Professional Documents
Culture Documents
Performance Appraisal
Performance Appraisal
Biases
PRs suffer from many biases, meaning factors
that “should not” influence raters but do, or
The author factors that influence ratings in ways about
Mark Cook is a Chartered Occupational Psychologist with which raters are unaware.
the Personnel Selection Research Group at University
College, Swansea, UK. Age, ethnicity, and gender
The pooled results of 40 separate research-
Abstract es[1] find a low negative correlation between
Argues that the conventional validation paradigm, which age and PRs, whereas age and objective
uses subjective performance or appraisal ratings as indices of performance correlate positively.
criteria, may be of doubtful validity. Discusses research This implies that PRs show substantial bias
into performance appraisal which documents four sets of against older persons.
problems which may reduce the usefulness of performance Similar large-scale analysis[2] finds a small
ratings as criteria. These problems include biases, politick- but consistent “own race” bias in PRs; whites
ing, impression management and undeserved reputation. favour whites, blacks favour blacks, etc.
Describes the inaccuracies to which these problems give Gender also distorts PRs. Sometimes PRs
rise and concludes that instead of selecting the right show simple gender bias[3], so that women
people for management, selection methods validated with masculine characteristics are rated more
against appraisal will simply perpetuate an unsatisfactory promotable. Sometimes the bias is subtler,
status quo. involving gender stereotype of the occupation,
so that raters who do not see management as a
traditionally female occupation tend to give
women managers lower PRs[4].
Physical appearance
Other biases in PRs are subtler. “Common
sense” argues that physical attractiveness is
not likely to influence PRs, because “beauty is
in the eye of the beholder” and anyway should
not influence behaviour in the workplace. In
Author’s address fact there is a very reliable broad consensus
Author’s name here about who is and who is not good-looking,
Address 1 and strong evidence that appearance does
Address 2 affect PRs. Two separate researches[5,6] find
Address 3 a strong link between attractiveness – rated on
Address 4 a five-point scale – and later salary level. Both
Address 5 find that each extra scale point of rated attrac-
e mail: insert here tiveness is “worth” $2,000-2,600 more in
salary. This implies that the salary difference
Journal of Managerial Psychology
Volume 10 · Number 7 · 1995 · pp. 3–7 between the most and least attractive could be
© MCB University Press · ISSN 0268-3946 as great as $13,000. The effect is far stronger
3
Performance appraisal and true performance Journal of Managerial Psychology
Mark Cook Volume 10 · Number 7 · 1995 · 3–7
for males, in “male” jobs, and in older age like/dislike, similarity of outlook, or
groups. PRs are strongly biased by appear- in-group/out-group effects, because this
ance, and the bias is not confined, as “com- information is not routinely recorded, and
mon sense” might argue, to young female often it is not readily accessible.
ratees. In fact it is probably strongest in tradi- This means that many sources of bias
tional management circles. render performance appraisal ratings less
accurate, but in ways which neither employer
Attitudes and values nor employee can easily identify. In technical
Nursing managers’ PRs are shaped more by terms, the performance ratings are contami-
the manager’s belief system than by the subor- nated, and the contamination is invisible.
dinate’s actual work performance[7]. The This in turn has further implications for
research uses a “laddering” technique to research on personnel selection. Between
explore nursing managers’ concepts of effec- 60 per cent and 80 per cent of researches on
tiveness. This means continuing to ask man- selection use supervisor rating as their criteri-
agers the question “why is that important” on of “true performance”. Bias in ratings,
until one eventually elicits answers like “self- from the sources described above, means that
fulfilment” or “pride in being a good manag- rating may often be a fairly poor criterion of
er”, which the researchers argue represent the true performance, which in turn implies that
managers’ fundamental values. The managers much research may actually underestimate
said that staff work behaviour shaped their the true validity of selection methods.
PRs, but “policy capturing” analysis showed
that the managers’ own values were the best
predictor of their PRs. Subordinates who Politicking
helped managers to feel proud of being a good
Organizations usually assume that managers
manager, or helped to make them feel ful-
want to make accurate PRs, and are prevented
filled, received better PRs.
only by unconscious biases which can be
cured by sufficient training or the right rating
In-group/out-group
format. The in-depth interview research flatly
One school of thought argues that most work
contradicts this and concludes that “political”
groups can be divided into an in-group,
considerations, or “private agendas”, nearly
whose members enjoy the supervisor’s confi-
always influence PRs. The managers list many
dence and concern and get assigned the more
reasons for giving inflated PRs: to maximize
challenging tasks, and an out-group, whose
merit increases, to protect persons whose
members are treated like “hired hands” and
performance is suffering because of personal
are assigned the mundane tasks. Research on
problems, to avoid “washing dirty laundry in
bank staff[8] shows that in-group members
get better PRs but do not perform any better public”, to avoid creating a badly written
on objective performance indices. This record, to avoid confrontation, and to rid
implies that the in-group members achieve themselves of people by promoting “up and
their position not by better work, but by some out”.
other path. The interviews also uncover managers’
reasons for making deliberately low PRs: to
Personal like/dislike shock someone back onto a higher “perfor-
“In-depth” interviews with managers who mance track”, to teach a rebellious subordi-
regularly make PRs reveal that three-quarters nate who is in charge, to send someone a
freely admit they allow liking to inflate PRs message that they should consider leaving the
and dislike to lower PRs[9]. The same analy- organization, and to build a well-documented
sis also reports that 83 per cent of managers record of poor performance to speed up
say being in a good or bad mood shifts the termination.
PRs they make – probably downwards in most Gross leniency in appraisal rating is very
cases, given that 73 per cent say they hate apparent, but subtler “politicking” may not
having to make PRs. be. Like biases based on appearance or atti-
PRs can be analysed to detect age, race or tude, subtler politicking contaminates the
gender bias (and usually are, to avoid equal appraisal rating, and makes it less accurate,
opportunities problems). PR data cannot be both as a measure of performance, and as a
analysed to detect bias created by appearance, criterion in research of selection methods.
4
Performance appraisal and true performance Journal of Managerial Psychology
Mark Cook Volume 10 · Number 7 · 1995 · 3–7
This has a further implication. Most selec- 7 Jolly, J.P., Reynolds, T.J. and Slocum, J.W., “Application
tion research uses performance rating as its of the means-end theoretic for understanding the
criterion of “true performance”. If perfor- cognitive bases of performance appraisal”, Organiza-
mance appraisal is not a good measure of true tional Behavior and Human Decision Processes,
Vol. 41, 1988, pp. 153-79.
performance, then selection methods validat-
ed against appraisal will not be selecting the 8 Vecchio, R.P. and Gobdel, B.C., “The vertical dyadic
linkage mode of leadership: problems and prospects”,
“right” people for management, but will
Organizational Behavior and Human Performance,
simply be perpetuating an unsatisfactory Vol. 34, 1984, pp. 5-20.
status quo.
9 Longenecker, C.O., Sims, H.P. and Gioia, D.A., “Behind
the mask: the politics of employee appraisal”, Acade-
References my of Management Executive, Vol. 1, 1987, pp. 183-
93.
1 Waldman, D.A. and Avolio, B.J., “A meta-analysis of 10 Padfield, P., The Battleship Era, Hart-Davis, London,
age differences in job performance”, Journal of 1972.
Applied Psychology, Vol. 71, 1986, pp. 333-8.
11 Wayne, S.J. and Ferris, G.R., “Influence tactic affect,
2 Kraiger, K. and Ford, J.K., “A meta-analysis of ratee and exchange quality in supervisor-subordinate
race effects in performance ratings“, Journal of interactions: a laboratory and field study”, Journal of
Applied Psychology, Vol. 70, 1985, pp. 56-65. Applied Psychology, Vol. 75, 1990, pp. 487-99.
3 Hartman, S.J., Griffieth, R.W., Crino, M.D. and
12 Bialek, H., Zapf, D. and McGuire, W., Personnel
Harris, O.J., “Gender-based influences: the promotion
Turbulence and Time Utililization in an Infantry
recommendation”, Sex Roles, Vol. 25, 1991,
Division, HumRRO FR-WD-CA.77-I l, Human Relations
pp. 285-300.
Research Organization, Alexandria, VA, 1977.
4 Landy, F.J. and Farr, J.L., “Performance rating”,Psycho-
logical Bulletin, Vol. 87, 1980, pp. 72-107. 13 Campbell, J.P., Dunnette, M.D., Lawler E.E. and Weick
K.E., Managerial Behavior, Performance, and Effec-
5 Frieze, I.H., Olson, J.E. and Russell, J., “Attractiveness
tiveness, McGraw-Hill, New York, NY, 1970.
and income for men and women in management“,
Journal of Applied SociaI Psychology, Vol. 21, 1991, 14 Parkinson, C.N., Parkinson’s Law, John Murray,
pp. 1039-57. London, 1958.
6 Roszell, P., Kennedy, D. and Grabb, E., “Physical 15 Heneman, R.L., “The relationship between supervisory
attractiveness and income attainment among Canadi- ratings and results-oriented measures of performance:
ans”, Journal of Psychology, Vol. 123, 1989, a meta-analysis”, Personnel Psychology, Vol. 39,
pp. 547-59. 1986, pp. 811-26.