ORGANIZATIONAL
BEHAVIOR
AND
HUMAN
DECISION
PROCESSES
40,
96-l 14 (1987)
Personnel Selection Decisions: Effects of Applicant Personality and the Letter of Reference SAMPOV. PAUNONEN,DOUGLASN.JACKSON,AND STEVENM. OBERMAN The University
of Western Ontario
Personnel selection tasks were staged in two experiments where job applicants described themselves in simulated employment interviews. The perceived competence of the job applicant was varied (high, neutral, low) in conjunction with the applicants’ personality attributes (characteristic of job incumbents or not characteristic). Judges then rated the suitability of the candidates for the described jobs. The effects of competence were very powerful in Study 1and masked the subsequent effects of the personality manipulation. In Study 2, the levels of competence were chosen to be less discrepant than in Study 1, and the personality characteristics were contrived so that opposing, rather than unrelated, types were presented as applying for the same job. The results indicated, again, a substantial main effect for the competence manipulation. There was a marked increase, however, in the personality effect, such that those applicants with personality characteristics congruent with the target job were judged as being more suitable for that position. These studies are interpreted as providing evidence for the validity and generalizability of certain personality conceptions people share about job incumbents. More generally, the data are viewed as supporting the utility of laboratory simulations of the employment interview. 0 1987 Academic PESS, Inc.
One of the hoary issues in personnel psychology concerns the type of assessment useful to personnel officers in making appropriate hiring decisions. The employment interview and the letter of recommendation continue to be two prevalent methods of evaluating prospective employees. But although the former method has been the subject of intensive empirical scrutiny (see Tenopyr & Oeltjen, 1982), the latter has not (Muchinsky, 1979). The purpose of this article is to report on two experiments that were designed to investigate aspects of both the employment interview and the letter of reference as they relate to personnel selection decisions. This research was supported by the Social Sciences and Humanities Research Council of Canada, Research Grants 410-85-0355to Sampo V. Paunonen and 410-83-0014to Douglas N. Jackson, and by the Ontario Mental Health Foundation, Research Grant 89584/86 to Douglas N. Jackson. Send reprint requests to Sampo V. Paunonen, The University of Western Ontario, London, Ontario N6A X2, Canada. 96 0749-5978187$3.00 Copyright AI1 rights
0 1987 by Academic Press, Inc. of reproduction in any form reserved.
PERSONNEL
SELECTION
97
The Employment Interview and Applicant Personality Whereas the employment interview has been the topic of vigorous empirical research, applicant personality as revealed in the interview has generally not been the specific focus (Tenopyr & Oeltjen, 1982). A few recent studies of the interview, however, have suggested that personnel hiring decisions for a variety of jobs can be reliably predicted using information about applicant personality characteristics. Rothstein and Jackson (1980) found that, in an experimental context, an applicant who portrayed himself in terms of high need for cognitive structure and orderliness and low autonomy, change, and impulsivity was consistently judged by raters as highly suitable for the job of accountant. Analogously, an applicant received high suitability ratings for the job of engineer when presenting himself as achieving, persistent, and high in the need for understanding, and low in the need for succorance and social recognition. But when judges evaluated the same applicant for the alternative job. they provided significantly lower suitability ratings, suggesting differential evaluation of job suitability, depending on the perceived fit between applicant personality and job. It is noteworthy that these sets of traits were not chosen arbitrarily for the Rothstein and Jackson (1980) study but had been identified previously (Siess & Jackson, 1970) as characteristic of people with interests resembling those of accountants and engineers, respectively. Similar findings have been reported by Jackson, Peacock, and Smith (1980) and by Rothstein (1983). Relevant to this area of research is the observation that actual employment interviewers have reliable and well-differentiated conceptions of the modal personality characteristics of several distinct occupational groups (Jackson, Peacock, & Holden, 1982). More important, it has been demonstrated that these conceptions are largely shared by people in general and, in particular, by college students. This observation is consistent with the conclusion of Bernstein, Hakel, and Harlan (1975) that substantial similarities exist between the perceptions and behaviors of student samples and professional interviewers. The Letter of Reference and Applicant Competence Although a relatively small amount of research has been published on the letter of reference and its influence on employment decisions, it has generally been observed that such reports are not predictive of job performance (Muchinsky, 1979). The common reason advanced for this predictive failure is that reference reports usually are quite homogeneous with respect to the evaluation of applicant attributes and qualifications. The lack of variability in these evaluations, with most applicants being characterized as somewhat desirable, results in poor discrimination
98
PAUNONEN,
JACKSON,
AND OBERMAN
among the candidates on job suitability (see Baxter, Brock, Hill, & Rozelle, 1981) and, hence, low correlations between reference reports and performance scores. The tendency of referees to give overly favorable evaluations of previous associates might be due in part to the perception that a negative comment would be misinterpreted as severe criticism, being so atypical. There may be some basis for this belief. For example, Tucker and Rowe (1979) have concluded that, when supplied with unfavorable information about an applicant, judges tend to attribute the cause of past success or failure in performance to the applicant rather than to external influences. Knouse (1983), however, recently reported somewhat contrary results, finding that a small amount of negative information in a letter of recommendation can actually improve the probability of an affirmative hiring decision if the applicant is otherwise presented with desirable qualities. Knouse surmised that a negative comment among many positive comments is perceived as reflecting an honest evaluation of the candidate unaffected by halo bias (see also Belec & Rowe, 1983). Purpose of the Present Research
In view of the issues concerning the processes underlying personnel selection decisions, we designed two laboratory studies to address the following questions: (a) Do interviewers view personality characteristics as differentially appropriate for different jobs? (b) Do letters of reference affect interviewer judgments if those reports refer to applicants of varying levels of competence? (c) Does perceived applicant competence interact with applicant personality in determining hiring decisions? Two studies are described in which hypothetical job candidates were presented to judges in simulated employment interviews. The judges were then asked to make several decisions about the suitability of the candidates for the specified jobs. Aspects of the job, the candidate, and the interview context were manipulated experimentally to evaluate factors potentially affecting personnel selection decisions. Both studies assessed the influence of applicant personality characteristics on hiring decisions, as well as the influence of perceived applicant competence as portrayed through a letter of reference. STUDY 1 Overview
and Hypotheses
An employment interview was staged in which college students judged the suitability of a hypothetical male applicant for a specific job. An applicant’s personality description was constructed to be either congruent with or unrelated to one of two targeted jobs-accountant or engineer.
PERSONNEL
SELECTION
99
Applicant personality information was conveyed to judges via audiotaped segments supposedly extracted from a recent job interview. The applicant’s level of job competence was manipulated to be high, moderate, or low by means of a letter of reference purportedly supplied by his previous employer. It was hypothesized that personality and target job would interact to determine suitability ratings such that higher ratings would ensue with higher perceived person-job similarity; but it was also hypothesized that this interaction would be tempered by perceived applicant competence. Thus, the highest suitability ratings should be obtained for job candidates who have job congruent personality attributes and who are highly competent. The lowest ratings should be obtained by candidates with incongruent attributes and low job competence. No hypothesis was made in the present study about the relative strength of applicant personality and applicant competence in influencing judges’ hiring/suitability decisions. Method
Subjects and Experimental Conditions Student judges were 72 male and female undergraduate psychology students receiving course credit for their participation. Subjects were randomly assigned to conditions in a 2 x 2 x 3 completely crossed factorial design, resulting in six judges per condition. Varied was the job target (accounting job or engineering job), the applicant personality type (congruent with accountant or engineer), and the level of applicant competence (high, moderate, or low). Target Jobs The two occupations chosen for evaluation in this study were those of accounting and engineering. These particular occupations were not chosen randomly but were selected following an examination of the factor analysis of personality and vocational interests reported by Siess and Jackson (1970). In that study, the Strong Vocational Interest Blank (SVIB) Accountant scale was found to define the same factor as was also defined by several personality traits measured by the Personality Research Form (PRF; Jackson, 1984). This factor was labeled impulse control versus expression and had the traits of order and cognitive structure at the positive pole and autonomy, change, and impulsivity at the negative pole. Thus, people who tend to show an interest in accounting tend to be high in order and cognitive structure and low in autonomy, change, and impulsivity. The SVIB Engineer scale defined a different factor in the Siess and Jackson study, one that was labeled as technically oriented
100
PAUNONEN , JACKSON, AND OBERMAN
achievement versus social recognition. This factor was orthogonal to the one marked by the Accountant scale and had the PRF scales of Endurance, Achievement, and Understanding as positively weighted personality variables and Social Recognition and Succorance as negatively weighted. The personality characteristics of accountants versus engineers, therefore, appear to be orthogonal or unrelated, rather than, say, direct opposites. The Siess and Jackson findings suggest to us that there are consistent and meaningful constellations of personality characteristics associated with employees in particular jobs. In the present investigation, judges were supplied with both a job title and a job description. Summary statements of job activities were provided to ensure that judges’ decisions about a candidate’s suitability were based on the same conception of the target job and were distinct from potentially inaccurate lay stereotypes. The description of the accountant job read as follows (adapted from Rothstein, 1983): The successful applicant for the position of professional accountant must be capable of keeping a complete set of records of financial transactions for a business establishment. Books must be balanced and reports compiled to show statistics such as cash receipts and expenditures, accounts payable and receivable, profit and loss, and other items pertinent to the operation of a business. The accountant will apply principles of business and finance in performing these professional activities.
Qualifications described in a professional engineering act were studied in drafting the engineer job description: The successful applicant for the position of a professional engineer must be capable of designing, composing of plans and specifications for, and evaluating various machines, vehicles, structures and apparatuses. The applicant must also be able to direct their construction, operation and maintenance. The purpose of the machines and apparatuses is for use by people in order to simplify various tasks thus reducing human errors and increasing productivity. The engineer will apply principles of mathematics, physics, and life sciences in performing these professional activities.
Care was taken to exclude any direct personality information in the job descriptions. Interview Materials Audiotaped recordings of simulated job interview excerpts were presented to judges. Based on the Siess and Jackson (1970) findings, one of the audiotapes had a male applicant (engineer type) respond to a few interviewer questions using self-descriptive statements related to being high in achievement, endurance, and understanding and low in the need for social recognition and succorance. On another tape, the same actor
PERSONNEL SELECTION
101
described himself as high in the need for cognitive structure and order and low in autonomy, the need for change, and impulsivity (accountant type). Each interview consisted of four or five interviewer questions to which the applicant responded with a few sentences. The applicant’s responses were varied systematically to convey the intended personality type. To illustrate, in response to a question asking the candidate to describe himself openly, the accountant type would reply I suppose that one thing I know about myself is that I’m the kind of person who likes to work consistently and regularly at one thing. YOUknow, I’m not interested in changing my life or my work just for the sake of change. I guess I just like the feeling of being attached to something that’s constant and predictable.
In response to the same question, the engineer type applicant would reply Well, one thing I know about myself is that 1 don’t really care what other people think of me. I mean, I always do my job and 1 do it well, probably better than most, but I’m just not the kind of person who is concerned about being socially proper. 1 guess I’m just not worried about impressing people. I like to work to please myself.
The responses of job candidates to the interview questions were constructed after studying the content of both the self-descriptive statements of the PRF, Form-E, and the PRF scale descriptions (Jackson, 1984, Table 1). These tapes were first employed in an independent study by Rothstein and Jackson (1980). Competence Manipulation
An applicant’s described job competence was varied for different experimental conditions. Subjects were told that, in addition to the information presented in the interview excerpts, they were to study a letter of reference from the applicant’s previous employer in arriving at their decisions about the candidate’s appropriateness for the target job. These letters of reference were varied for different judges to be relatively positive, neutral, or negative evaluations. This was achieved by using a standardized recommendation form as part of the letter of reference. A onepage checklist was printed for this study which contained seven items. Each item had five labeled rating categories (see Stone & Meltz, 1983, p. 23 1). The seven items on which an employer or supervisor was ostensibly requested to rate an employee were quality of work, quantity of work, technical job knowledge, the ability to plan and direct work, appearance, the ability to get along with others, and service awareness. In each condition of the present study, the applicants’ presumed ratings were held constant (and relatively neutral) on all variables except for (a) job knowl-
102
PAUNONEN,
JACKSON,
AND OBERMAN
edge and (b) the ability to plan and direct work. The positive letter of reference was constructed so that the highest category was checked for these two items (“expert; has superior knowledge” and “self-sustaining; plans and carries out own work in a superior manner,” respectively), whereas the negative letter had the lowest categories marked (“very limited knowledge; needs frequent assistance” and “always waits to be directed”). The neutral letter had the center categories checkmarked (“possesses acceptable knowledge” and “requires occasional work direction”). A covering letter, complete with appropriate company letterhead, was included in the materials. This was ostensibly from the prospective employer requesting the previous employer to complete the recommendation form. Procedure
and Dependent
Variables
Subjects were assigned randomly to one of the experimental conditions of the 2 x 2 x 3 factorial design. Each judge studied a job description (accountant or engineer), read a letter of recommendation (positive, negative, or neutral) and listened to taped segments from a job interview where the applicant divulged only information about his personality (congruent with accountant job or engineer job). Judges were then asked to complete several ratings concerning the hypothetical job application. These were (a) applicant suitability, (b) expected success, (c)judge’s willingness to hire for target job, (d) expected satisfaction with job, (e) expected satisfaction with co-worker interactions, (f) job knowledge, and (g) ability to work independently. The latter two items were included as checks to determine whether or not judges correctly perceived the applicant competence manipulation. All judgments were recorded on 7-point rating scales. Results and Discussion
To determine the extent to which the various experimental groups appropriately perceived the manipulation of applicant competence, the dependent variables of job knowledge and ability to work independently were examined. The mean job knowledge ratings on the 7-point scales for the groups with positive, neutral, and negative letters of reference were, respectively, 6.04, 3.92, and 2.67. All pairwise contrasts were significantly different at the .Ol level. The corresponding means for the ratings of ability to work independently were 5.00, 3.46, and 2.33. Only the neutral letter versus negative letter contrast was not significant (p > .Ol). Overall, it appears that the competence manipulation was appropriately perceived by the student judges. The seven dependent variables were intercorrelated. The results are presented in Table 1. The variables of job suitability, willingness to hire,
103
PERSONNEL SELECTION
TABLE 1 INTERRCORRELATIONSOFSEVENRATINGSOFHYPOTHETICALJOBAPPLICANTSFOR STUDY]
1. 2. 3. 4. 5.
Suitability of applicant Expected success at job Willingness to hire Expected satisfaction Expected co-worker relations 6. Job knowledge 7. Ability to work independently
1
2
66 65
57
30
21 12 62 60
19 58 42
3
4
5
6
25 20 52 43
34 16 03
15 04
52
Note. Decimals omitted and N = 72. For r > .31. p < .Ol
and expected success were highly intercorrelated, with a mean value of .63. These variables were accordingly combined into a composite variable by summing across the three for each judge. The average suitability composite score for each experimental condition is given in Table 2. Inspecting the means in Table 2, it is apparent that there is evidence for the hypothesized relation between perceived job competence and decisions about candidate suitability. The more competent candidate generally received higher ratings of suitability. The correlations between the suitability composite and ratings of job knowledge and ability to work independently support this observation with the respective values of .79 and .68. To explore further this and other hypothesized effects, the composite suitability data were analyzed in a 2 x 2 x 3 completely crossed, fixed effects ANOVA. The summary table is presented in Table 3. The letter of reference exerted a powerful main effect on overall suitability judgments as evident in the large F ratio for that factor (p < .OOOl). TABLE 2 MEAN COMPOSITERATINGS OFJOB SUITABILITY FORSTUDY 1, BY CONDITION Target job Personality type
Engineer
Accountant
Engineer Accountant
High competence 14.16 14.16 13.17 15.15 Moderate competence
Engineer Accountant
Il.82 8.67
Engineer Accountant
9.99 8.01
11.34 11.82
Low competence 9.15 8.85
Note. Composite scores can range from 3 to 21 and a higher score indicates higher overall ratings of suitability.
104
PAUNONEN,
JACKSON,
AND OBERMAN
This confirms the hypothesis that applicant job competence will in part determine decisions about his or her desirability for that job (see also Rasmussen, 1984). The predicted personality type x job target interaction, although in the expected direction, was not significant at the .05 level (p = .062). That is, whereas the overall ratings of suitability were generally higher for the accountant personality type applying for the accountant job and for the engineer personality type applying for the engineer job than when person-job characteristics were incongruent, this effect was not as strong as it was found to be in other studies where applicant competence was not manipulated (e.g., Rothstein & Jackson, 1980). None of the other effects approached statistical significance 0, < .05) in this analysis. The analysis of variance was also carried out on the ratings of (a) expected employee satisfaction with job and (b) expected satisfaction with co-workers. The letter of reference had a marginally significant effect on job satisfaction ratings (F(2,60) = 3.12, p = .052), such that the more competent candidates were generally expected to be somewhat more satisfied with their jobs. No other effects were found in this analysis. For the variable of satisfaction with co-worker interactions, the only statistically significant effect was that due to personality type; the accountant personality was judged to have a greater likelihood of getting along with his fellow employees (F(1,60) = 8.67, p < .Ol), regardless of perceived competence or target job. This effect was probably due to the fact that the engineer type applicant had described himself in the interview as being low in the need for succorance or support from others. That applicant may have been unintentionally depicted as someone who is cool toward interpersonal relationships. Returning to the composite suitability ratings, there are two possible reasons for the relatively weak personality x job target interaction in this study. First, the occupations of accounting and engineering define orthogonal clusters of attributes, whereas the characteristics for some occupations define opposite attributes (Siess & Jackson, 1970). This would TABLE ANOVA
3
SUMMARYTABLE FORCOMPOSITESUITABILITY RATINGS FORSTUDY 1 Source
df
Mean square
F ratio
Personality type (A) Target job (B) Competence (C) Personality x Job (A x B) Within cells
1 1 2 1 60
12.50 10.89 163.72 26.89 7.49
1.67 1.46 21.84* 3.60
Note. All interaction terms not illustrated were nonsignificant (p > .05). *p<.OOO1.
PERSONNEL
SELECTION
105
mean that the contrast between the ideal personality types for the present pair of occupations would not be as salient as it could be. Consequently, judges might be more likely to ascribe higher suitability ratings for the incongruent conditions in this study as opposed to a study where target jobs called for more discrepant personality types. The second reason for the relatively small personality x job interaction in this study is conceivably due to the massive main effect of the competence manipulation. An inspection of Table 2 indicates that high overall ratings of suitability were provided by judges for all conditions in which the applicant was described as extremely competent, regardless of person-job match. Analogously, relatively low ratings are apparent for all applicants described as low in competence. It appears that someone who is portrayed as having superior job knowledge and as being self-sustaining is judged to be a good candidate for either job, whereas someone with limited knowledge of the job and who always waits to be directed is seen as generally undesirable. Because of this judgmental process, the predicted job x personality interaction should only occur in the neutral competence conditions. Although mean ratings in Table 2 tend to support this interpretation, we addressed the issue directly in a second study. STUDY 2 Overview
and Hypotheses
An employment interview was staged similar to that of Study 1. Again, college students judged the suitability of a hypothetical applicant (female, in this case) for one of two specific jobs. Relative to Study 1, however, the personality traits characteristic of the two jobs were more distinct. New audiotapes of interview segments were created for personality types associated with the occupations of office work and author-journalism. Competence was again manipulated to be high, moderate, or low by means of a letter of reference, but differences in the level of competence between conditions were contrived to be smaller than those differences of Study 1. It was hypothesized that the personality x target job interaction would be larger than that of the previous study and that the competence effect would be smaller. Thus, we proposed that decisions about applicant suitability can be manipulated by systematically varying the differential salience of personality and competence information. Method
Subjects and Experimental Conditions Student judges were 132 male and female undergraduate psychology students receiving course credit for their participation. Subjects were randomly assigned to conditions in a 2 x 2 x 3 completely crossed facto-
106
PAUNONEN,
JACKSON,
AND OBERMAN
rial design, resulting in 11judges per condition. Varied was the job target (office worker job or author-journalist job), the applicant personality type (congruent with office worker or author-journalist), and the level of applicant competence (high, moderate, or low). Target Jobs The two occupations chosen for evaluation in this study were those of office work and author-journalism. The reason for choosing two occupations different from those of Study 1 was to assess the generalizability of the previous findings to new jobs that are more contradictory than are accountant and engineer in terms of the presumed modal personality types of employees. Again, the specific occupations were chosen following further examination of the Siess and Jackson (1970) factor analysis of personality and vocational interests. In that study, it was reported that the Author-Journalist and Office Worker scales of the SVIB defined the opposite poles of a factor also defined by several personality traits. One pole of this bipolar factor was marked by the SVIB scales of AuthorJournalist, Lawyer, and Advertiser, and by the PRF scales of Impulsivity, Change, and Autonomy. The other end of the factor revealed a proximity among the SVIB scales of Office Worker and Accountant and the PRF scales of Order and Cognitive Structure. Thus, individuals who tend to show a similarity to, say, journalists, also tend to describe themselves as rather impulsive, changeable, and autonomous, and lacking in orderliness and in the need for predictability in their environment. The opposite traits would characterize those having an interest in office work or accounting. Although we believe that most people are familiar with the occupations of office work and author-journalism, and a substantial amount of information about the jobs is conveyed by the titles alone, we supplied each judge with a brief job description. The respective job titles used on the experimental materials were those of payroll clerk and newspaper reporter. Interview Materials and Competence Manipulation Audiotaped transcripts of segments of a simulated job interview were presented to student judges. A dialog of each interview was constructed to portray a female applicant who represented either a newspaper reporter personality type, as identified by Siess and Jackson (i.e., high in impulsivity, need for change, and autonomy, and low in orderliness and need for cognitive structure), or a payroll clerk type having the oppositive traits. Each interview was constructed in a manner similar to that of Study 1. The same seven-item recommendation form used in Study 1 was included as part of the presumed letter of reference. Moreover, the same
PERSONNEL
SELECTION
107
two items, job knowledge and ability to plan and direct work, were used in the competence manipulation. Less extreme ratings, however, were attributed to the high and low competence applicants. Recall that these items were rated on 5-point scales and, in the first study, the extreme endpoints were used for the ratings of high and low competence. In the present study, the second and fourth rating categories were used for the two extreme groups. Thus, the most competent candidate was described as someone who has “well-rounded knowledge; seldom needs assistance,” and who “plans and carries out work well; requires little supervision.” The least competent candidate was described as someone whose “knowledge is adequate to perform minimum job requirements” and who “carries out only the most obvious tasks without follow-up.” The neutral candidate description was the same as that of the preceding study. Procedure and Dependent Variables Student judges were randomly assigned to one of three competence conditions (positive, neutral, or negative letter of reference). For each condition, half the judges were presented with the author-journalist type candidate and half with the office worker type candidate. In addition, half of each group was told that the job for which the candidate was applying was that of newspaper reporter and the other half was told the job was that of payroll clerk. Brief job descriptions were supplied. The research design, therefore, was a completely crossed 2 x 2 x 3 factorial experiment. After judges studied the written materials and listened to the taped segments, they were asked to complete several ratings concerning the hypothetical job application. These ratings were the same as those used in Study 1, namely, (a) applicant suitability, (b) expected success, (c) judges’ willingness to hire for target job, (d) expected satisfaction with job, (e) expected satisfaction with co-worker interactions, (f) job knowledge, and (g) ability to work independently. The latter two items, again, were included as checks to determine whether or not the judges correctly perceived the manipulation of applicant competence. In this study, subjects were further requested to estimate the personality characteristics of the described job applicants. After completing the ratings of job suitability, judges were given 176 self-descriptive personality statements. These statements constitute a half-form of the PRF in which the same trait dimensions are represented but with half the number of items in each scale. Judges were asked to estimate how characteristic is each statement of the interviewee. Their responses were recorded on 9-point rating scales where a 9 indicated a judgment that the statement was extremely characteristic of the applicant, and a rating of 1 indicated the statement was extremely uncharacteristic. It was expected that the
108
PAUNONEN,
JACKSON,
AND OBERMAN
personality ratings of the office worker type would differ from those of the author-journalist type in a direction consistent with their respective personality descriptions. Results and Discussion Ratings of Personality In order to determine whether or not the personality type manipulation was correctly perceived by the student judges, personality ratings on the PRF items were combined into their respective scales and compared for the two applicant description conditions, averaged over the other factors. As illustrated in Table 4, the mean personality scale scores for the office worker and author-journalist types differed appreciably. And these differences were in the expected direction, overall. For example, the authorjournalist type was rated as significantly higher in the marker scales of Autonomy, Change, and Impulsivity (i.e., traits implied by the interview dialog) and lower in the marker scales of Cognitive Structure and Order than was the office worker type. Other differences were found on the nonsalient scales and, as can be seen in Table 4, most of these differences TABLE 4 MEAN PRF TRAIT RATINGS OF Two JOB CANDIDATESEXPRESSINGOPPOSITE
PERSONALITYCHARACTERISTICS
Abasement Achievement Affiliation Aggression Autonomya Change” Cognitive structure” Defendance Dominance Endurance Exhibition Harmavoidance ImpulsivityO Nurturance Order” Play Sentience Social recognition Succorance Understanding
Offke worker
Author-journalist
Difference
5.03 4.92 5.01 4.62 3.84 2.81 6.11 5.19 4.00 5.07 3.98 7.48 3.32 5.35 7.85 3.80 4.89 5.41 5.61 5.28
4.58 4.35 4.98 5.02 6.41 6.95 3.65 5.15 4.66 4.58 5.24 4.02 7.01 4.71 2.55 5.90 5.53 4.53 3.77 4.50
0.45 0.57 0.03 -0.40 - 2.57* -4.14* 3.12* 0.04 - 0.66 0.49 - 1.26* 3.46* -3.69* 0.64* 5.30* -2.10* - 0.64* 0.88 1.84* 0.78*
Note. Scores can range from 1 to 9. N = 66. PRF = Personality Research Form. a Scale used in personality descriptions. * p < .OOl.
PERSONNEL
SELECTION
109
were in the direction expected by the general constellation of personality traits associated with the respective job applicants. (Only differences exceeding the .OOl level are interpreted here because of the large number of t statistics computed.) For instance, the office worker was perceived to be lower than the author-journalist in exhibitionism while being higher in harmavoidance. These data support the conclusion that the characteristics of office worker and author-journalist are correctly perceived by student judges in experimental simulations such as we have described. Ratings of Suitability The variables of job knowledge and ability to work independently were analyzed for evidence of the success of the competence manipulation. On 7-point scales, the mean job knowledge ratings for the positive, neutral, and negative letters of reference were, respectively, 4.27, 4.14, 3.41. Note that these values are less discrepant than those reported in Study 1 where competence was manipulated to be more extreme. Only the difference between the positive and the neutral letter of reference was not statistically significant (p < .05). For the variable of ability to work independently, the respective mean ratings were 4.91, 3.77, 3.46. Only the difference between the neutral and the negative letter of reference was not significant at the .05 level. It appears, therefore, that the manipulation of competence was successful despite the fact that this manipulation was more subtle than that in Study 1. The seven independent variables were intercorrelated. The three variables of job suitability, willingness to hire, and expected success correlated with a mean value of 58. As in Study 1, therefore, these three variables were summed to form a composite measure of job suitability. The mean composite suitability scores, averaged within experimental conditions, are presented in Table 5. The means of Table 5 indicate some effect of the competence manipulation. In general, ratings of suitability were higher for candidates in the high-competence conditions and declined as the level of competence declined. There is also an indication in Table 5 of the hypothesized interaction between personality type and target job. For each level of competence, the ratings of suitability are generally greater for the personalityjob congruent conditions than for the incongruent conditions. These effects were explored statistically in a three-factor, fixed effects ANOVA. The ANOVA summary table is reproduced in Table 6. Note that the competence manipulation, although less extreme than that of Study 1, produced a highly significant main effect on suitability ratings. Note also the statistically significant personality x job interaction. In contrast to the data of Study 1, this interaction strongly supports the hypothesis that job candidates having personality characteristics con-
110
PAUNONEN,
JACKSON, AND OBERMAN
TABLE 5 MEANCOMWSITERATINGSOFJOB SUITABILITYFOR~TUDY~, BYCONDITION Target job Personality type Author-journalist Office worker Author-journalist Office worker Author-journalist Office worker
Author-journalist
Office worker
High competence 10.64 12.09 13.55 11.18 Moderate competence 10.81 10.00 13.64 10.91 Low competence 8.00 10.09 11.82 9.27
Note. Composite scores can range from 3 to 21, and a higher score indicates higher overall ratings of suitability.
gruent with those of job incumbents are perceived as being more suitable for those jobs than are applicants having characteristics that are incongruent with the job. There was one other significant effect in this analysis, which involved a main effect of personality type. An inspection of Table 5 indicates that the office worker personality type was generally perceived as being more suitable than the author-journalist type for either job. This effect could be due to the differential desirability of the personality characteristics associated with each type (cf. Jackson et al., 1980, p. 305). None of the other effects approached statistical significance (p < .05) in this analysis. Analyses of variance were applied to the variables of expected job satisfaction and expected satisfaction with co-worker interactions. Although no significant main effects or interactions emerged with respect to the latter variable, this was not the case for ratings ofjob satisfaction. A main effect of personality type (F(1,120) = 15.03, p < .OOl) indicated that TABLE 6 ANOVA SUMMARYTABLEFORCOMPOSITESUITABILITYRATINGSFORSTUDY~ Source
df
Mean square
F ratio
Personality type (A) Target job (B) Competence (C) Personality x job (A x B) Within cells
1 1 2 1 120
69.82 22.09 50.89 98.45 6.51
10.72* 3.39 7.81** 15.12**
Note. All interaction terms not illustrated were nonsignificant (p > .05). *p < .Ol. ** p < ,001.
PERSONNEL
SELECTION
111
judges perceived the office worker type to be more satisfied with her work than the author-journalist type, for either job. Furthermore, a highly significant personality x job interaction (F(l,120) = 34.35, p < .OOOl)showed that the personality type postulated to be congruent with the job was expected to be more satisfied in that position than was the personality type that was incongruent with the job. Note that the competence manipulation had no significant effect on ratings of expected job satisfaction. GENERAL
DISCUSSION
Although the employment interview has received considerable research attention in recent years, studies directed at applicant personality variables have been few. One possible reason for the relative dearth of research on the role of personal attributes of applicants in affecting hiring decisions may be reviews of past research which purport to document meager associations between personality and job performance measures (e.g., see Guion & Gottier, 1965). In response to these reviews, it has been argued that the low degree of association reported is attributable in part to the predictor measures generally used in such research. For example, in Guion’s (1976; Guion & Gottier, 1965) reviews of the literature he cites several studies where personality measures failed to correlate with job success. The measures, however, included the Rorschach and the TAT as well as several personality inventories that have dubious psychometric properties. Moreover, in many cases measures of personality have been used in a shotgun approach to prediction where several variables are evaluated simultaneously and without a priori theoretical justification; for example, using the MMPI scales to predict accident rates in truck drivers (Parker, 1953). These and related problems have been recognized by Guion and others (e.g., Rothstein, 1983) but have nevertheless probably discouraged research on personal attributes vis-a-vis personnel selection. The results of the present studies have largely supported some prior research and theorizing in suggesting that applicant personality characteristics are interpreted by job interviewers as important components in the hiring process. The higher the perceived person-job match, the greater the ratings about the candidate regarding suitability for the job and expected success and the greater the interviewer’s willingness to hire. Our findings, of course, do not deal directly with the validity of interviewer perceptions with respect to actual performance on the job; that is, personality may have little influence in determining job performance (Guion, 1976). But it seems reasonable to expect that, all else being equal, people sharing preferences, attitudes, temperamental characteristics. and behavioral dispositions might be more productive
112
PAUNONEN,
JACKSON,
AND OBERMAN
working together than when such commonalities are absent. Personality conflicts would be minimized while mutual cooperation and morale would be increased. It also seems reasonable to expect that certain personality characteristics are directly relevant to the successful performance of some job activities and not others. Achievement and endurance, for example, might be more important for the job of engineer than for that of secretary. If our analysis is correct, it would suggest that the hiring decisions of our subjects that are based on personality information alone are logical decisions, formulated in the attempt to be accurate in predicting how well a candidate will perform on the job. If personality factors in fact have no relation to job performance for the occupations we have investigated, then the present studies must be interpreted in terms of inappropriate biases that operate in shaping personnel selection decisions, at least with respect to ratings of expected job success. But although the preference for selecting a person of a certain type for a specific job may be ill-founded from the perspective of performance, it cannot be said that the preference is based on an invalid stereotype about the personality characteristics of the occupational group. The personality traits we chose as representing accountants, engineers, payroll clerks, and newspaper reporters were not based on intuition or folklore. Rather, the modal types were determined empirically (Siess & Jackson, 1970) by considering the characteristics of people who displayed substantial similarity to persons in those -occupations. Inasmuch as the college student judges in our studies were able implicitly to identify these person characteristics as being associated with job incumbents, their ratings have some degree of accuracy. These ratings, furthermore, reflect constructions not only of the students employed in this research, but of professional interviewers as well (Jackson ef al., 1980, 1982). The manipulation of applicant competence in Study 1 produced an extremely powerful effect on judgments of applicant suitability, so powerful in fact that the applicant personality characteristics only showed marginal evidence for the predicted personality x job interaction. In the second study, where differences in competence level were more subtle and differences in personality were more salient, the competence manipulation largely maintained its influence on suitability ratings, whereas the personality manipulation showed a remarkably enlarged effect. From the perspective of the decisions of actual interviewers, this finding is interesting in that it suggests reference reports (a) are interpreted as important components in predicting future success at the job and (b) can override concerns about person-job fit when competence is perceived to be extremely high or extremely low. The validity of our assertions, of course, depends on the generalizability of our experimental simulation with respect to realistic job settings. We also note that we made no formal at-
PERSONNEL SELECTION
113
tempt to determine statistically the relative sizes of the experimental effects, as these are largely dependent on the specific manipulations used in a study. In Study 2 compared with Study 1, for instance, the competence effect was deliberately mitigated by using less extreme levels for the highand low-competence conditions, and the personality x job interaction was increased by purposely choosing to investigate two jobs for which the modal personality types of incumbents show more contrast. We also add here that the distinction between competence and job experience may be an important one, as the latter variable has been reported to have little effect on personality-based hiring decisions (Jackson et al., 1980; cf. Rasmussen, 1984). In summary, our investigation can be interpreted as supporting the conclusions that, for certain occupations, student judges (and by inference, actual interviewers) (a) have reliable and largely accurate conceptions about the personality characteristics of prototypical job incumbents, and (b) use this personality information in both realistic and artificial employment contexts in determining the suitability of job applicants, but (c) modify their judgments of suitability appropriately to account for differences in applicant competence. Our data also lead us to conclude that a point is ultimately reached on the competence continuum where the level is perceived to be so high (or so low) that the applicant’s personality characteristics become nonsalient by contrast and cease to have an effect on hiring decisions. We speculate that this critical level of competence/incompetence varies from one occupation to another. For example, although personality characteristics may fall in the shadow of job knowledge in contributing to the effective performance of an accountant or engineer, the reverse might be true of a guidance counselor or minister. The idea raises interesting possibilities for future research. REFERENCES Baxter, J. C., Brock, B., Hill, P. C., & Rozelle, R. M. (1981). Letters of recommendation: A question of value. Journal of Applied Psychology, 66, 296-301. Belec, B. E., & Rowe, P. M. (1983). Temporal placement of information, expectancy, causal attributions, and overall final judgments in employment decision making. Cunadian Journal of Behavioral Science, 15, 106- 120. Bernstein, V., Hakel, M. D., & Harlan. A. (1975). The college student as interviewer: A threat to generalizability? Journal ofApplied Psychology, 60, 266-268. Guion, R. M. (1976). Recruiting, selection, and job placement. In M. D. Dunnette (Ed.), Handbook of industrial and organizational psychology (pp. 777-828). Chicago: Rand McNally. Guion, R. M., & Gottier, R. F. (1965). Validity of personality measures in personnel selection. Personnel Psychology, 18, 135- 164. Jackson, D. N. (1984). Personality Research Form manual. Port Huron, MI: Research Psychologists Press. Jackson, D. N., Peacock, A. C., & Holden, R. R. (1982). Professional interviewers’ trait
114
PAUNONEN,
JACKSON, AND OBERMAN
inferential structures for diverse occupational groups. Organizational Behavior and 29, l-20. Jackson, D. N., Peacock, A. C., & Smith, J. P. (1980). Impressions of personality in the employment interview. Journal of Personality and Social Psychology, 39, 294-307. Knouse, S. B. (1983). The letter of recommendation: Specificity and favorability of information. Personnel Psychology, 36, 331-341. Muchinsky, P M. (1979). The use of reference reports in personnel selection: A review and evaluation. Journal of Occupational Psychology, 52, 287-297. Parker, J. W. (1953). Psychological and personal history data related to accident records of commercial truck drivers. Journal of Applied Psychology, 37, 317-320. Rasmussen, K. G. (1984). Nonverbal behavior, verbal behavior, resume credentials, and selection interview outcomes. Journal of Applied Psychology, 69, 551-556. Rothstein, M. (1983). Predicting job performance from impressions of personality. Unpublished doctoral dissertation, The University of Western Ontario. Rothstein, M., & Jackson, D. N. (1980). Decision making in the employment interview: An experimental approach. Journal of Applied Psychology, 65, 271-283. Siess, T. F., & Jackson, D. N. (1970). Vocational interests and personality: An empirical integration. Journal of Counseling Psychology, 17, 27-35. Stone, T. H., & Meltz, N. M. (1983). Personnel managemenf in Canada. Toronto: Holt, Rinehart & Winston. Tenopyr, M. L., & Oeltjen, P. D. (1982). Personnel selection and classification. Annual Review of Psychology, 33, 581-618. Tucker, D. H., & Rowe, P. M. (1979). Relationship between expectancy, causal attributions, and final hiring decisions in the employment interview. Journal of Applied PsyHuman Performance,
chology,
64, 27-34.
RECEIVED: December 3, 1985