3.» nun 3“. 1 5 ) T34»! . wanflfimuba? ‘ . . a. . .{7 :5: V .l144p.132: 3...}... ‘ I. ‘ . vhrbqrin u.- uh... .. u . 3...; 7. 1:, 3.1., .. Xx... .9:sz : - . . it»: 5:... :\r-\ ifiv .z. Sly) 21 53“. ' ‘ l\\\\\\\\\\\\\\l\l\“ll‘l“l\\\\\\\\\\\\\\l l LIBRARY 1 “9’75 Michigan State 1 University This is to certify that the thesis entitled The Effects of Race, Gender, Community Involvement and Social Skills on Social Desirabilitg Response presented y Robert Taylor Anderson has been accepted towards fulfillment of the requirements for I , degree in Psychology Major professor Date 5/20/94 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE ll RETURN BOXto mwommum ywrnoord. TOAVOID FINEsmunonorbdonddoduo. 'gh DATE DUE DATE DUE . i t -“ - - [.J-LJ ! - MSU IcAnAfllmuflvo Adlai/Equal Opponmlty lm THE EFFECTS OF RACE, GENDER, COMMUNITY INVOLVEMENT AND SOCIAL SKILLS ON SOCIAL DESIRABILITY RESPONSE BY Robert Taylor Anderson A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1994 ABSTRACT THE EFFECTS OF RACE, GENDER, COMMUNITY INVOLVEMENT AND SOCIAL SKILLS ON SOCIAL DESIRABILITY RESPONSE BY Robert T. Anderson Social desirability' response (SDR), the tendency to endorse items indicative of behavior that is accepted but rarely performed and behavior that is not approved of but often committed, is often seen as a response bias which potentially distorts other answers given on any particular measure. In. personnel selection” persons’ scores may' be discarded or altered to reflect this response bias because applicants who give responses believed to be untrue of most people are suspected to have given false responses on other items as well. Analyses show that these scores may' be unnecessarily' altered. due t1) the jpossible influences of community involvement and social skills on individuals’ responses to social desirability items. It was hypothesized that race and gender would also have an effect on social desirability response, but these hypotheses were not supported. ACKNOWLEDGEMENTS I would like to acknowledge Professor Neal Schmitt for his constant availability and patience in providing me with support and guidance in completing this thesis. I thank him for the original idea which was the impetus for this work. I would also like to acknowledge the two remaining members of my thesis committee, Kevin Ford and Mike Lindell, for their views and comments. Finally, I thank Matthew Smith for tirelessly serving as a patient ear to listen to my ideas, frustrations and successes throughout the Master’s thesis process. iii TABLE OF CONTENTS LIST OF TABLES ..................................... V LIST OF FIGURES .................................... vi INTRODUCTION ....................................... 1 RELEVANT LITERATURE ................................ 4 COMMUNITY INVOLVEMENT ......................... 11 SOCIAL SKILL .................................. 15 SOCIAL DESIRABILITY ........................... 21 RACE .......................................... 25 GENDER ........................................ 28 HYPOTHESES ......................................... 30 METHODS ............................................ 33 SAMPLE ........................................ 33 PROCEDURES .................................... 33 MEASURES ...................................... 34 DATA ANALYSES ................................. 37 RESULTS ............................................ 39 DISCUSSION ......................................... 49 RECOMMENDATIONS .................................... 56 FURTHER RESEARCH ................................... 61 APPENDIX A (PAULHUS SELF-DECEPTION SCALE) .......... 63 APPENDIX B (PAULHUS IMPRESSION MANAGEMENT SCALE)...64 iv APPENDIX C (ABLE ITEMS) ............................ 65 APPENDIX D (COMMUNITY INVOLVEMENT SCALE) ........... 66 APPENDIX E (SOCIAL SKILLS SCALE) ................... 67 APPENDIX F (EXTROVERSION SCALE) .................... 70 LIST OF REFERENCES ................................. 71 LIST OF TABLES TABLE 1: MEANS, STANDARD DEVIATIONS, RELIABILITIES AND CORRELATION MATRIX OF OBSERVED VARIABLES ........ 40 TABLE 2: RELATIONSHIPS OF EXTROVERSION, ABLE AND SOCIAL SKILLS SCALES SCORES (BY PERCENTILE) ......... 47 vi LIST OF FIGURES FIGURE 1: HYPOTHESIZED RELATIONSHIP BETWEEN RACE, GENDER, COMMUNITY INVOLVEMENT, SOCIAL SKILLS AND SOCIAL DESIRABILITY RESPONSE ........................ 32 FIGURE 2: PATH ANALSIS OF COMMUNITY INVOLVEMENT, SOCIAL SKILLS AND SOCIAL DESIRABILITY RESPONSE (REVISED MODEL) ..................................... 42 vii INTRODUCTION In personnel selection, measures are often used to determine the extent to which an applicant's responses to a questionnaire or any self-report are truthful. One type of these measures is a social desirability scale, which gives an indication of the individual’s attempt to present a favorable impression of personal qualities. This I'attempt" is often seen as dishonest, a purposeful or idiosyncratic ruse to gain employment. Because ”lying" has such negative connotations and implications, the term 'impression management" is often used to refer to this response bias. A response bias (social desirability is only one of many) is any response or response style which may improperly influence the answers provided by an applicant. In such an instance, the construct validity of the selection instrument is brought into question because the actual content of the tool may not be addressed. The individual presenting a favorable evaluation in the measure may not be giving a true indication of her/his typical behavior or attitude, but rather the answers that he/she believes will be viewed as socially appropriate. In evaluating the applications of individuals, 2 employers often set a "cutoff“ score for acceptability or rejection of these applications based on the number of socially desirable responses given by an individual. Apparently, these businesses believe that anyone who gives an unexpected number of positive items cannot be trusted. Individual differences in the tendency to supply socially desirable responses do not seem to be taken into consideration. Due to other factors which may influence socially desirable responding, the assumption that these responses are no more than an effort to deceive or present one’s self in a favorable light may be problematic. Furthermore, an accepted cutoff point across individuals may be inappropriate. Taking into account a person's community involvement, frequency of interpersonal interaction, and social skill, socially desirable responses may be true responses. Given that this "response bias" of social desirability may, in fact, be an indication of true behavior, we should examine if social skill as defined by one's subgroup norms or culture plays a role in determining these responses. Furthermore, if such differences imply adverse impact for certain groups who may be more likely to give such responses, research in this area is definitely warranted. Therefore, I will present further information on social desirability as well as the tools that are used to measure it. This will be followed by proposed reasons why community 3 involvement and social skill may play a role in individual differences in responses on social desirability measurements. Based upon this reasoning, I will provide a number of hypotheses pertaining to the issues of community involvement, social skill, race, and gender, as they relate to social desirability response. RELEVANT LITERATURE The major problem with social desirability, as seen by researchers and practitioners, is the effect it has upon questionnaire validity (e.g. Bernreuter, 1933). During the last 40 years, social desirability response has been a concern in measuring personality (e.g. Edwards, 1953) and self—reports of sensitive behavior (Goode & Hart, 1952), among other constructs. Explanations for socially desirable responding (SDR), as well as measurements of SDR are numerous. Personality constructs previously credited with influencing SDR include need for approval (Crowne & Marlowe, 1964), repression— sensitization (Byrne,l964), censure avoidance (Allaman, Joyce, & Crandall, 1972) and self-deception (Paulhus, 1984). Rarely do researchers credit the individual with providing truthful responses due to social adjustment (McCrae & Costa, 1983) or learning acquired from social reinforcement (Edwards, 1957). Although these latter two are seen as more plausible in this paper, it is not the purpose of this paper to deny that any individuals may, in fact, respond dishonestly. Certainly some individuals will portray themselves dishonestly to an extent in a self—report. The S mistake on the part of the researcher and practitioner, I believe, is the generalization of certain levels of response across individuals. Some individuals who respond to these items honestly, may receive a high score because of a real tendency to behave in the socially desirable fashion. Honest responses by these individuals may result in scores that are inflated above a critical cutoff score. In addition, this interpretation questions the construct validity of SDR. A method believed to give a more accurate score on content measures is to partial out correlations between the content measure and the social desirability scalels). Another tactic to improve the validity of scores believed to be contaminated with social desirability is to adjust the raw score in relation to the score on an SDR measure (e.g. Norman, 1967). This is done by regressing the content score on SDR. The corrected score is the residual found in the regression equation. Instead of adjusting an individual's scores, however, the scores may be discarded if the SDR level is above a critical cutoff score. A cutoff score may be determined by purposely providing favorable or unfavorable responses, or by providing some respondents instructions to fake good or fake bad (e.g. Helmes & Holden, 1986). According to my hypothesis, however, these cutoff scores may be set too low or set inappropriately because they do not account for cultural or subgroup differences in SDR . 6 Furthermore, research (Holden & Fekken, 1989) indicates that the more well-known instruments used to determine levels of SDR are not highly correlated. Factor analyses of these measures generally yield two factors that explain SDR (Borkenau & Ostendorf, 1989). One cluster is associated with the general anxiety factor of the MMPI (Block, 1965) and the second with another factor of the MMPI (Wiggins, 1964) which measures agreeableness and traditionalism. Paulhus (1986) proposes that these two SDR factors represent (a) self- deceptive positivity (an honest but overly positive self- presentation) and (b) impression management (self- presentation tailored to an audience). While the latter of the two factors is indeed plausible in many instances, the contention is with the fonmer, "Individuals may give an honest but overly positive self-presentation“. What standards are used to determdne “overly positive“? How do we kggw_that this is a misrepresentation of actual behavior? This is the evaluation to be made with the majority of social desirability scales. Three of the social desirability measures that have received the most attention from researchers are a) Balanced Inventory of Desirable Responding (Paulhus, 1984, 1988) b) Edwards (1957) Social Desirability Scale, and c) Marlowe— Crowne Social Desirability Scale (Crowne & Marlowe, 1960). The Balanced Inventory of Desirable Responding (Paulhus, 1984, 1988), or BIDR, is designed to measure the 7 constructs of self—deception and impression management (mentioned above). The focus of the items pertaining to self-deception is ego enhancement. Thus, the development of these items involved overly positive judgments of individual attributes. The impression management dimension of this scale was based on the belief that individuals would be less likely to admit to the frequency of negative actions performed while exaggerating the amount of positive behaviors performed. These rationally developed items involve overt behaviors (e.g. I never take things that don't belong to me) so that any distortion is interpreted as a deliberate lie. The 40 items are rated by the respondents on a seven- point scale. For each extreme response (6 or 7) an additional point is given to each total score on the entire measure. This rationale ensures that only individuals who dramatically emphasize positive traits or behaviors will score highly on the measure. The scale was administered to a sample of 884 religious adults (no information was given as to the racial make-up of the sample) by Quinn (1989). Means for the self-deception items were 7.6 (s.d.=3.1) and 7.3 (s.d.=3.1) for men and women, respectively. Means found for the impression management items were 7.3 (s.d.=3.1) and 8.9 (s.d.=3.1) for men and women. The internal consistencies for the self- deception measure ranged from .68 to .80 while reliabilities 8 ranging from .75 to .86 were found for the impression management scale. The measure as a whole had a reliability of .83 (Paulhus, 1988). A second scale of social desirability was constructed by Edwards (1957). In doing so, he selected 150 items from several MMPI scales (F, ”plus-getting“ and test validity; L, the Lie scale; K, the dissimulation or "faking good'I and “faking bad“ scale; and the Manifest Anxiety scale [Taylor, 1953]). These items were given to ten judges (no race or gender specified) to rate each item's social desirability. The final version incorporated the 39 items which best discriminated between high and low scorers. These items are judged to have either extremely high or extremely low desirability ratings. The self-deception factor of social desirability seems to be measured by this scale, as indicated by high correlations with measures of adjustment and personality, which have been related to self-deception (Taylor & Brown, 1988). A measure's high correlation with this scale is interpreted as evidence of a response bias which is overly positive. With this measure, also, an "overly positive“ score may be incorrectly inferred. The Marlowe-Crowne Scale consists of 33 True-False items concerning everyday behavior. The scale was constructed using 50 items of behavior that were designed to be culturally approved but performed by almost no one in the population. These items were judged by a panel of ten 9 individuals, faculty members and graduate students in the Department of Psychology at the Ohio State University, to determine if the items met the above criteria (it was not mentioned if any women or minorities were included on this panel). Those items which received an agreement of 90 percent or greater (47 items) were included. This version of the scale was administered to 76 introductory psychology students (again, no numbers of females or minorities were given). An item analysis found 33 items with satisfactory (.05) discrimination between high and low scorers on the scale. Those 33 items were used for the final scale. Internal consistency and the test-retest correlation for the scale were both .88. This scale is based on the belief that the majority of people, at some time, perform the negative behaviors and fail to perform the positive behaviors. Therefore, denial of the undesirable items and endorsement of the desirable items are viewed as responding in a socially desirable manner. While the Marlowe-Crowne scale does correlate with self- deception, it correlates more highly with impression management. Again, as mentioned before, it is my assertion that “impression management" may be a result of honest responses on the part of some identifiable subgroups. Paulhus (1986) gives several recommendations and cautions in interpreting high correlations between self- reports and SDR. Studies have shown that controlling SDR 10 reduces the predictive validity of content measures (Borkenau & Amelang, 1985; Kozma & Stones, 1988; McCrae, Costa, Dahlstrom, Barefoot, Siegler, & Williams, 1989; McCrae & Costa, 1983; Ruch & Ruch, 1967). SDR measures that involve self—deception (such as the Edwards or Marlowe- Crowne scales), according to Paulhus (1986), apparently will lower the predictive validity of content measures that tap the construct of self-deception such as well-being, perceived control, and self-esteem. Paulhus (1986) does, however, recommend that impression management be controlled in specific job-selection situations. This would be done by determining a cut—off score using a mean difference between subjects instructed to give a favorable impression and those not given such an instruction. Paulhus (1986) concedes, though, that in some personnel selection situations a high scorer on an impression management scale may actually be beneficial (e.g. public relations). Another potential inadequacy of most SDR scales is that they contain I'improbable" responses. If it is hypothesized that the majority of individuals will not be responding truthfully when these items are endorsed, then there should be little differentiation among applicants and few applicants would endorse such items according to this definition of I'accurate" responses. If this were the case, then it would be of little benefit to include these ll measurement tools. They would not be measuring socially desirable behaviors, or the propensity to perform socially desirable behavior, because there would be little range between people, giving no indication of individual differences in behaviors within actual situations. However, there are individual differences in responding to these items, and it is possible that the individuals who have “aberrant“ scores do respond truthfully. These differences in responses may be related to community involvement and social skill as defined culturally or normatively by certain subgroups in our society. In this case, it cannot be directly determined how the responses to these social desirability scales should be evaluated. If my thesis is correct,:members of these subgroups cannot be evaluated properly using the standards of the majority group. Differences in interpersonal interaction and social competency and tendency to respond in a socially desirable manner may be indicative of real cultural differences and not indicative of inappropriate job behavior. These differences, I believe, are influenced by one's community, community involvement, and the social competency that results from this involvement. COMMUNITY INVOLVEMENT Involvement in the community can serve several functions which may increase individual levels of interpersonal interaction and social competency. Among these 12 are: 1) development of group goals, community or societal cohesion, 2) a problem solving medium to deal with problems encountered within the community, and 3) provision of an atmosphere of sociability (Smith, 1966). Participation in these community groups lends itself to situations where socially desirable behavior is essential to make sufficient progress to the collective goal. It is difficult to argue against the notion that increased exposure to such interpersonal situations would increase individual knowledge of accepted and expected behavior when in the presence of others. It would also follow, then, that these individuals who are aware of the more socially accepted behaviors would be more likely to perform these behaviors than someone who is not familiar with the aspects of interpersonal etiquette. Not only would these behaviors constitute social desirability, they would become the social norm. These relationships with others in the community would essentially dictate the actions which are expected of its members (Fischer, 1982), such as to be polite in public with friends and avoid shaming those within the community (socially desirable behaviors). Furthermore, the strength of interpersonal ties in the community would impose a degree of social control upon each individual involved in the community (Fischer, 1982). Therefore, more socially desirable behaviors would be 13 demonstrated and required in interpersonal situations. Individuals living in such a communal environment would be much more concerned with how they are viewed by others and would be equipped with the necessary skills to avoid negative evaluation by others. In relation to social desirability scales, if these individuals are concerned with their presentation on a questionnaire, they should also be as concerned, or more concerned, with their behavioral presentation among others. In interpersonal situations, particularly among peers, these individuals would be under closer scrutiny and any deviant behavior would be subject to direct scorn or verbal disapproval. Therefore, individuals who wish to be seen in a positive light would adjust their actual behavior accordingly. Consequently, responses on a social desirability instrument would be reflective of actual behavior in such situations as depicted in the questionnaire items. It could be argued, of course, that these individuals would not be acting out of kindness or goodwill in their everyday behavior. They may only present a favorable appearance to avoid rejection by others. This may well be the case, but the end result remains the same. For example, an individual may avoid presenting gossip about others, regardless of how much the individual may be tempted to do so, if she/he realizes that gossiping is unacceptable 14 behavior. Therefore, if a questionnaire item asks whether or not the individual gossips about others, the answer would still be “no” and that answer would be truthful. Finally, levels of community involvement are affected by individual perception of problems within the community and a need to form an alliance to achieve group goals (Hallman, 1984). When situations arise to make community members aware of sindlar needs and circumstances, more developed forms of community involvement are created (Warren & Warren, 1977). Korte (1988) also found that neighborhoods where members were well acquainted with one another had more community involvement, which is consistent with previous findings (e.g. Unger & Wandersman, 1983). This involvement and acquaintance would serve to further strengthen the ties within the community. I believe that this involvement is different from the desire for social interaction with others. I see this community involvement as developing from an awareness of community problems and a perceived obligation to help solve these problems (regardless of anticipated social interaction). Often, individual tendencies to interact or not interact may be put aside because the community itself is of main concern. This is exemplified by the saying "If not me - who? If not now — when?'. In this case, an individual who usually prefers not to interact with others may view the problems of the community as more important than her/his reluctance to 15 participate in a potential social atmosphere. It is my contention that perceptions related to community involvement may also be affected by race and gender, where certain subgroups may feel disadvantaged in‘ society (to be addressed later). SOCIAL SKJILL; Unfortunately, the literature relevant to social skill has focused upon maladaptive or maladjusted behavior (Doll, 1953; Trower, Bryant, & Argyle, 1978; Wine & Smye, 1981). Therefore, research on the normal population in the area of social skill is limited. The term social skill (often used interchangeably with “social competence" or ”interpersonal competence") can incorporate many facets of interpersonal behavior such as empathy (Hogan, 1969), sociability (Cheek & Buss, 1981), and self-monitoring (Snyder, 1979). Thus, one accepted definition would be impossible. For the purposes of this research, “social skill“ will refer to individuals’ experience in, and seeking of, interpersonal interaction and the skills which may be acquired through such interaction. These skills include being comfortable around others, approaching and making new acquaintances, and the ability to present themselves and their organizations positively. Two major themes in the literature on social skill are collaboration and adaptability. Collaboration refers to the desire of the individual to obtain not only personal goals, but facilitate goal achievement of others as well. According 16 to Pearce (1976), this is indicative of social skill. This ability to collaborate with others would appear to be a necessary result of working with others to achieve goals, such as in a community organization. Consequently, gaining such an aspect of competence is one skill which would enable one to be more comfortable around others and approaching new acquaintances. The ability to respond appropriately in changing situations, or adaptability, is also considered a major characteristic of the socially competent person (Steffen & Redden, 1977; Sundberg, Snowden & Reynolds, 1978). An aspect of this element is behavioral flexibility, which refers to one’s ability to adjust one’s behavior according to each encountered situation and the constraints it entails. This adaptability would appear crucial in presenting one’s self or one’s organization in everyday, as well as, novel situations. To acquire these skills, Bandura (1965) tells us that certain behaviors are acquired through social learning, which would entail observing others. When we witness individuals being rewarded (e.g. a smile or pleasant conversation) we are more likely to acquire and exhibit these behaviors than if we had not been exposed to such behavior at all. According to Goffman (1959), we engage in verbal and nonverbal impression management when we interact with others. We also rely on others for cues when an 17 ambiguous situation is presented (Friedman, 1979). I, therefore, see it as a reasonable assumption that the more models an observer witnesses, the better generalizations that individual can make regarding appropriate (socially ‘ desirable) behavior across people and situations. The present study, therefore, views social skill as a learned skill. This is consistent with the views of McFall and Twentyman (1973), who propose that socially unskilled individuals never acquired the appropriate responses in interpersonal situations. Those more experienced with interpersonal situations are seen as more likely to have acquired the necessary knowledge of what is expected of each individual in social interaction. Thus, through personal experience, or the observing of others, these individuals have the opportunity to assess what is appropriate and inappropriate behavior. Mere exposure to others’ behavior is not sufficient to acquire and imitate such behavior. However, I believe acting in a more socially desirable manner is, however, more likely among these individuals than among those who lack such experience and skills. In fact, several aspects of social skill have been found to be significantly correlated with social desirability. Riggio (1986) found three such aspects (social sensitivity, social expressivity, and social control) which were correlated with the Marlowe-Crowne Social Desirability Scale (—.31,.26,and .48; all 18 significant). Social sensitivity was defined as having the knowledge of generally accepted behavior and norms (ability to assess appropriate and inappropriate behavior). Socially sensitive individuals are also seen as being attentive to the behavior of others (learning through observation). Therefore, these individuals may be more concerned with the appropriateness of their own behavior. The negative correlation may be due to the fact that the definition in this study also included social anxiety and taking the behaviors of others too personally, which may inhibit social interaction. Those having the knowledge of appropriate behavior, yet not overly self-conscious are of more concern in my thesis. A second aspect of social skill is social expressivity. This skill involves the ability to engage others in social interaction and initiate conversations with Others (approaching and making new acquaintances). Also found in Riggio’s study was that extroversion correlated (p< .01) with the social expressivity dimension of social skill. This would, of course, facilitate opportunities to interact with more individuals, and thus enhance one’s knowledge of interpersonal skills. Therefore, one would expect that a measure of extroversion would correlate significantly with social skills in this current study as well. The third skill, social control, involves social self- presentation (ability to present one’s self or one’s 19 organization positively). These individuals are socially adept and self-confident and able to adjust their behavior according to each social situation. This indicates the knowledge of appropriate behavior which would result in the person acting in a more socially desirable manner. Of course, correlations with the Marlowe-Crowne scale do not, in themselves, imply causality. However, I see it as unlikely that a person can wish to be seen as socially desirable or avoid negative evaluation independently of others. In other words, a person who would endorse socially desirable responses would most likely do so because s/he is influenced by others or by the opinion of others. Even in the event that the individual responds so as to view herzhimself in a favorable light, s/he would still probably use social norms to determine what a 'favorable light“ is. Therefore, it is unlikely that a person can be aware of which responses or actions are socially desirable without exposure to social situations. Because the development of social skills appears to require social interaction, it is not seen as plausible that social desirability directly causes social skill. While the concern to be seen as socially desirable may cause one to seek the skills to do so, the skills can only be acquired through actual experience. The concern to be socially desirable cannot be evident to others unless socially desirable behavior is performed. It is the position of this research that these 20 skills actually lead to socially desirable behavior and responses. An anticipated shortcoming of the above finding of Riggio (1986) is that all four scales (social sensitivity, social expressivity, social control and social desirability) were measured through self-reports, which are often seen as subject to personal bias. However, social competency measured through self-reports has been found to correlate with peer-reports of social competency as well (Buhrmester, Furman, Wittenberg, and Reis, 1988). It is important that an individual’s perception of his/her own behavior is shared by those who actually observe the behavior of that individual because it lends validity to such measures, at least for the concept of social skill. Furthermore, friends of the individual would be most familiar with that person’s typical behavior. Thus, the measurement of social skill through self-reports may indeed be an accurate description of personal behavior and interaction and verified by those who are the objects or observers of such behavior. The content of social desirability items is often related to social skill or social skills. Thus, the responses given to these items may be indicative of true behavior and/or beliefs about one’s behavior. These behaviors may be more prominent among particular subgroups of the population. In use with personnel selection 21 (pertaining to job-related behaviors), therefore, scores may be inappropriately corrected, resulting in a negative impact upon these subgroups. SfiIAL DESIRABILITY Ironically, a high score on a social desirability scale may be indicative of the person having adopted the social values of her/his cultural subgroup, but among researchers and practitioners, these high scores may be seen as undesirable self-enhancement. Social desirability, as stated earlier, is often viewed as simple impression management or self-deception (Paulhus, 1984) which is thought to distort other responses on a questionnaire or application. However, the arguments of the research presented above suggest other possible reasons why individuals may endorse socially desirable responses on a questionnaire. These reasons include indications of actual behavior and the avoidance of misrepresentation. The items I believe that may have content which will be affected by the above arguments are presented in Appendix A. For example, an item from Paulhus’s self-deception scale “My first impressions of people usually turn out to be right“, could actually tap a person's ability to interpret the actions (verbal and nonverbal) and behaviors of others. This, as mentioned earlier, is a characteristic of social sensitivity, or being attentive to others. Therefore, perhaps the socially sensitive individual is more adept in 22 her/his observations and judgments of other people, which could lead to an "exaggerated" endorsement of this item. Also, in Paulhus’s impression management scale (items I believe are influenced by my arguments are presented in Appendix B) an item such as 'I have said something bad about a friend behind his or her back" could be influenced by one’s community. It may be that if the listener is also a friend of the person being insulted, the listener may sense a responsibility to tell this person. This would perhaps not be seen as gossip, but rather as “looking out" for his or her friend. As another example, the item.'I sometimes try to get even rather than forgive and forget” is related to community involvement. Attempts to "get even“ with individuals within the same organization may prove to be counterproductive in accomplishing group goals. As argued above community involvement and social skill may influence an individual to actually behave in a more socially desirable manner, regardless of sincere or insincere motivation. It has been suggested (Loo, 1980) that in some situations lie scales (social desirability scales and lie scales may be used interchangeably) may provide truthful responses and that positive responses to such scales may be the result of conditioning into socially conformist behaviors (such as within a community). Not- withstanding actual negative affect or ambivalence toward others, individuals who actually behave in a socially 23 desirable manner are those most likely to possess the necessary skills to be accepted among others and those persons who are most desirable in work situations in which it appears that measurement of social desirability would be justified. Another possible reason why these individuals may have elevated levels on a social desirability scale is that they are attempting to avoid being seen as impolite or inconsiderate, as the expected responses (by those who see social desirability as impression management or self- deception) would indicate. It is stated implicitly by those who view social desirability as impression management or self-deception that the only truthful answer is l'no". Any individual responding otherwise is either intentionally lying to the evaluator of the questionnaire or unintentionally to themselves. This may not be the case at all. It may be impression management, but in a different sense. Impression management, as seen by Paulhus (1984) involves intentional distortion of actual behavior patterns, but the type of impression management which is possible may involve preservation of actual behavior patterns. For example, an item which asks, "I never hesitate to go out of my way to help someone in trouble" (Crowne & Marlowe, 1960) and only provides possible responses of “True“ or ”False” simply does not allow any differentiation with respect to actual behavior patterns. 24 If one individual almost always helps people without delay and one individual ggygg helps people without delay, the absolutely true answer for both individuals would be 'No'. However, the individual that almost always helps may not ' wish to be seen as someone who never or rarely helps (which is possible with a 'no' response). Therefore, that individual may respond 'Yes', giving a more representative indication of his/her typical behavior. This could further escalate the score of individuals who actually do engage in more socially desirable behaviors and wish to respond in a manner which would reflect that behavior. Therefore, a wider range of responses (e.g. “almost always“, 'sometimss') must be provided for these individuals before one can assume that he/she is lying. Otherwise, a person who actually behaves in a socially desirable manner cannot give an honest and accurate representation of his/her behavior. If he/she usually helps without hesitating, a "False'I response could be interpreted as a lack of concern for others, which would not be true of that individual. On the other hand, if the individual responds “True“ it is seen as a lie. Such items from the ABLE (Assessment of Background Life Events) scale are presented in Appendix C. If the arguments presented above are accurate, these response scales for social desirability measures may be inadequate. Use of these dichotomous response scales could contribute to the problem of discerning the truthfulness of 25 responses on social desirability scales. PLAQE. Another factor which may predispose an individual to socially desirable behaviors and thus, socially desirable' responses, is culture. Obviously, cultural differences exist among various races, such as between Blacks and Whites. One of these differences may be the extent to which members in these racial communities interact with one another (within the same race). As contended earlier, the amount of interaction with others, within one's subgroup, may facilitate the development of social competency, as defined by that culture. The hypothesized result is that individuals who achieve such competency are more likely to behave and respond to questionnaires in a socially desirable manner. It is hypothesized that Blacks will be more likely to attain social competency (as defined by that culture) due to the higher frequency of interaction within one’s culture among Blacks than among Whites. In addition, as previously noted, involvement in the community is often predicated by a perceived sense of necessity to join together to achieve group goals (Hallman, 1984). Due to not only perceived, but statistical, disadvantages among Blacks, it is more likely that these groups will seek and form community groups to assess these needs and goals. In fact, strong bonds between families and communities is seen as characteristic of Blacks (Presser, 1980; Rainwater, 1970; Stack, 1974), whereas 26 Whites engage in less informal involvement (Mitchell, Barbarin, & Hurley, 1981). Two major strengths of the Black community, as noted by Hill (1971) are a strong connection with church groups, and that, in many ways, the entire Black community is seen as a family. The family-like atmosphere among the Black community, I believe, would serve to increase and strengthen the social ties of each of the members. The resulting social control would cause a greater adherence to the social expectations and norms within the community. These social norms would most likely correspond with behaviors that are viewed as socially desirable. The church, among Blacks, provides a natural setting for social interaction (Gary, 1990). Gary (1990) believes that the Black church serves additional roles: 1) economic development, 2) educational development, 3) group and identity values, 4) leadership development, 5) social support, 6) protest and political development, and 7) psychological support. Of particular interest among the above functions of the Black church are protest and political development. This function may also be the result of another motive to be involved in the community -— racial consciousness (Shingles, 1981). Increased group consciousness has also been the result of the efforts by Blacks to fight against racism (Gooley, 1990). Blacks, who often view themselves as treated unfairly throughout American history (Wilson, 1980), may 27 feel that they must depend on their own resources to improve their status in society. This would lead to increased community involvement and community groups. White males, however, perhaps see little need to form such alliances or have such alliances readily available to them to cope in a society where they are the majority. Therefore, they are less likely to meet only among themselves and in more diverse settings, such as Blacks (and other minorities) and women. As previously hypothesized, the result would be greater interpersonal interaction, greater cultural social competency, and the exhibition of more socially desirable behavior among Blacks than among Whites. In addition, Blacks must exhibit more behavior flexibility because of their minority status. This is due to the fact that they must adapt not only to the Black subculture, but to the majority White subculture as well (Boykin and Toms, 1985). A Black person raised in a Black community must, in most instances, interact with the majority population, especially in obtaining employment. In such an instance, there would be great motivation to present one’s self favorably, in light of perceived racial inequality with regard to achievement for Blacks in America (Wilson, 1980). Whites on the other hand, could conceivably not interact significantly with members of a minority group and still achieve any level of success. Therefore, they may not achieve the degree of behavioral flexibility necessary 28 to adapt to situations where they may have to interact with others of another subculture. GENDER Gender may also play a role in social desirability scales due to similar reasons given for race. Women, also, often perceive dissimilarity in their treatment by society in comparison to men. This could be an underlying reason for the development of women’s movement groups such as the National Organization for Women (NOW). Eagly (1987) found that gender differences in emotion (emotion skills and social skills being interrelated) are necessary for both men and women to adapt to their roles in our culture. Brody and Hall (1993) state that this is evident in women’s lower status and power. Women also face discrimination in the workplace where they may feel the need to adapt to a male— dominated environment. WOmen, therefore, would learn to exhibit more social skills than would men. In addition, a review by Brody and Hall (1993) found that women are superior to men at interpretation of facial, voice, and nonverbal cues. This ability to decode the expressions of others would facilitate one's exhibition of the appropriate (socially desirable) response. Furthermore, females are also more personal in their self-disclosures (Brody, Hay, & Vandewater, 1990), which could be interpreted as being more comfortable around others. Also, according to 29 Brody and Hall (1993), women display more warmth and happiness (emotions which are related to affiliation and self-consciousness) than men. Affiliation is also shown in women’s tendencies to express their own shortcomings while accentuating positive aspects of others. These characteristics of nonverbal decoding, self-disclosure, and affiliation are seen as indicative of social skill and, apparently, women display more of these traits than do men. Therefore, if my hypothesis is correct, women will have higher self-reports of social skill and social desirability. In summary, responses on social desirability scales could be influenced by an individual’s involvement in the community, experience with others, and the hypothesized result of this interaction -- social skill. Further, Blacks and women, I believe, would be involved in more social groups in the community than White males due to political groups designed not only for the community as a whole, but for their subgroups in the population as well (the latter being less prevalent among White males). This increased involvement in the community and other people and resultant social skill should produce heightened scores on the social desirability measures used in personality tests. If Blacks and women do score higher on social desirability scales, while responding accurately with respect to their beliefs about their own behavior, the validity of personnel selection tests incorporating these 3O scales could be jeopardized. I believe that this could lead to inappropriately lowered scores for these two groups, leading to adverse impact when these measures are used. My arguments with respect to the interrelationships of race, gender, community involvement, and social skill and their impact on measures of social desirability are summarized in Figure 1 and in the following hypotheses. Based upon the above reasoning, the following hypotheses are offered: ER: The scores of Black individuals will be higher than those of White persons on social desirability measures. IF: Race will also be correlated with community involvement and social skill as defined above such that Blacks will achieve higher scores than Whites on these measures. PP: The relationship between race and social desirability scores will be mediated by community involvement and social skill as indicated in Figure 1. PP: The scores of women will be higher than those of males on social desirability measures. PP: Gender will also be correlated with community involvement and social skill as defined above such that women will achieve higher scores than White males on these measures. 31 l?: The relationship between gender and social desirability scores will be mediated by community involvement and social skill as indicated in Figure 1. PF: Higher scores by Blacks on social desirability scales will result in adverse impact when scores on these scales are used to “correct" scores on other measures. IF: Higher scores by women on social desirability scales will result in adverse impact when scores on these scales are used to “correct" scores on other measures. 32 waOdwmm >._._.__m_m>._0>z_ 722225.00 .mmosz . w0I up mmDGE ._.Zm_2w0._0>z_ EZDEEOO w0mmv wszdmm—m >._._.__m._0>z_ FEZDEEOO m0 m_w>.__m>._0>2_ ._<_00m K F. EZDEEOO No. 43 The chi—square test of model fit for the model that did not include direct effects of community involvement on social desirability measures yielded a value of 51.03 with 11 degrees of freedom. This was significant indicating lack of model fit. However, various other fit indices indicated the model fit the data relatively well. The GFI of .92 and the AGFI of .79 indicate that the model explained a substantial proportion of the variance. A second model including direct paths from community service directly to each social desirability measure yielded a chi-square value of 49.84. The first model is nested within the second model, so a difference chi-square was computed to test the combined direct effects of community service on social desirability. This chi—square (1.19) was non-significant (df=3). There were no changes in the goodness of fit indices or significant improvement upon the hypothesized path model as a function of the added direct paths. Therefore, I conclude that these paths should not be included in the model. The path analyses gave significant support for the hypothesized effects of community involvement on social skills, social skills on social desirability response, and social skills as a mediator of the relationship between community involvement and social desirability for two of the three indices of social desirability. The exception was a lack of an effect of social skills or community involvement 44 on the impression management dimension. The path coefficient relating community involvement to social skills was .17 (t=5.42, df=6, p<.05) but there were no direct effects on the measures of social desirability. The direct effects of community involvement were .02, .00, and .02 for the ABLE scale, self—deception, and impression management, respectively, all of which were nonsignificant. As previously mentioned, the construct of social skills as measured by the scale used in this study did account for a significant amount of variance with respect to the ABLE items and self—deception but did not correlate with responses on the impression management scale. The path coefficient for the ABLE items was .18 (t=3.38, df=6, p<.05) and for the self—deception items, .13 (t=4.04, df=6, p<.05), but only .06 (t=l.51, df=6, p>.05) for impression management. Although there were no race and gender effects, eliminating the possibility of changes in adverse impact resulting from use of social desirability scale corrections, analyses were performed to determine the number of individuals that would be affected by correction in scores to reflect high scores on social desirability measures. Typically corrections are made only in extreme cases. For example, examinees’ scores may be discarded when they obtain social desirability scores above the 90th percentile on the scales. Therefore, only the number of individuals scoring 45 above the 90th percentile (and the 50th percentile, as a further example) on the ABLE scale (because it had the highest correlation with social skills) was determined (ABLE - x=3.49, 90th percentile >6). The same procedure was performed for scores on the social skills scale at the 75th and 50th percentiles (social skills - x=30.46; 75th percentile >34; 50th percentile >30). I took this approach because it was not the focus of this study to determine how many people in general may be affected by social desirability correction. Rather, it was my point to find how many people who may be reporting genuine behaviors (as measured by the community involvement and social skills scales) could be affected. It was never my intention to contend that social desirability scores should never be corrected, but that it should be determined if these scores are a function of individual community involvement and social skills as hypothesized above in this study. If this is the case, I believe, that these scores should not be corrected. Instead, analyses were performed to determine the possible impact of businesses correcting or discarding applications as a result of individuals’ high scores on social desirability measures (see Table 2). Only one test was chosen for analyses because a composite test battery score could not be computed due to the fact that the scoring weights for the entire battery have not yet been determined. n‘ (I) (I3 El: ‘bll ‘Iau'. DA“ 5‘," § 46 Also, for some tests in the battery (i.e. cognitive ability), social desirability corrections would not be appropriate. Therefore, scores on the extroversion scale (Costa & McCrae, 1978) were chosen as an example of correction effects on a selection test. Conceptually, it seems that the extroversion measure (correlation with ABLE scale: r=.17; p<.05) would be susceptible to SD bias and empirically extroversion was correlated with both SD and social skills measures. I believe this is so because individuals scoring high on the extroversion scale would be more likely to interact with others (in the community and elsewhere) which would enable to acquire the appropriate social skills. In addition, these individuals (who have attained these skills) would feel more confident in social situations and would engage others in discussions. In Table 2, I present the number of individuals scoring within the the top 10th, 25th, and 50th percentiles on the extroversion scale (see note under Table 2). Of these individuals, I further determined how many scored above the top 90th and 50th percentiles on the ABLE scale while also scoring above the top 50th and 75th percentiles on the social skills scale. This yielded the number of individuals who would be "inappropriately" affected by correction of the extroversion scale using scores on the ABLE. 47 TABLE 2 RELATIONSHIPS OF EXTROVERSION, ABLE AND SOCIAL SKILLS SCALE SCORES (BY PERCENTILE) Extrover. #selected ABLE Social Skills #corrected 8.4% 31 90th 50th/75th 4/4 50th 50th/75th 10/9 23.8% 67 90th 50th/75th 10/6 50th 50th/75th 20/14 46.7% 114 90th 50th/75th 16/8 50th 50th/75th 29/18 * The ’number corrected' reflects the number of scores that would be subject to correction at the 50th and 75th percentiles of social skills scale, respectively. ** The entire sample of 227 was used for these analyses instead of the race and gender identifiable sample of 162 due to the lack of correlation between race and gender and these dependent variables. *** Due to identical scores on the extroversion scale, precise 10th, 25th, and 50th percentiles could not be determined. 48 As is obvious in the last column of Table 2, a substantial number of applicants (up to 1/3) would have been penalized for high scores on the ABLE scale when an alternative interpretation of the ABLE score is that they possessed a high knowledge of social skills. If scores on the social skills scale are indeed an accurate indicator of an individual’s typical behavior, these corrections would be unfair to these applicants. However, it should be noted that I picked a scale for which it was highly likely that these results would be found. Whether similar relationships hold for a broader range of personality and biodata instruments remains to be determined. DISCUSSION As indicated by the results of the path analyses, community involvement does play a role in the development of social skills. Those individuals more involved in the community adhere to a more socially desirable code of behavior (as measured by the scales provided) which may be manifested in their everyday actions if their responses to the scale items are an accurate manifestation of these actions. Although the degree of variance in community involvement may have been limited due to the sample used in this study (discussed below), its ability to predict social skill was not eliminated. This is not to say that there is not a reciprocal effect. It could also be that those who attain social skills through community involvement would also have a higher propensity to participate in more community activities due to their confidence and desire to interact with others. Also, social skills, as predicted, did have an effect on aspects of social desirability response. Also as hypothesized, social skills was a mediator of the relationship between community involvement and social desirability responses. Apparently, as addressed in the 49 so introduction, mere exposure to others is not sufficient to modify individual behavior. There must be some indication that certain interpersonal skills were acquired as a result of that personal interaction which is provided by the social skills scale. These findings support my hypothesis that an individual’s community involvement often leads to an increase in her/his level of social skills. In addition, extroversion had significant correlations with social skills and community involvement. Therefore, it is likely that there is a relationship between these three constructs. One possible explanation is that those individuals who are more extroverted acquire more social skills, are more confident during social interactions and consequently seek to be involved with community activities. Conversely, it is also plausible that those extroverted individuals seek contact with others (e.g. through community activities) and acquire higher levels of social skills. The data collected in the study, however, prevent a determination of causality. Unexpected, however, was the lack of an effect on the dimension of impression management. Upon further review, though, this is seen as reasonable. The ABLE and self— deception items pertain to individual perceptions about one’s self, are open to interpretation, and are less evaluative than are the impression management items. Perhaps individuals are more aware of the “correct“ response 51 on the impression management scale because the items lack subtlety. In other words, this scale is easier to fake. The self-deception scale, however, contains items which do not necessarily reflect a negative image on the respondent and the socially desirable response may change according to what characteristics are valued by the evaluator of the scale. For example, the item.'I am_a completely rational person', can have at least two interpretations. A completely rational person may be excellent for a business executive related position. However, in the case of an advertising applicant, this may be indicative of a lack of creativity. Therefore, I assert that where the socially appropriate response is more ambiguous, individuals would have a stronger tendency to simply respond as honestly as possible and the responses of individuals with higher levels of social skill would be in the more desirable direction. This would account for the significant effect of social skills on self-deception despite the low standard deviation (.1363) of the self-deception scale (lower than impression management, .1533). It would appear that the previously mentioned aspects of social skill (social sensitivity, social expressivity, and social control) do apply to the content of the social desirability items. Because it has been previously found that self-reports of socially competent behavior are relatively accurate (Buhrmester, Furman, Wittenberg, & Reis, 52 1988), I believe that these respondents are more familiar with these dimensions of social skills. Contradictory explanations for the above results could include a possible social desirability component in the community involvement and social skill scales. This is not a significant concern for at least two reasons: 1) the anonymity and lack of evaluation of the participants in the study, and 2) the fact that the use of biodata items generally decreases social desirability (Shaffer, Saunders, & Owens, 1986). Although I would have preferred to have a larger sample of Blacks and women in this study, I must concede that the correlations found were remarkably lower than expected even given the relatively small proportion of these groups in the sample. Therefore, explanations other than sample size must be offered. I believe that the differences between race and gender may have been minimized as a result of all examinees being incumbents. I previously viewed this as an advantage in anticipation of more honest answers given to the questionnaires due to the anonymity offered to the examinees. However, it is possible that one of the variables, social skills, could actually be learned through job experience. It was my argument that the more interaction one has with others, the greater one’s social skills will become. Gaining the cooperation of others is, in fact, a necessary requirement for an effective agent. Therefore, 53 individuals who did not possess such skills at the onset of employment with this agency may have learned them through others (e.g. their partners or witnesses), perhaps after previously-learned inappropriate or inadequately-learned ‘ interpersonal techniques proved ineffective. Further examination of the correlations with race and gender shows that not only are the correlations for gender (male=1, female=2) and social desirability low, but they are negative. It was found that social skills for women are negatively related to social desirability response. A possible explanation for this is again a function of the positions held by the applicants. Perhaps the female examinees acquire social skills through community involvement (as hypothesized and supported by data) but do not engage in socially desirable behavior or respond in a socially desirable manner because they feel that such behavior would be ineffective. They could believe that they may not be respected or taken seriously due to their gender unless they portray a more “abrasive“ personality. With regard to community involvement, I overlooked the possibility that many incumbents may have joined the agency due to concern over their individual community or the nation’s community as a whole. This concern could very well extend itself to other forms of assistance throughout the community, not specifically related to race or to gender he.g. public education, crime, drugs). If this is the case, 54 perceived responsibility to the community or desire to solve problems therein may be a shared characteristic across individuals in the sample, regardless to race Therefore, this difference, too, may have been minimized due to the‘ sample tested. Another reason why the results may not have been as anticipated is the restriction of range for socio—economic status. It is likely that all of the examinees were relatively equal on this dimension and lived in similar neighborhoods. It is also possible that as members of minority status moved into neighborhoods which are not predominately Black they did not have individuals of similar ethnicity with whom to develop any formal or informal community groups based on race. Therefore, racial (differences may have been minimized because members of ssubgroups could possibly have a stronger commitment to the czommunity if their efforts will directly and specifically :3erve to improve conditions for their respective subgroups within that community. If no such effect is possible, Imotivation to become involved with community programs may decrease . An additional consideration is that the job incumbents, onuce assigned, are not allowed to work within a particular pr