Journal of Occupational Psychology (1990), 63, 63-76 @ 1990 The British Psychological Society

Printed in Great Britain 63

Decision making in the graduate selection interview: A field study

Neil Anderson* MRCIESRC Social and Applied Psychology Unit, University of Shdfild, SheffieldSI 0 ZTN,

UK

Viv Shackleton Aston Business School, Birmingham

The influence of candidate non-verbal behaviour (NVB) upon interviewer impression formation in the graduate selection interview was investigated using a modified Bmnswik lens model approach. The impact of three dysfunctions in interviewer decision making found to be salient in previous research was also examined: similar-to-me effect, personal h i n g bias and prototype bias. Thirty-eight graduate interviewers participated in the study and completed assessments on 330 interviewees for 14 diverse occupational groups. Interviewer outcome decisions were found to be substantially linearly dependent upon impressions of candidate personality which were in turn linearly dependent upon candidate facial area NVBs. Overall evaluations correlated r = .50 with ratings of similarity-to-self and r = .64 with ratings of personal liking, indicating the pervasive bias of these two factors. Virtually no significant differences were found in accepted candidate personality profiles across the 14 job functions, suggesting the marked impact of prototype bias upon interviewer decision making. Implications for graduate recruit- ment practices, particularly usage of the selection interview, are discussed.

The popularity of the annual spate of visits by UK employers to universities and polytechnics in order to conduct initial interviews with graduates shows few signs of diminishing (Connor & Prior-Wandesforde, 1986). There is considerable circumstantial evidence, however, that graduate selection procedures lack predictive validity. Unaccept- ably high turnover rates have been reported in two recent large-scale surveys (Brennan & McGeevor, 1987; Parsons, 1985), with over halfofall graduates having changed job twice or more in the three years following graduation. Reasons for this high turnover appear to centre upon graduands’ beliefs that their first job is a ‘stepping-stone’ to another position (Brennan & McGeevor, 1985), coupled with low job satisfaction caused at least in part by unrealistically high job expectations (Hutt & Parsons, 198 1; Mabey, 1986; Williamson, 198 1).

The ‘milkround’ interview constitutes the first stage of screening applicants for many employers of graduates and research focusing upon interviewer decision-making processes

Requests for reprints.

64 Neil Anderson and Viu Shackleton is therefore valuable as such studies promise to elucidate dysfunctions in information processing. Indeed, existing research testifies to a range of errors in information processing by interviewers, all of which combine to reduce the predictive efficacy of assessments (Arvey & Campion, 1982). Three dysfunctions found to be salient in previous research are examined in this study: similar-to-me-effect, personal liking bias and prototype bias.

Similar-to-me e&t

A consistent finding across several studies is that candidates with similar biographical backgrounds, attitudes and perceived personalities to the interviewer are rated more favourably than those who differ in these respects (Baskett, 1973; Dalessio & Imada, 1984; Peters & Terborg, 1975; Wexley & Nemeroff, 1974). Rand & Wexley (1975) term this bias the ‘similar-to-me effect’ and conclude that the tendency for interviewers to recruit in their own self-image exerts substantive influence upon their decision-making process. Unfortunately, no studies have considered this effect in the context of the UK graduate interview and so no definitive conclusions may be drawn regarding its impact upon interviewer decision making.

Personal liking biaj

Another bias affecting outcome decisions is the interviewer’s personal liking for the candidate. Keenan (1977), in a major study of graduate interviews, found positive and significant associations between interviewer ratings of liking and overall evaluations. Overall ratings correlated r = .5 1 and r = .54 with ratings of personal liking and liking as a potential work colleague. As Keenan points out, the ‘common stereotype of the good interviewer is that of the dispassionate observer who remains emotionally detached while carefully evaluating each candidate’. On the basis of these findings, this stereotype appears to be fundamentally inaccurate. More recent research by Graves & Powell (1988) supports this assertion. Interviewer ratings of the probability of a job offer being made correlated Y = .35 with personal liking evaluations and r = .45 with ratings of perceived simi- larity, corroborating the influence of personal liking bias and similar-to-me effect upon interviewer decision making.

Prototype bias

The third bias concerning this study may be termed lprototype bias’ (Anderson, 1988 a). Controversy exists over whether interviewers possess distinct notions of occupation- specific stereotypes of suitable personalities for particular job functions, or whether interviewers merely utilize a general ‘prototype’ (Cantor & Mischel, 1977) of the model candidate regardless of which occupational group is the target for recruitment.

Jackson, Peacock & Holden (1982) surveyed 132 recruiters and asked them to imagine a typical employee in 15 diverse occupational groups. Respondents then rated the applicability of a series of personality traits to these imaginary individuals. Marked differences were found between the perceived personalities for each occupational group. Jackson et al . argue that these findings support the view that interviewers possess occupation-specific stereotypes of suitable personality.

Decision making in interviews 65 Diametrically opposed to this conclusion is the viewpoint that, since interviewers are

attempting to cope with an overload of verbal, non-verbal and documented information, they simplify their complex decision-making task by screening against perceptions of a universally appropriate personality prototype regardless of the occupational group in question (Anderson, 1988 a) .

Since no studies could be located which address the influence of prototype bias in the graduate interview, this potential dysfunction in interviewer decision making was incorporated into the present study design.

Candidate non-verbal behaviour

In addition to examining the influence of the similar-to-me effect, personal liking bias and prototype bias, this study investigated the role of candidate non-verbal behaviour (NVB) in interviewer impression formation. In an earlier laboratory experiment in this research programme (Anderson, 1988 b) , it was found that candidate NVB, particularly the facial area NVBs of eye-contact and facial expressions, exerted significant influence upon interviewer impression formation. Clearly, however, the external validity of such laboratory simulations of the interview needs to be demonstrated and the findings ratified against field study research using actual graduate interviews (see also Gifford, Fan N. G. & Wilkinson, 1985). The field study reported here thus transposes elements of the experimental design into the setting of real-life interviews in order to evaluate the perceptual links between candidate NVB and interviewer impression formation.

Method

Study design

A modified Brunswik lens model (Brunswik, 1956) was used to operationalize the study design as shown in Fig. 1. The model consists of four sets of variables:

(a) candidate non-verbal behaviour; (b) interviewer impressions of candidate personality; (c) interviewer perceptions of occupational group suitability; (4 interviewer outcome decision making.

Stage 1 of the lens model, ‘impressioti formation perceptual links’, represents relationships between candidate NVB and interviewer impressions of personality. Stage 2, ‘occupational group suitability perceptual links’ models the relationships between personality impressions and perceived suitability for different occupations. The final part of the model depicts the relationship between perceived suitability and outcome decision making, although it is of course acknowledged that factors other than impressions of personality affect outcome decisions. Perceived personality similarity and interviewer liking for the candidate are incorporated as moderator variables in the model on the basis of past study findings and therefore stand in relation to all other personality factors.

Sample

A random sample of 69 organizations conducting graduate interviews was contacted through the Aston University Careers Service and a total of 38 recruiters from 26 organizations agreed to participate in the study.

m (j\

Suitability Candidate Interviewer impressions Non-verbal behaviour -> of candidate personality -} decision making -) Outcome decision

Stage 1: ImDression formation DerceDtual links

Stage 2: Occuoation

\ Eyecontact Unsuccessful

Hand gestures

Head movements Immature

Enthusiastic Posture /

Insensitive

Pleasant

Postural changes

Facial exoressions \

Accepthject candidate

Similar-to-me

Liked

b

Figure 1. Modified Brunswik lens model of interviewer impression formation and suitability decision making.

Decision making in interviews 67 Twenty-three of the 38 participant interviewers were personnel managers, 15 were line managers. Numbers of graduates interviewedperannum varied, with 18 interviewers conducting less than 50 interviews each year, 11 conducting between 5 1 and 150 interviews, and 9 interviewers conducting over 150 interviews each year. All participants were conducting initial graduate interviews usually lasting between 30 and 45 minutes with final year students either at the university or on their own premises. Interview formats varied between highly structured, guided interviews and unstructured, informal interactions, but most interviews were semi- structured.

Procedtlre

Instructions to interviewers requested that immediately following each interview they should complete the two-part Interviewee Assessment Form (IAF) developed for the earlier laboratory experiment. Briefly, the first part measured six aspects of candidate NVB: eye-contact, hand gestures, head movements, posture, postural changes and facial expressions. The second part measured 1 1 personality factors: 1) boring-interesting, (2) relaxed-nervous, (3) weak-strong, (4) generally unsuccessful in life-generally successful in life, ( 5 ) active-passive, ( 6 ) immature-mature, (7) enthusiastic-unenthusiastic, (8) insensitive-sensitive, (9) pleasant-unpleasant, (10) honest-dishonest, and (1 1) dominant-submissive. Variables on both parts of the IAF were rated by interviewers on nine-point Likert-type scales. The IAF was supplemented with a number of overall assessment scales evaluating: ( i ) personal liking, (ii) perceived personality similarity, (iii) occupational group suitability and (iv) an additional dichotomous rating of the accepdreject outcome decision. It should be noted that this outcome decision related to second interview stage only and did not constitute the ultimate hiring decision.

Response rates

A total of 358 usable IAFs were returned which were categorized into 14 occupational groups as shown in Table 1. The occupational group titles were derived from the job titles given by participants. Several IAFs were discarded as being unsuitable for these 14 occupational groups, thereby reducing the number included to 330.

Table 1. Response rates across occupational groups

Occupational group N

Royal Air Force officer 25 Royal Navy officer 27 Army officer 20 Police officer 80 Mechanical engineer 19 Electronics engineer 27 Industrial chemist 19 Pharmacist 11 Optician 21 Marketing trainee 19 Retail sales trainee 15 Banker 12 Accountant 14 Computer programmer 21 Total 330

68 Neil. Anderson and Viu Shackleton

Results

Difjferences between accepted and rqected candidates

Initial analysis of the IAFs was undertaken to establish whether any significant differences existed between interviewer ratings of accepted and rejected candidates. Table 2 displays means, standard deviations and t test values resulting from this analysis.

Table 2. Differences in interviewer ratings of accepted and rejected candidates

Accepted group Rejected group (N = 203) (N = 127)

Mean SD Mean SD t P Eye contact Hand gestures Head movements Posture Postural changes Facial expressions

Boring Relaxed Weak Unsuccessful Active Immature Enthusiastic Insensitive Pleasant Honest Dominant

Liked Similar-to-me

Occupational group suitability

6.52 4.67 5.00 5.66 4.57 6.26

6.44 4.01 6.09 6.52 3.35 6.60 3.26 6.03 2.85 2.5 1 4.21

2.93 5.23

3.32

1.57 1.82 1.65 1.85 1.77 1.52

1.52 1.81 1.53 1.42 1.50 1.56 1.55 1.36 1.20 1.27 1.45

1.31 1.85

1.59

5.61 4.68 4.73 5.59 4.02 5.13

4.36 5.16 4.31 4.85 5.09 5.13 5.12 5.61 3.91 2.91 5.42

4.72 6.78

6.80

2.07 4.29 2.18 -0.01 1.93 1.32 2.12 0.30 2.08 2.44 1.79 5.94

2.05 9.87 2.23 -4.90 1.78 9.26 1.65 9.42 1.96 -8.52 1.73 7.77 2.03 -8.82 1.59 2.49 1.28 -7.54 1.62 -2.40 1.70 -6.68

1.75 -9.92 1.69 -7.80

1.69 -18.61

.001 n.s. n.s. n.s. .05 .001

.001

.001

.001

.001

.001

.001

.001

.05

.001

.05

.001

.001

.001

.001

Amongst interviewer ratings of candidate NVB, ratings of greater eye-contact (t = 4.29, p < .001) and more positive facial expressions (t = 5.94, p < . O O l ) differ most significantly between the accepted and rejected groups, whilst ratings of more frequent postural changes differentiate less significantly (t = 2.44, p < .05). This table also displays highly significant differences between mean ratings of accepted and rejected candidate personalities, whereby successful candidates are rated as more interesting, relaxed, strong, successful, active, mature, enthusiastic, pleasant and dominant (p < .OOl), and more sensitive and honest (p < .OS). Candidates accepted for second interview are liked significantly more than those rejected (t = -9.92, p < .OOl ) , and also perceived as more similar to the interviewer (t = -7.80, p < .OOl ) .

Decision making in interviews 69

The latter results suggest that interviewers were prone to both personal liking bias and the similar-to-me effect. To check this point, Pearson’s product moment correlation coefficients were calculated between ratings of occupational group suitability and those of personal liking and perceived similarity. Overall evaluations correlated r = .64 with personal liking and r = .50 with perceived similarity, confirming the susceptibility of interviewer assessments to these two dysfunctions.

Accepted candidate personality profiles

Although acceptance ratios varied considerably across different occupational groups (40 per cent of police officer candidates were accepted whereas 95 per cent of army officer applicants were accepted for second interview), z tests revealed only three significant differences in interviewer ratings of accepted candidate personalities. Accepted police officer trainees were rated more active than industrial chemists (z = 2.14, p < .05), and also more honest than marketing trainees (F = 2.7 1 , p < .05). Similarly, army officer trainees were rated higher on honesty than marketing trainees ( z = 2.56, p < .05). All other z tests, however, revealed non-significant differences, strongly suggesting that interviewers were recruiting for a prototypical graduate personality rather than for occupation-specific personality stereotypes. The ideal graduate for all occupational groups was perceived as interesting, relaxed, strong, successful in life, active, mature, enthusias- tic, sensitive, pleasant, honest and dominant.

Impression-fomtion model: Stage I Returning to the hypothesized lens model of interviewer impression formation, a series of hierarchical multiple regression analyses with stepwise introduction of variables was computed in order to elucidate perceptual links between candidate behaviour and interviewer impression formation. In the first multiple regression analysis, interviewer personality ratings and moderator variable ratings were used as dependent variables and ratings of candidate non-verbal behaviour as independent variables. The results are set out in Table 3.

On all dependent variables, candidate facial area NVBs of eye-contact and positive facial expressions account for significant proportions of total variance. It is therefore apparent that interviewer impressions of candidate personality were substantially linearly dependent upon candidate facial NVB. These findings support those of the earlier laboratory experiment and generate several important implications which are considered below.

Imprasion-formation modeil: Stage 2

In order to delineate interviewer perceptual links between impressions of personality and perceptions of occupational group suitability, a further set of multiple regression analyses was computed using ratings on the 11 personality scales as the independent variables and suitability ratings as the sole dependent variable. The results are shown in Table 4.

Ratings of personality account for almost 60 per cent of the total variance in occupational group suitability ratings (total R-squared = .56), with interviewer percep-

70 Neil Anderson and Viv Shackleton Table 3 . Stepwise multiple regression analyses with personality ratings as dependent variables and non-verbal behaviour ratings as independent variables

Independent Dependent variable variables Multiple R R2 F' Beta

Boring

Relaxed

Weak

Unsuccessful

Active

Immature

Enthusiastic

Insensitive

Pleasant

Honest

Dominant

Liked

Similar-to-me

FE EC H G FE EC HM PC FE EC FE EC HM FE EC EC FE

HM P FE EC P PC FE P

HM EC H G FE EC HM FE

H G EC FE EC FE EC P

H G FE EC PC

.60

.65

.66

.41

.47

.49

.50

.42

.49

.39

.45

.46

.45

. 5 1

.34

.39

.41

.43

.47

.53

.54

.55

.28

.30

.3 1

.31

.32

.52

.56

.57

.30

.35

.38

.36

.39

.52

.57

.58

.58

.43

.46

.47

.36

.42

.43

.17

.22

.24

.25

.18

.24

.15

.20

.21

.20

.26

.12

.15

.17

. I8

.22

.28

.29

.30

.08

.09

.I0

.I0

. l l

.27

.31

.32

.09

.13

.14

.13

.16

.27

.32

.33

.34

.I8

.21

.22

185.04 .60 35.94 .27

3.09 .08 65.54 -.41 24.29 -.26

8.72 .15 2.47 .09

71.85 .42 24.74 .26 58.14 .39 21.36 .24

4.09 - . I 1 83.00 - .45 27.27 -.26 42.83 .34 15.37 .21 6.36 -. 14 4.45 - .11

93.22 -.47 28.54 -.26

3.56 .09 3.23 -.09

25.20 .27 3.28 -. 10 3.30 .I0 3.13 .10 2.69 -.11

118.44 -.52 22.79 -.23

3.72 .10 32.77 -.30 13.08 .19 6.05 -. 13

49.03 -.36 10.05 -.17

120.94 -.52 26.42 -.25

3.41 .08 2.26 .07

73.69 - .43 9.83 - . I6 5.14 - .11

F ratios significant at p < .05. FE = facial expressions; EC = eye-contact; PC = postural changes; P = posture; HM = head movements; HG = hand gestures.

Decision making in interviews 71

Table 4. Stepwise multiple regression analysis with occupational group suitability ratings as the dependent variable and personality factor ratings as independent variables

Independent variables Multiple R R2 F* Beta

Unsuccessful .64 .4 1 224.52 -.64 Enthusiastic .71 .51 168.04 .37 Immature .73 .54 125.69 -.22 Pleasant .74 .55 99.99 .15 Weak .75 .56 83.86 -. 18

* F ratios significant at p C .05.

tions of general success or failure of the candidate in life emerging as the most influential factor accounting for over 40 per cent of the variance (R-squared = .41, F = 224.52, p < .OO 1). Perceived enthusiasm accounts for a further 10 per cent of total variance when entered into the multiple regression analysis. Immature, pleasant and weak add a further 5 per cent to total variance, but beyond these variables, all other factors do not add significantly to total variance. These results therefore highlight the extent of the linear dependence of interviewer perceptions of occupational group suitability upon personality factors.

In order to establish the influence of the two moderator variables, liked and similar-to- me, upon these 11 personality scales, a final set of multiple regression analyses was computed. In this case the personality factors constituted the dependent variables, whilst the moderator variables were entered into the stepwise equations as independent variables.

As Table 5 shows, the moderator variable ‘liked’ accounts for significant proportions of the variance in all personality factors, ranging from R-squared = .16 on the dependent variable of insensitive up to R-squared = .61 on pleasant. Indeed, only on the former does the moderator of ‘similar-to-me’ account for a greater proportion of variance than liked. These multiple regression analyses illustrate the pervasive bias of personal liking upon interviewer personality assessment and suggest that, on most personality traits, personal liking bias is more influential than the similar-to-me effect.

Outcome decision-making perceptual links

A single Pearson correlation coefficient was calculated between ratings of occupational group suiability and interview outcome decisions in order to examine the final part of the lens model. These two variables correlate r = .73 ( p < .Ol>, indicating a strong association between interviewer impressions of occupational group suitability and the accept-reject decision. This correlation coefficient was squared, giving an R-squared value of .53. Thus, it may be stated that occupational group suitability predicts the outcome decision .53 and, as such, reflects the weight of personality considerations in graduate interview decision making.

Summarizing the results of these multiple regression analyses, the findings are displayed in relation to the lens model in Fig. 2. The linear dependence of the four sets of variables is quantified by the inclusion of the relevant R-squared values. Only the most

72 Neil Anderson and Viv Shackleton

Table 5 . Stepwise multiple regression analyses with personality factors as the dependent variables and moderator variables as independent variables

Dependent variable

Independent variables Multiple R R2 Beta

Boring

Relaxed

Weak

Unsuccessful

Active

Immature

Enthusiastic

Insensitive Pleasant Honest Dominant

Liked Similar-to-me Liked Similar-to-me Liked Similar-to-me Liked Similar-to-me Liked Similar-to-me Liked Similar-to-me Liked Similar-to-me Liked Liked Liked Similar-to-me Liked

.65

.69

.48

.51

.55

.60

.59

.61

.60

.64

.59

.61

.64

.66

.40

.78

.42

.44

.48

.42

.48

.23

.26

.30

.36

.35

.37

.36

.41

.35

.38

.40

.43

.16

.61

.18

.19

.24

237.59 150.17 97.02 56.62

143.64 90.86

174.13 96.05

180.56 111.34 176.78 99.09

223.14 125.12 63.81

5 17.76 69.86 76.56 50.25

-.65 -.29

.48

.20 - .55 -.27 -.59 -. 18

.60

.27 -.59 -.20

.64

.20 -.45

.78

.42

.44

.26

* F ratios significant at p < .05.

influential NVBs of candidate facial expressions and eye-contact are included to simplify the specification model.

The model clearly illustrates that outcome decisions are highly dependent upon perceptions of occupational group suitability (R-squared = .53, p < . O l ) , which are significantly dependent upon impressions of personality (total R-squared = . 5 6 , p < .O 1) which, in turn, are partially dependent upon candidate facial NVBs (total R-squared ranges from .08 to .42). This figure hence represents a specification model of interviewer impression formation and outcome decision making resulting from this study.

Discussion

Four principal findings emerged from this study of graduate interviewer decision making:

(1) Interviewer impressions of candidate personality were linearly dependent to a substantial extent upon candidate facial non-verbal behaviours.

( 2 ) The person-perception biases of similar-to-me effect and, more influentially, personal liking for the candidate affected interviewer impression formation.

Candidate Interviewer impressions Suitability non-verbal behaviour --) of candidate personality -) decision making -) Outcome decision

Stage 2: Occupational group suitability perceptual links f f Stage 1:

Impression formation perceptual links

Rz f f

4 UJ

Figure 2. The linear dependence of interviewer outcome decisions and suitability ratings upon impressions of candidate personality and non-verbal behaviour. Linear dependence between the moderator variables, similar-to-me and liked, and the 11 personality factors is detailed in Table 5 .

74 Neil Andenon and Viv Shackleton (3) Interviewers screened against the ubiquitous personality criterion of a suitable

graduate prototype for diverse occupational groups. The suitable graduate was perceived as interesting, relaxed, strong, successful, active, mature, enthusiastic, sensitive, plea- sant, honest and dominant.

(4) Interviewer impressions of candidate personality were highly predictive of occupa- tional group suitability ratings and outcome decisions, indicating the importance of personality considerations in graduate interviewer decision making.

Several implications arise from these findings which can be considered in relation to the two-stage lens model of impression formation developed for the study.

Stage I Impression-fomtion perceptual links

In common with the earlier laboratory experiment (Anderson, 1988 b), the findings of this study illustrate the linear dependence of interviewer impression formation upon microsco- pic facial area of candidate NVB. It seems that interviewees would be well advised to maintain high levels of eye-contact with the interviewer and to display frequent positive facial expressions so as to maximize their chances of success. Moreover, these findings reiterate the potential for self-presentation training for graduate interviewees which concentrates upon facial NVBs.

Stage 2 Occupational group suitability perceptual links

It is significant that the participating interviewers applied the criterion of a universally appropriate personality prototype when determining whether to accept or reject candi- dates. Returning to the occupation-specific stereotype vs. the ubiquitous prototype debate, the results of this study show that interviewers failed to differentiate between appropriate personalities for widely divergent occupational groups.

These findings directly contradict the argument advanced by Jackson and his co- workers that interviewers perceive distinct stereotypes of suitable personality for different occupations (Jackson etal., 1982; Rothstein &Jackson, 1980). Even though interviewers did not distinguish between suitable personalities for different occupational groups, it is unreasonable to justify this finding by arguing that there may be a single most appropriate personality type for such diverse occupations. In fact, it seems highly unlikely that the personality criteria for an army officer should be synonymous with those for an optician or an accountant. Yet, these results suggest that interviewers do look for similar personality characteristics in all graduate applicants.

Other important implications arise in relation to the study findings on the similar-to- me effect and personal liking bias. Since interviewer ratings of perceived personality similarity correlated highly with overall evaluations, it is apparent that interviewers were, at least to some extent, recruiting in their own image. This ‘clone syndrome’ almost certainly acts as an additional source of bias in interviewer decision making as the criterion is unspecified and unvalidated, and, in the longer term, may be self-perpetuating.

Personal liking for the candidate was found to be a pervasive bias affecting the processes of impression formation and outcome decision making in this study. Liking correlated significantly with all other ratings of personality, suggesting considerable halo effects in

Decision making in interviews 75

assessments. More critically, liking correlated highly with ratings of occupational group suitability ( Y = .64). This finding shows a notably stronger association between liking and overall assessment than that reported by Keenan (1977), where these variables correlated r = .5 1. Despite Keenan’s plea that interviewers should guard against this bias affecting their overall evaluations (Keenan, 1978), it appears that this remains a potent source of error in interviewer decision making.

Conclusion

In conclusion, the image of graduate interviewer decision making which emerges from this study is not an encouraging one. Interviewers were prone to several errors in information processing, the combined effect of which is almost certainly to reduce the predictive validity of outcome assessments. The purpose to which the interview is devoted in graduate recruitment therefore requires careful re-appraisal. As Herriot ( 1985) rightly advocates:

What needs to be avoided is the use of the interview as a quasi-personality test, with interview behaviour being used as evidence of personal characteristics (p. 35).

Ironically, these findings show that in fact the graduate interview is used predominantly as a personality assessment device and that perceptual links between candidate behaviour, especially facial NVBs, and interviewer impressions of personality constitute the chain of reasoning underlying interviewer impression formation.

Herriot’s ideal stands in stark contrast to the actual purpose and functioning of the graduate selection interview. On the basis of these study findings, however, it can be argued that interviewer information processing appears so flawed as to corrupt the candidate assessment purpose of the interview. The interview, it may be concluded, is not suited to candidate personality assessment objectives and should thus be reduced in status to a role more befitting the predictive efficacy of interviewer decision making.

References

Anderson, N. R. (1988 a). Interviewer impression formation and decision making in the graduate selection interview: A theoretical and empirical analysis. Unpublished PhD thesis, University of Aston, Birming- ham, UK.

Anderson, N. R. (1988 b). Interviewer impression formation: An experimental investigation. Internal memorandum no. 1006, Social and Applied Psychology Unit, University of Sheffield, UK.

Arvey, R. D. & Campion, J. E. (1982). Theemployment interview: Asummary and review ofrecent research. Personnel Psychology, 3 5 , 28 1-322.

Baskett, C. D. (1973). Interview decisions as determined by competency and attitude similarity.Joumal of Applied Psychology, 57, 343-345.

Brennan, J. L. & McGeevor, P. A. (1985). C N A A Graduates: Their Experiences and Views Twelve Months after Graduation, HELM Working Paper No. 4. London: Council for National Academic Awards.

Brennan, J. L. & McGeevor, P. A. (1987). C N A A Graduates: Their Employment andtheir ExperienceafterLeuuing College: A Summary Repwt of a Survey of C N A A Graduates. London: Council for National Academic Awards.

Brunswik, E. (1956). Perception and the Representative Design of Psychological Experiments. Berkeley; CA: University of California Press.

Cantor, N. & Mischel, W. (1977). Traits as prototypes: Effects on recognition memory. JournalofPersonality and Social Psychology, 35 , 38-48.

Connor, H. & Prior-Wandesforde, G. (1986). The graduate milkround: Its changing role and pattern of use. IMS report no. 124, Institute of Manpower Studies, University of Sussex, Brighton.

76 Neil Anderson and Viu Shackleton Dalessio, A. & Imada, A. S. (1984). Relationships between interview selection decisions and perceptions of

applicant similarity to an ideal employee and self: A field study. Human Relations, 37, 67-80. Gifford, R., Fan, N. G. & Wilkinson, M. (1985). Nonverbal cues in the employment interview: Links

between applicant qualities and interviewer judgments. Journal of Applied Psychology, 70, 729-736. Graves, L. M. & Powell, G. N . (1988). An investigation of sex discrimination in recruiters’ evaluations of

actual applicants. Journal of Applied Psychology, 73, 20-29. Herriot, P. (1985). Give and take in graduate selection. Personnel Management, May, 33-35. Hutt, R. &Parsons, D. (198 1). Getting the drift ofgraduate mobility. PersonnelManagement, March, 29-3 1. Jackson, D. N. , Peacock, A. C. & Holden, R. R. (1982). Professional interviewers’ trait inferential structures

Keenan, A. (1977). Some relationships between interviewers’ personal feelings about candidates and their

Keenan, A. (1978). Interviewing for graduate recruitment. Personnel Management, February, 3 1-35. Mabey, C . (1986). Graduates into Industty. Aldershot, Hants: Gower. Parsons, D. (1985). Graduate recruitment and retention, 1980-1984. IMS report no. 107, Institute of

Manpower Studies, University of Sussex, Brighton. Peters, L. H. & Terborg, J. R. (1975). The effects of temporal placement of unfavorable information and of

attitude similarity on personnel selection decisions. Organizational Behavior and Human Performance, 13,

Rand, T. M. & Wexley, K. N. (1975). Demonstration of the effect, ‘similar-to-me’, in simulated employment interviews. Psychological Reports, 36, 535-544.

Rothstein, M. & Jackson, D. N. (1980). Decision-making in the employment interview: An empirical approach. Journal of Applied Psychology, 65, 27 1-283.

Wexley, K. N . & Nemeroff, W. F. (1974). Effects of racial prejudice, race of applicant, and biographical similarity on interviewer evaluations of job applicants. Journal ofsocial and Behavioral Sciences, 20, 66-78.

Williamson, P. (198 1). Early Careers of 1970 Graduates. Unit for Manpower Studies Research, paper no. 26, Department of Employment, London.

for diverse occupational groups. Organizational Behavior and Human Performance, 29, 1-20.

general evaluation of them. Journal of Occupational Psychology, 50, 275-283.

279-293.

Received 29 September 1988; revised version received I 7 March I989