical analysis had to be used for comparing the two groups in this study . Several references are provided here for the reader interested in becoming more familiar with nonparametric statistical tests ( Conover , 1980 ; Hayek , 1969 ; Lehmann , 1975 ; Quade , 1966 ) . <p> To answer the questions raised in this study and to provide the appropriate statistical analysis , we used four nonparametric tests to determine the differences in ran.kings between younger and older students . These four nonparametric tests almost always gave the same conclusions regarding the significance of the findings and probabilities at the traditional level of .05 , and thus reinforced one another in terms of the reported results of this research . The following four nonparametric procedures and tests were used @ @ @ @ @ @ @ @ @ @ two-sample test with normal approximation and continuity correction Kruskal-Wallis test with a chi-square approximation Median two-sample test with normal approximation Median one-way analysis with a chi-square approximation Results <p> Table 1 contains the medians , ranks , and quartile deviations of the terminal ends for the 80 students aged 21 or younger and for the 91 students aged 22 or older . The value ranking differences for the two age groups indicated that five terminal values had a spread of five or more between the rankings of the younger and older students . An exciting life , a world at peace , equality , inner harmony , and social recognition had spreads of five or more between younger and older student rankings . <p> Table 2 contains the medians , ranks , and quartile deviations of the instrumental means for the 80 students aged 21 or younger and for the 91 students aged 22 or older ; these data indicate that six instrumental values had a spread of five or more between the rankings of the younger and older students . Cheerful , forgiving , helpful , intellectual , logical , @ @ @ @ @ @ @ @ @ @ and older student rankings . <p> First impressions are often inaccurate , however , and they are not scientifically determined . A quick look at the medians and quartile deviations reported in Tables 1 and 2 give a slightly different picture than that seen when simply comparing ranks . We also used the four companion tests of probability and significance to examine the frequency distributions of all the rankings by every student . Wilcoxon , Kruskal-Wallis , median two-sample , and median one-way probability tests were applied to all 36 Rokeach values . <p> Just as first impressions are right only about half the time , 6 of the 11 values that passed the visual inspection of the 5 or more rankings spread test failed the statistical analysis examination . To pass the probability benchmark , we required that all four probability tests of significance be in agreement , but we did stretch the traditional limit of .05 to .06 to capture two values . Two terminal ends , a world at peace and inner harmony , and four instrumental means , cheerful , forgiving , logical , and polite , @ @ @ @ @ @ @ @ @ @ two groups of students . Thus , only five values survived the probability standard : three terminal ends , an exciting life , equality , and social recognition , and two instrumental means , helpful and intellectual . <p> In a final analysis , we first investigated significance by examining spreads of five or more between younger and older student rankings of the terminal and instrumental values . At this point , the question was , Can any other of the 36 total values pass the probability criterion ( all four probability tests at .06 or greater significance ) even though they failed the observation benchmark ( five or more ranks between younger and older student rankings ) ? The answer was yes . An examination of the total data base showed that one more terminal end , freedom , and one more instrumental mean , honesty , passed the probability requirement . Although the ranking difference for freedom was only two , when the frequency distributions were tested for significance differences , the point .003 level was surprisingly obtained in all four tests . Honesty was also added as a @ @ @ @ @ @ @ @ @ @ although they both ranked it as number 1 . Table 2 reported a median difference of 3.0 for the two age groups on honesty . Discussion <p> For a finding to qualify as a conclusion in this study , we required that it pass the four-fold probability standard . Three out of the four ethical values identified earlier in this report qualified as conclusions because they were significantly valued differently by younger and older students , as determined by the probability standard . Along with past research and human intuition , this study showed that ethical values were significantly different between the junior and senior students . Surprisingly , however , the younger students placed a higher value on equality , freedom , and honesty than did the older students . Remembering that , in the Rokeach Value Survey , ranks with lower numbers reveal higher priority and importance , younger students ranked freedom number 3 ; older students ranked it fifth in order of importance . Younger students ranked equality number 8 ; older students ranked it 13th . Both student groups ranked honesty number 1 , but the median @ @ @ @ @ @ @ @ @ @ older students . <p> Our results showed that age and ethics are related . Historically , older people have been shown to be more ethical than younger people . The reverse ' was true in this study , perhaps because there was not enough difference in age between the two groups . Only 22 of the 171 student sample were over 25 ; only 8 were older than 30 . The older students were more often products of the 1980s ; the younger students were products of the 1990s . But for a number of reasons , we believe that perhaps the best explanation for our findings is that colleges and universities are doing a better job in the 1990s than in the 1980s in teaching ethical values and environmental concerns . Part of the explanation may also be youthful idealism . <p> With regard to nonethical values , two terminal values , an exciting life and social recognition , and two instrumental values , being helpful and intellectual , passed the fourfold probability test . Again , surprisingly , older students , by five ranks , valued an exciting life @ @ @ @ @ @ @ @ @ @ put less value on social recognition than did their elders ( by six ranks ) . Younger students also valued being helpful more than the students 4 to 7 years older ( by seven ranks ) and being intellectual ( by six ranks ) . Based on the results of this study , in which the age range was admittedly extremely narrow , younger college-aged baccalaureate students seem to be significantly more ethical than older college-aged undergraduate students . <p> TABLE 1 <p> Medians , Ranks , and Quartile Deviations for Two Age Groups PREFORMATTED TABLE <p> TABLE 2 <p> Medians , Ranks , and Quartile Deviations for Two Age Groups PREFORMATTED TABLE <p> <p> 
##4001152 ABSTRACT . Interpersonal dependency and locus of control orientation were studied in obese and nonobese samples . The Interpersonal Dependency Inventory ( Hirschfeld , Klerman , Gough , Barrett , &; Korchin , 1977 ) and the I-E Scale ( Rotter , 1966 ) were administered to 106 obese adults in outpatient treatment for obesity . The I-E Scale was also administered to 97 nonobese control subjects . As predicted , the obese subjects were significantly more internally oriented than has been previously identified in obese populations . These findings challenge currently held assumptions about locus of control in obese groups . Although the obese and nonobese subjects maintained similar overall dependency scores , the obese subjects were more likely to show low levels of autonomy in comparison with the nonobese subjects . Interpersonal dependency and locus of control dimensions associated with obese conditions may serve as useful predictor variables influencing obesity treatment approaches and outcome . <p> THE PERSONALITY DYNAMICS and psychological characteristics of obese individuals have been exhaustively researched ( Kaplan &; Kaplan , 1957 ; McReynolds , 1982 ; Slochower , 1987 ) . Although @ @ @ @ @ @ @ @ @ @ ( Brownell &; Foreyt , 1986 ) , in most contemporary paradigms researchers view obesity as multiply determined ( Kuldau , Rand , &; Tucker , 1982 ; Nash , 1987 ; Striegel-Moore &; Rodin , 1986 ) . In several investigations , researchers have examined psychological variables in obese individuals and have focused on personality correlates assumed to contribute to and maintain obesity . <p> Numerous research findings have associated personality characteristics of obese people with dependency , passivity , low assertiveness , self-consciousness , and low self-esteem ( Jacobs &; Wagner , 1984 ; Klesges , 1984 ; Mattlar , Salminen , &; Alanen , 1989 ; McReynolds , 1982 ; Stein , 1987 ) . Hirschfeld , Klerman , Gough , Barrett , and Korchin ( 1977 ) reported that interpersonal dependency is highly associated with the psychogenesis of addictive and psychiatric disorders . Hirschfeld et al . devised the Interpersonal Dependency Inventory ( IDI ) , which measures how much individuals need to associate with , interact with , and rely on others . As a personality construct , inter-personal dependency is conceptualized from the psychoanalytic theory @ @ @ @ @ @ @ @ @ @ and the ethological theory of attachment . To date , there have been no reported studies that have used the IDI with obese populations . I hypothesized that the obese sample in this study would score higher on all levels of dependency , as measured by the IDI , than the normative sample in the Hirschfeld et al . ( 1977 ) study . <p> The psychological construct of locus of control ( LOC ) orientation of obese populations has also been studied extensively . Locus of control , developed by Rotter , Seeman , and Liverant ( 1962 ) within the context of Rotter 's ( 1954 ) social learning theory , refers to how much control individuals believe they have over their lives . Researchers report that internal control orientation may be an important factor associated with successful weight reduction and the maintenance of weight loss ( Balch &; Ross , 1975 ; Cohen &; Alpert , 1978 ; Jeffrey , 1974 ; Kincey , 1983 ; Saltzer , 1982 ) . However , contradictory findings have indicated that control orientation is not a reliable predictor for successful weight @ @ @ @ @ @ @ @ @ @ MacDonald , 1977 ) . <p> In several studies that measured locus of control in obese populations , using Rotter 's ( 1966 ) Internal-External ( I-E ) Scale , researchers found that obese adult subjects ' mean scores were within the external range of control orientation ( Balch &; Ross , 1975 ; Gormanous &; Lowe , 1975 ; Tobias &; MacDonald , 1977 ) when compared with Rotter 's ( 1966 ) original normative samples . Although most research supports the notion that obese populations demonstrate an external control orientation , results from recent studies challenge this assumption. -1 ( Mills , 1991 a , 1991 b ) examined LOC in both male and female obese adult populations before the subjects entered an outpatient obesity treatment program and discovered that the subjects demonstrated an internal control orientation . Therefore , I further hypothesized that the obese subjects in this study would demonstrate an internal locus of control . Method <p> The sample consisted of 106 obese adults ( 93 women and 13 men ) and 97 adult controls ( 51 women and 46 men ) . The mean age @ @ @ @ @ @ @ @ @ @ 14.5 ) , and the mean age for the control group was 21 years ( SD = 2.7 ) . The obese subjects were involved in an intensive outpatient treatment program for obesity and were recruited from two medically supervised weight-loss programs located in large metropolitan cities in the Midwest . The control group was recruited from three introductory psychology classes from Lewis University . The obese subject pool consisted of moderately to morbidly obese adults who had been medically diagnosed according to clinical standards for obesity . The subjects were at least 50 lb over their ideal weight ( according to established normal height and weight tables ) when they entered the treatment program . The obesity treatment program consisted of comprehensive medical , psychological , behavioral , and nutritional interventions . Rotter 's ( 1966 ) I-E Scale and the Interpersonal Dependency Inventory ( Hirschfeld et al. , 1977 ) were administered in a standardized fashion to the obese subjects during the initial screening process before they entered treatment for obesity . The I-E Scale was also administered to the control group in a standardized group format . For @ @ @ @ @ @ @ @ @ @ participation was voluntary . Results <p> The LOC mean score for the total obese sample ( N = 106 ) on Rotter 's ( 1966 ) I-E Scale was 7.7 ( SD = 3.8 ) . The mean for the female obese subjects ( N = 93 ) was 7.8 ( SD = 3.8 ) , and the mean for the male obese subjects ( N = 13 ) was 7.3 ( SD = 4.0 ) . Results of a two-tailed t test that compared these groups were not statistically significant , t(104) = .46 , p > .05 . The results suggest that there were no differences in LOC orientation between the female and male obese subjects . The LOC mean score for the total control group ( N = 97 ) was 10.1 ( SD = 3.8 ) . The mean for the female control subjects ( N = 51 ) was 9.78 ( SD = 3.6 ) , and the mean for the male control subjects ( N = 46 ) was 10.22 ( SD = 3.9 ) . No statistical differences were found between the female and @ @ @ @ @ @ @ @ @ @ , p > .05 . However , the results of a two-tailed t test that compared the obese subjects ' and non-obese subjects ' LOC scores were statistically significant , t(201) = 4.33 , p < .001 . As predicted , the obese subjects demonstrated a significantly greater internal control orientation than the nonobese controls . The findings also support the notion that the obese adults in this study had higher degrees of internal LOC than has been previously recognized in obese populations . <p> The relationship between LOC and a reported history of family obesity was examined in the obese sample . The LOC mean score for the obese subjects who reported a history of obesity in their families ( N = 90 ) was 8.0 ( SD = 3.9 ) , whereas the mean LOC score for the obese subjects with no history of family obesity ( N = 16 ) was 6.1 ( SD = 3.2 ) . Results of a two-tailed t test that compared these two groups were statistically significant , t(104) = 2.15 , p < .05 . These results suggest that the obese subjects @ @ @ @ @ @ @ @ @ @ , and the obese subjects with no history of family obesity were very internally oriented . <p> The mean scores on the IDI were examined for the obese sample ( N = 66 ) and for the original normative sample ( N = 220 ; Hirschfeld et al. , 1977 ) . On Scale 1 of the IDI ( Emotional Reliance on Another Person ) , the-me , an score for the entire obese sample was 39.0 ( SD = 7.9 ) . The mean for the female obese subjects ( N = 58 ) was 38.9 ( SD = 8.2 ) , and the mean score for the male obese subjects ( N = 8 ) was 39.7 ( SD = 5.2 ) . Results of a two-tailed t test that compared the obese women and men were not statistically significant , t(64) = .28 , p > .05 . The mean scores of the obese subjects and the mean scores of the original normative sample showed no statistically significant differences on reported levels of emotional dependence on others , M = 39.2 ( SD = 7.8 ) , t(284) @ @ @ @ @ @ @ @ @ @ Scale 2 ( Lack of Social Self-Confidence ) , the mean score for the obese sample was 30.1 ( SD = 8.1 ) . The mean for the obese women was 30.6 ( SD = 8.4 ) , and the mean score for the obese men was 26.0 ( SD = 4.4 ) . A two-tailed t test that compared these groups revealed statistically significant results , t(64) = -2.41 , p < .05 . These results suggest that the obese women had lower levels of self-confidence in social situations than the obese men . A two-tailed t test that compared mean scores of the obese sample with mean scores of the normative sample ( M = 29.8 , SD = 6.4 ) showed no statistically significant results , t(284) = .27 , p > .05 . Although the obese men showed a slight trend toward more social self-confidence than the obese women overall there were no significant differences between the obese and the nonobese groups . <p> On IDI Scale 3 ( Assertion of Autonomy ) the mean score for the total obese sample was 28.1 ( SD = 6.31 @ @ @ @ @ @ @ @ @ @ ( SD = 5.88 ) , and the mean for the obese men was 33.2 ( SD = 7.36 ) . The results of a two-tailed t test that compared the female and male obese subjects were statistically significant , t(64) = 2.55 , p < .01 . The obese women seemed to be less autonomous than the obese men in this study . When the obese sample was compared with the control sample ( M = 30.2 , SD = 6.1 ) , there were statistically significant differences between mean scores , t(284) = -2.40 , p < .05 . Overall , the obese subjects in this study demonstrated less assertion of their independence and autonomous functioning than those in the nonobese control group . <p> On IDI Scale 4 ( Total Dependency ) , the mean score for the obese sample was 97.3 ( SD = 14.6 ) . The mean for the obese women was 97.0 ( SD = 15.2 ) , and the mean for the obese men were 99.0 ( SD = 9.6 ) . A two-tailed t test that compared these groups showed no statistically @ @ @ @ @ @ @ @ @ @ . There seemed to be no differences between the obese men and women for overall dependency . When the mean score for the obese sample was compared with the mean score for the nonobese controls ( M = 99.2 , SD = 9.2 ) , there was no statistically significant difference , t(284) .... 1.04 , p > .05 . There appeared to be no differences in total dependency between the obese sample and the control sample . <p> The mean scores on the IDI and the 1 E Scale for the obese , subjects who reported a history of alcoholism in their family of origin ( N = 19 ) were compared with the scores of the obese subjects who reported no history of family alcoholism ( N = 39 ) . The subjects with a history of family alcoholism ( M = 26.1 , SD = 4.4 ) revealed results that were significantly different on Scale 3 ( Assertion of Autonomy ) from the results for the subjects with no family alcoholism , M = 29.1 , SD = 7.3 , t(56) = 2.68 , p < .05 . @ @ @ @ @ @ @ @ @ @ t(56) = .35 , p > .05 ; Scale 2 , t(56) = .10 , p > .05 ; Scale 4 , t(56) = -.51 , p > .05 ; or locus of control , t(61) = 1.34 , p > .05 . <p> Pearson product-moment correlations were computed between scores on the I-E Scale and the four scales of the IDI . Locus of control scores were not significantly correlated with any of the factor scale scores on the IDI . However , the correlation coefficients between Total Dependency and Scale 1 ( Emotional Reliance on Another Person ; r = .73 , p < .001 ) , Scale 2 ( Lack of Social Self-Confidence ; r = .76 , p < .001 ) , and Scale 3 ( Assertion of Autonomy ; r = .40 , p < .001 ) were statistically significant . These results suggest that ( a ) the greater the obese subjects ' need for emotional reliance on others , ( b ) the lower their level of self-confidence in social situations , and ( c ) the less autonomous their behavior , the greater @ @ @ @ @ @ @ @ @ @ Scale I were significantly correlated with scores on Scale 2 ( r = .40 , p < .001 ) . This result suggests that the greater the subjects ' need for dependent attachment and reliance on others , the more they lacked self-confidence in social environments . Discussion <p> The results of this study suggest that both female and male obese groups are more internally oriented than has been previously reported . The obese subjects in this study demonstrated an internal locus of control ; however , most of the empirical findings in the obesity literature assert that obese populations have an external control orientation . Adults who seek treatment for obesity often report feeling that they do not have control over factors that influence the development and maintenance of their obesity . The sample in this study apparently perceived that generalized control over their lives was contingent on personal actions . In other words , the obese subjects felt in control of their lives despite feeling out of control over their weight status and food-related behaviors . This finding suggests that locus of control may be more circumscribed for obese @ @ @ @ @ @ @ @ @ @ of perceived control of and efficacy in specific food- and eating-related situations . <p> The obese subjects ' motivations for seeking treatment may also reflect their internal stance . Seeking professional help for their condition may be associated with a desire to assume control of and responsibility for their lives rather than accepting their fate or " biological " destiny . Obese individuals who are internally oriented may possess certain indigenous strengths that can be used as potential predictor variables and applied toward behavior change to influence their obesity . <p> The finding that the obese subjects with no history of family obesity were significantly more internally oriented than the obese subjects with a reported history of family obesity could possibly reflect internalized views toward food , eating , and weight status , conditioned eating patterns , parental identification , role modeling , and attitudes toward diet and nutrition . <p> Contrary to prediction , there were no significant differences between the obese sample and the nonobese control sample on overall dependency attributes . Apparently both groups demonstrated similar needs for approval and dependent attachment to others . These results raise @ @ @ @ @ @ @ @ @ @ needs than nonobese populations . Perhaps obese individuals modulate or fulfill various dependency needs via food and eating-related behaviors , whereas nonobese individuals find other avenues to satiate these inner yearnings and concerns . <p> The scores on the specific scales of the IDI did indicate that the obese subjects were less autonomous than the nonobese normative sample . The obese group apparently preferred to be with others rather than to be alone , and their self-esteem was contingent on approval and acceptance from others . This preference may have been due to the impact of obesity on self-esteem , the developmental capacity for individuation , thwarted levels of self-expression , and dependent rather than independent gratification . <p> The findings also suggest that the obese women were less autonomous and had lower levels of social self-confidence than the obese men . These gender differences may have been due to prevalent socialization practices that influence conditioned patterns of behavior and relatedness of men and women . In addition , the gender differences may reflect unique developmental differences in personality structure , internal needs , and dynamic conflict . <p> The obese @ @ @ @ @ @ @ @ @ @ assertion of autonomy than those subjects who had no such history . Individuals who are raised in alcoholic home environments often feel abused and deprived ( whether in reality or fantasy ) . These conditions could influence a heightened need for dependent attachment and approval from others . <p> Although locus of control was not correlated with interpersonal dependency in the obese sample , total dependency was correlated with all three dependency scales independently . Therefore , the greater the need for emotional reliance on another person , the greater the overall dependency . Lower levels of social self-confidence and low assertion of independent motives were also associated with higher levels of overall interpersonal dependency . In addition , the greater the subjects ' emotional reliance on another , the less self-confidence they demonstrated in social situations . <p> The obese sample in this study showed a trend ( although not statistically significant ) toward greater vulnerability on the dependency needs and conflict items ; the nonobese control sample showed less vulnerability . This trend suggests that food- and eating-related behaviors may have an important impact on dependency states in obese @ @ @ @ @ @ @ @ @ @ more likely to turn to food to meet specific dependency needs . Future research that examines dependency attributes in a larger clinically obese sample may provide more cogent evidence that supports a link between obesity and greater interpersonal dependency . <p> Regarding treatment , obese patients with relatively greater overall dependency needs , higher reliance on others , lower levels of social self-confidence , and fewer capacities for autonomous behavior are probably more likely to have difficulty with obesity treatment efforts than are patients with normal dependency needs . <p> <p> 
##4001153 ABSTRACT . The influence of five dimensions of teaching ( organization , knowledge acquired , encouragement of questions , instructor 's knowledge , and presentation clarity ) on 40 undergraduate students ' judgments of teaching effectiveness was investigated . The students rated completed evaluation forms ( each with a different configuration of responses ) as to the teaching effectiveness of an instructor who might receive that given pattern of evaluation responses . The students ' subjective reports of the dimensions they believed most influential ( subjective weights ) were compared with their statistically derived judgment " policies " ( statistical weights ) . Analyses revealed that there were significant differences between the mean statistical and subjective weights on two of the five evaluation items ( knowledge acquired and presentation clarity ) . <p> STUDENT EVALUATION of instruction has been examined in recent research that has included topics such as ( a ) validity of evaluation ratings as a measure of teaching effectiveness ( e.g. , Abrami , d'Apollonia , &; Cohen , 1990 ; Baird , 1987 ; Cranton &; Smith , 1990 ; Gigliotti &; Buchtel , @ @ @ @ @ @ @ @ @ @ instructor or course characteristics and evaluations ( e.g. , Drews , Burroughs , &; Nokovich , 1987 ; Marlin &; Gaynor , 1989 ; Murray , Rushton , &; Paunonen , 1990 ) , and ( c ) the relationship between evaluation ratings and the instrument or process of evaluation ( e.g. , Abbott , Wulff , Nyquist , Ropp , &; Hess , 1990 ) . Few studies , however , have satisfactorily examined the underlying judgment processes used by students who are asked to assess teaching effectiveness . It is well documented that humans are not always rational and consistent in the ways they make judgments ( Kahneman , 1991 ) . Consequently , an examination of the basic judgment processes involved in student evaluation of instruction may lead to an improved understanding and interpretation of the vast amounts of past research . <p> Whitely and Doyle ( 1976 ) , in their work on implicit theories of teaching , confronted the issue of student evaluations from a judgment process perspective . Implicit theories are the assumptions that students make about covariations in various teaching behaviors . If , @ @ @ @ @ @ @ @ @ @ and Y are positively correlated , that student will rate an instructor high on behavior Y after having observed only behavior X. Students are thought to systematically apply their idiosyncratic theories across various evaluation situations ( Larson , 1979 ) . That is , students ' implicit theories are hypothesized to remain constant over time and to systematically influence instructor or course evaluations . Unanswered questions remain , however , regarding ( a ) whether students actually apply their implicit theories consistently and ( b ) whether students can accurately articulate their implicit theories . In a broader sense , the question becomes one of whether students have accurate insight into the factors that influence their judgments of teaching effectiveness . <p> In the present study , we examined whether student judgments of teaching effectiveness , based on the presence or absence of certain course dimensions , are consistent with postjudgment self-reported estimates of the relative influence of the dimension variations on the judgments . To investigate this issue , we applied an empirical judgment paradigm known as policy capturing ( Hammond , 1954 , 1955 ) . In this paradigm @ @ @ @ @ @ @ @ @ @ materials ( " profiles " ) . Within each profile , levels of the various cues are manipulated ( e.g. , present or absent ) , and subjects are requested to provide a judgment ( e.g. , 1 to 5 rating ) on each of the profiles . The ratings are then regressed on the cues ( coded , for example , as 0 = cue absent and 1 = cue present ) , revealing the objective weight each subject applied to each cue while making the judgments . The statistical determination of the cue weightings for each subject is the process of capturing the subject 's judgment policy . <p> Past policy-capturing research has demonstrated that an individual 's subjective assessment of his or her own judgment policy ( i.e. , the major factors that subjects report as having contributed to their judgment ) is often at odds with the statistical ( i.e. , captured ) assessment of the same judgment . Such a phenomenon has been reported with judgments of clinical psychologists and diagnostic assessments of hyperactive children ( Ullman &; Doherty , 1984 ) , and with the @ @ @ @ @ @ @ @ @ @ Kozloski , &; Hults , 1991 ) . The current research assessed the relationship between statistical and subjective weightings in evaluations of teaching behavior . Method <p> Subjects <p> The subjects were 40 students ( 13 male , 25 female , and 2 not indicated ) enrolled in introductory psychology courses at a large southeastern university ( median age , 19 years ) . The subjects receive . d course credit in exchange for their participation in the research . <p> Stimulus Materials <p> The stimulus materials for the study were a series of profiles , each consisting of a teaching evaluation form containing five statements : ( a ) " Each class period was well-organized. " ( b ) " The course significantly increased my knowledge or competence in the subject . " ( c ) " The instructor encouraged students to ask questions and express opinions , " ( d ) " The instructor seems knowledgeable and well-informed on the subject . " ( e ) " The instructor presented the material clearly . " Each statement had response options of yes or no . These five questions were @ @ @ @ @ @ @ @ @ @ 1979 ) . For this study , each of the evaluation forms had been ostensibly completed ( yes or no circled for each of the five items ) by a " student " prior to the study . On each evaluation page the subjects were asked to respond to the statement " If a professor received the following course evaluation from a student ... How would you rate the quality of the professor as a teacher ? " The subjects responded on a 5-point scale ranging from very poor ( 1 ) to very good ( 5 ) . A sample profile is provided in the Appendix . <p> The stimulus materials were constructed so that all subjects provided judgments on all possible combinations of cue-level arrangements . In other words , the subjects judged all possible combinations of yes and no responses on all five evaluation form items . Given that each of the five items on the evaluation form had two response options ( yes or no ) , 32 yes no response configurations were possible . Eight duplicate profiles were developed to assess the reliability of the subjects @ @ @ @ @ @ @ @ @ @ order and constructed into stimulus booklets with a cover page of instructions . <p> Procedure <p> The subjects participated in the study in groups of 5 to 10 . They were each given a stimulus booklet , and they worked through the 40 judgment tasks at their own pace , assessing each evaluation form individually and providing ratings on the 1 to 5 scale . The subjects were instructed not to view their previously judged profiles once they started the procedure . Their task was not to complete the course evaluation form but to evaluate 40 completed course evaluations and provide judgments of the teaching effectiveness of a professor who might receive a given pattern of responses from a student on an evaluation form . <p> When they had completed the judgment tasks , the subjects provided ( in the form of a percentage ) their own estimates of the weights given to each of the five evaluation-form items in making their judgments . Specifically , the instructions read : " You have now completed all the ratings , but we would like you to provide us with some additional information @ @ @ @ @ @ @ @ @ @ us how much ' weight ' you gave each of the items in making your ratings . In other words , how much attention did you pay to each question on the evaluation when deciding how to rate each professor ? ( NOTE : Your total percentage should sum to 100% ) . " Thus each subject allocated percentage points according to how much each of the five items influenced the previous judgments . Results <p> Reliability Analysis <p> To calculate reliability estimates for each subject , we correlated the ratings on the eight duplicate profiles ( Presentation 1 and Presentation 2 ) . Reliability estimates ranged from 1.0 to 0.41 , with an average reliability of 0.86 ( SD = .13 ) . This analysis demonstrates that the subjects generally maintained a consistent judgment policy throughout the series of profiles . <p> Determination of Statistical Weightings <p> The yes or no responses ( coded 0 =- no and 1 = yes ) were used as predictors in multiple regression analyses of each subject 's 32 judgments ( 1 to 5 ratings ) of teaching effectiveness ( the second presentation of @ @ @ @ @ @ @ @ @ @ ensure cue orthogonality ) . The resulting regression equation was considered the subject 's judgment policy . For each subject , squaring the beta weights associated with each evaluation item provided a measure of the proportion of the variance accounted for in that subject 's judgments by variations in the given evaluation items . <p> Assessment of Statistical and Subjective Weightings <p> To be better able to interpret a comparison of the subjects ' statistical and subjective weightings , we followed a procedure similar to that used by Ullman and Doherty ( 1984 ) and normalized the squared beta weights . The normalization procedure involved , for each subject 's data , dividing each weight by the total sum of the weights and multiplying by 100 ( thus , each subject 's statistical weights summed to 100 ) . <p> The means and standard deviations were calculated across s , 11 subjects for the statistical and subjective weightings of each evaluation item ( see Table 1 ) . Data for five of the subjects were not included in this analysis , one because of an error in the stimulus materials and @ @ @ @ @ @ @ @ @ @ to 100% ( a copy of all subjects ' individual statistical and subjective weightings is available from the first author ) . Five t tests for paired samples were performed to compare the statistical and subjective weight means for each evaluation item . Results revealed that the weightings did not differ significantly on Items 1 , 3 , and 4 ( see Table 1 ) . However , on Items 2 and 5 , significant differences were revealed between the statistical and subjective weight means . ( Because multiple t tests were performed , a Bonferroni correction was applied to control for experiment wise error rate . This procedure involves dividing the standard accepted level of significance .05 by the number of t tests to determine the required significance level . Thus , in the present case , a given t statistic needed to reach the .01 significance level .05/5 to be considered significant . ) Discussion <p> This study was designed to investigate the relationship between statistical and subjective weightings of judgments of teaching effectiveness . Analyses revealed no significant differences for the following evaluation items : class was well @ @ @ @ @ @ @ @ @ @ . The students in our sample evidently had a reasonably accurate degree of insight into some of the factors that determined their judgments . Yet , the statistical and subjective weighting means differed significantly for the remaining two items ( course increased my knowledge and material presented clearly ) ; thus , student insight is certainly not highly accurate . The finding that the students subjectively overestimated the impact of the item material presented clearly can perhaps be understood from the perspective that lecture presentation may be one of the most salient and memorable features of a course . Thus , when our subjects reported ( subjectively ) the factors they believed influenced their ratings , they relied on the most available information in memory ( i.e. , use of the availability heuristic ) . <p> Perhaps the most interesting result of our analyses of the statistical weightings was that the item that stated " The course significantly increased my knowledge or competence in the subject " was the most influential ( highest mean ) in the assessment of teaching quality . Unfortunately , however , the subjects ( subjectively ) @ @ @ @ @ @ @ @ @ @ . This finding is indeed encouraging ; Evidently these subjects did feel ( at least as expressed through their judgments ) that the amount learned in a course was a crucial variable to consider when they evaluated teaching effectiveness . The challenge is to make students realize this fact at the subjective or conscious level . <p> Limitations of the Current Research <p> Because of the artificial nature of the task in the present study , sometimes referred to as a " paper people " paradigm , we are constrained in generalizing the findings . But as Gorman , Clover , and Doherty ( 1978 ) noted , the use of a policy-capturing strategy allows for ( a ) a high degree of control and ( b ) manipulation of factors that are difficult to otherwise regulate . In addition , the paper people approach allows for the presentation of identical , standardized stimulus materials across subjects . Thus , although the policy-capturing approach lacks a degree of representativeness , it provides a great deal of control in the research environment . In fact , Ullman and Doherty ( 1984 ) @ @ @ @ @ @ @ @ @ @ setting should be more reliable and predictable than in the real world . <p> Future Directions <p> The current research was an initial attempt to examine the judgment processes of students who are asked to evaluate courses or instructors . Future studies should relate individual students ' performances on a policy-capturing task with evaluations of an actual course . The crucial question for researchers is whether accuracy in assessing one 's own judgments is related to accuracy or validity in the actual judgment of teaching effectiveness . Research that uses the same methodology should also be undertaken with different samples of students and different evaluation items to replicate the current findings . An interesting extension of the current research would be an assessment of faculty judgment policies of evaluation forms . A comparison of student and faculty judgment policies at the same institution might be an enlightening and informative venture . <p> TABLE 1 <p> Mean Percentages , Standard Deviations , and t-Test Results for Statistical and Subjective Weightings PREFORMATTED TABLE <p> APPENDIX <p> Sample Evaluation Profile <p> IF A PROFESSOR RECEIVED THE FOLLOWING COURSE EVALUATION FROM A STUDENT ... PREFORMATTED @ @ @ @ @ @ @ @ @ @ PROFESSOR AS A TEACHER ? PREFORMATTED TABLE <p> <p> 
##4001154 ABSTRACT . To test Herzberg 's ( 1966,1987 ; Herzberg , Mausner , &; Snyderman , 1959 ) motivator-hygiene theory , I examined the relationship between work values and job satisfaction . Educators ( N = 386 ) from 18 Canadian secondary schools were asked to report the degree to which they experienced 16 work values and five dimensions of job satisfaction . Factor analysis suggested five sets of work values : intrinsic work-related , intrinsic work-outcome , extrinsic job-related , extrinsic job-outcome , and extrinsic people-related . Regression analyses identified the best predictors for each of the five dimensions of job satisfaction . The results support and extend Herzberg 's theory . In addition to those identified by Herzberg , other factors contributed portions of variance to the five dimensions of job satisfaction . <p> WORK VALUES denote the degree of worth , importance , and desirability of what happens at work ( Knoop , 1991 ) ; job satisfaction indicates an individual 's general attitude toward the job ( Brooke , Russell , &; Price , 1988 ) . Valuation is a qualitative act , an @ @ @ @ @ @ @ @ @ @ . Satisfaction refers more to gratification of needs and wants . An individual 's judgment of what takes place at work would be expected to influence his or her satisfaction with the job . <p> The seemingly different yet theoretically connected conceptions of work values and job satisfaction are reflected in process and content theories of job satisfaction ( Locke , 1976 ) . Process theories are attempts to specify values considered causally relevant to job satisfaction . Content theories are focused on identifying values conducive to , but not necessarily causal to , job satisfaction . <p> Herzberg ( 1966,1987 ; Herzberg et al . 1959 ) , in his well-known , much criticized , yet still very relevant motivator-hygiene theory , proposed several characteristics consistently related to job satisfaction and dissatisfaction . Intrinsic factors -- achievement , recognition , work itself , and responsibility -- were recurrently mentioned by respondents as sources of satisfaction . Extrinsic factors -- company policy , supervision , salary , relationship with peers , status , and security -- were frequently named as causes of job dissatisfaction . Herzberg concluded that job satisfaction and @ @ @ @ @ @ @ @ @ @ promote growth needs ; dissatisfaction depends on hygiene factors that serve lower-order needs . <p> Herzberg 's theory has been criticized for purporting to explain job satisfaction , rather than work motivation , without really measuring job satisfaction ( Caston &; Braito , 1985 ) . The present research tests Herzberg 's theory by measuring work values and job satisfaction separately and examining them concurrently . Two questions were asked : ( a ) How separate are work values and job satisfaction ? ( b ) Which set of values best predicts different aspects of job satisfaction ? Method <p> Subjects <p> The subjects were 386 volunteers from secondary schools : teachers ( N = 245 ) , department heads ( N = 100 ) , and principals ( N = 41 ) from five school boards in a large metropolitan area of Canada . Most of the subjects were married ( 278 ; 72% ) and held a bachelor 's degree ( 264 ; 68% ) . More than half were women ( 197 ; 51% ) . The average age was 43 years ; the average tenure was 12 @ @ @ @ @ @ @ @ @ @ Hatfield , Robinson , and Huseman 's ( 198 ' 5 ) Job Perception-Scale . The 20-item measure , modeled after Smith , Kendall , and Hulin 's ( 1969 ) Job Descriptive Index , was used to assess five facets of job satisfaction : work itself , pay , opportunities for promotion , supervision , and coworkers . The 16 work values were derived from Elizur 's ( 1984 ) research . The respondents were asked to what degree they currently experienced each of these values in their job . A 5-point scale was used to measure both sets of responses . <p> The questionnaires were distributed by part-time graduate students to principals , department heads , and teachers in 18 schools . Most of the administrators ( 92% ) , a majority of the department heads ( 64% ) , and nearly half the teachers ( 46% ) who received questionnaires completed them . Results <p> Factor Analyses <p> To examine whether the reported job satisfaction dimensions were distinct from the work values , I performed two factor analyses . In the first method , which Cranton and Smith @ @ @ @ @ @ @ @ @ @ as units of analysis . In the second method , called idiosyncratic , I used deviations from the means as the unit of analysis . The first method relies on overall averages , whereas the second method takes into account individual differences in the scores . Both methods led to similar factor structures but explained different amounts of variance . In each case , the 21 variables emerged into five fairly distinct factors . However , the variance for the normative method was 34% , whereas the variance for the idiosyncratic method was 59% . <p> Two factors contained intrinsic values toward work ( see Table 1 ) . The first factor included seven work values and one satisfaction dimension : exercising responsibility , achievement through work , influence over one 's work , doing meaningful work , being able to use one 's abilities and knowledge , independence in doing one 's work , contributing to society , and job satisfaction with the work itself . This factor is called Intrinsic Work-Related Values because the variables seem to stem from the content and worth of the work itself . <p> @ @ @ @ @ @ @ @ @ @ job status , recognition , influence in the organization , and pride in working in the organization . These values together are called Intrinsic Work-Outcome Values because they seem to be manifestations resulting from work . <p> Two other reported factors consisted of extrinsic values directed toward the job , not the work . The third factor contained two values and one job satisfaction dimension : benefits , job security , and satisfaction with pay . This factor was therefore called Extrinsic Job-Outcome Values . The fourth factor was made up of only two values : hours of work and working conditions . It was called Extrinsic Job-Related Values . <p> The fifth factor consisted solely of three dimensions of satisfaction : opportunities for promotion , supervision , and co-workers . The other two dimensions , satisfaction with work ( Factor 1 ) and pay ( Factor 3 ) , loaded partially on this factor ( .44 and . 17 , respectively ) . The fifth factor thus represents aspects of job satisfaction , because it does not contain any of Elizur 's values . This factor was called Extrinsic People-Related @ @ @ @ @ @ @ @ @ @ element is present in this factor and absent in the other factors . <p> Correlation Analyses <p> To discover which values best predicted each of the five dimensions of job satisfaction , I performed simple correlation and regression analyses . Pearson correlations indicated that each of the five dimensions of job satisfaction was statistically , but not always meaningfully , correlated with most of the work values . For satisfaction with the work itself , correlations ranged from .54 for achievement to . 12 for benefits or security . For satisfaction with pay , correlations ranged from .36 for status and benefits to .06 for pride in the organization . For satisfaction with opportunities for promotion , correlations ranged from .38 for recognition and .34 for influence over work and the organization to .07 for security . For satisfaction with supervision , correlations ranged from .31 for pride in the organization and .31 for esteem or achievement to .07 for security . Last , for satisfaction with co-workers , correlations ranged from .27 for influence in the organization to .03 for benefits . Overall , of the 80 relationships , 64 @ @ @ @ @ @ @ @ @ @ Regression analyses identified the best predictors of the five dimensions of job satisfaction . For satisfaction with work , four predictors entered the regression equation at a significant level ( p = .05 ) : achievement ( R 2 = .30 ) , use of abilities ( R 2 = .07 ) , pride in the organization ( R 2 = .02 ) , and meaningfulness of work ( R 2 = .02 ) . The total variance explained by the four predictors was .41 . <p> For satisfaction with pay , five work values were identified as significant predictors : job status ( R 2 = .12 ) , benefits ( R 2 = .07 ) , influence over work ( R 2 = .02 ) , pride in the organization ( R 2 = .02 ) , and working conditions ( R 2 = .01 ) . These five predictors explained 24% of the variance in the dependent variable . <p> Only three predictors were found for reported satisfaction with opportunities for promotion : recognition ( R 2 = .14 ) , influence over work ( R 2 = @ @ @ @ @ @ @ @ @ @ 2 = .0l ) . Altogether , these three variables explained 18% of the variance in satisfaction with promotions . <p> As one would have expected from the factor structure , the other two job satisfaction dimensions did not produce many or meaningful predictors . For satisfaction with supervision , pride in the organization ( R 2 = .10 ) and esteem ( R 2 = .04 ) explained a total of 14% of the variance . For satisfaction with co-workers , status ( R 2 = .07 ) , independence in work ( R 2 = .02 ) , and pride in the organization ( R 2 = .02 ) explained 11% of the variance . Discussion <p> The factor structure indicates that for this sample of educators three of the five reported dimensions of job satisfaction ( promotions , supervision , and co-workers ) differed from work values and constituted a separate factor . The common element of these dimensions seemed to relate to people . The fourth dimension ( satisfaction with the work itself ) related to intrinsic work values , and the fifth dimension ( pay ) @ @ @ @ @ @ @ @ @ @ job satisfaction dimensions were clearly reported to be separate from work values . <p> The factor called Intrinsic Work-Outcome Values points to the possibility that there is a second dimension of work that differs from Intrinsic Work-Related Values aspects . This finding seems logical because it contains the assumption that work can affect an individual directly through a set of first-level values or indirectly through a set of second-level values . The first-level values seem to have an introverted , subjective component , whereas the second-level values have more of an extraverted and objective nature . <p> Satisfaction with pay fits in well with other extrinsic job outcomes . The second extrinsic factor ( Extrinsic Job-Related Values ) , consisting of working conditions and hours of work , seems to measure an extrinsic dimension of satisfaction distinct from Extrinsic Job-Outcome Values . In short , the factor structure suggests five dimensions of job satisfaction that are somewhat different from those traditionally proposed : satisfaction with ( a ) the work itself , ( b ) work outcomes , ( c ) the job itself , ( d ) job outcomes , @ @ @ @ @ @ @ @ @ @ The regression analyses showed somewhat divergent results from the factor analysis . The first two factors contributed to the variance of nearly all job satisfaction dimensions . The intrinsic work-related values of achievement , influence over work , meaningfulness of work , abilities and knowledge , and independence contributed to the variance of satisfaction with work , pay , promotion , and co-workers . Similarly , the intrinsic work-outcome values of esteem , job status , recognition , and pride also contributed to all five job satisfaction dimensions . Pride in the organization explained some of the variance in all five dimensions . Two extrinsic values -- benefits and working conditions-explained some of the variance in satisfaction with pay . Thus , intrinsic values contributed to all five dimensions of job satisfaction , but extrinsic values contributed to only one dimension . <p> The results support and extend Herzberg 's theory . Not only did the variables Herzberg termed satisfiers -- achievement , recognition , the work itself , and responsibility -- load clearly on the intrinsic dimension of job satisfaction , but the results also suggest other values that may @ @ @ @ @ @ @ @ @ @ over work , being able to use one 's abilities and knowledge , having independence in doing one 's work , contributing to society , receiving esteem from others , gaining job status , and having influence and pride in the organization . In a later paper , Herzberg ( 1966 ) attested to the existence of other psychological needs that can motivate individuals to act . <p> The regression analyses supported the hypothesis that other values also influence job satisfaction . Although Herzberg 's sense of achievement accounted for the greatest amount of variance in satisfaction with the work itself , three other factors not identified by Herzberg added another 11% to the variance . All of the predictors for this dimension were intrinsic values . Indeed , as one would have expected for this dimension , most of the predictors were intrinsic work-related values . One might wonder which aspect of job satisfaction is tapped by the values of the second factor ( Intrinsic Work-Outcome Values ) . Adding and measuring such a factor would be logical because satisfaction with the work leaves out satisfying feelings arising from the @ @ @ @ @ @ @ @ @ @ 's theory . Convenient hours of work , which would include flextime , partially reflect Herzberg 's company policy and work conditions . This two-item factor ( Extrinsic Job-Related Values ) explained only 5% of the overall variance . The other extrinsic factor , Extrinsic Job-Outcome Values , also supports Herzberg 's hygiene factors . Thus the present research suggests that two extrinsic forces may contribute to job satisfaction or , as Herzberg phrased it , to an absence of dissatisfaction . Both forces relate to factors that motivate action to avoid discomfort . <p> The fifth factor ( Extrinsic People-Related Values ) contained three job satisfaction dimensions . Herzberg classified all of them , concerning co-workers , supervisors , and promotions , as hygiene factors . He would have claimed that these three variables can , at best , lead to an absence of dissatisfaction , but not to satisfaction . He would have argued further that these three variables can therefore not be considered job satisfaction dimensions , or they can be so considered only to a limited extent . <p> The only variable that did not fully support @ @ @ @ @ @ @ @ @ @ hygiene factor in his framework , it emerged as an intrinsic work-outcome value in this study . In the regression analysis , status was the main predictor of satisfaction with pay . Possibly this sample of educators , practicing in a relatively affluent geographical area , linked status to pay and to social position . <p> Except for satisfaction with work , the variance explained for the other four reported dimensions of job satisfaction was relatively small ; thus , other predictors of job satisfaction were not examined here . Because most of Herzberg 's factors were included in this research , it is apparent that his factors explain only a small percentage of each of the dimensions of job satisfaction . If the causal model of job satisfaction is accurate and values do combine to determine aspects of job satisfaction , then the results suggest that there may be two work dimensions , two job dimensions , and one people dimension . <p> TABLE 1 <p> Factor Loadings , Means , and Standard Deviations for Variables PREFORMATTED TABLE <p> <p> 