saog.
New
for
ve An
‘in,
rin,
000
Yo
deree
satis
na of
dt
A META-ANALYSIS OF SELF—SUPERVISOR,
SELF-PEER, AND PEER-SUPERVISOR RATINGS
MICHAEL M. HARRIS, JOHN SCHAUBROECK
‘Kranet Graduate School of Management,
Prdue University
Reviews of self-supervisor, sef-peer, and peer-supervisor ratings have
ssenerally concluded that there is at best # modest correlation between
sifferent rating sources. Nevertheles, there has been much inconsistency
across studies. Accordingly, a meta-analysis was conducted. The resulie
indicated a relatively high correlation between peer and supervisor atings
(= .62) but only a moderate correlation between self-supervisor (p
35) and self-peer ratings (p= .36). While rating format (dimensional
versus global) and rating scale (ait versus behavioral) had litle impact as
"moderators, job type (manageril/ professional versus blue-
HARRIS AND SCHAUBROECK 49
Journal of Vocational Behavior (1970-1986), Organizational Behavior and
Human Decision Processes (Formerly Organizational Behavior and Human
Performance) (1966-1986), and Personnel Psychology (1956-1986), Con-
ference papers from the Midwest and National proceedings of the Academy
of Management were also examined. Computer-based searches were con-
ducted on Psyc-Info (Psychological Abstracts) for the period since 1967
and the ABI Inform data base for the period since 1975, References for
‘other studies were obtained from these sources to expand the research base.
‘Three judgments were made conceming the collection and analysis of
data. First, only those reliability estimates based on accepted formulas
‘were selected 10 determine the mean reliabilities actoss studies. ‘This ex-
cluded, among other estimates, unadjusted average inter-item correlation
coefficients. Second, laboratory studies were deliberately excluded from
the analysis. This was done because the conditions under which laboratory
ratings occur grossly exaggerate the observation time factor. More impor-
tantly, assigned roles cannot be taken to reflect true perspectives of super-
visors and peers (Ilgen & Favero, 1985). Finally, some studies contained
‘multiple measures of performance. Since these measures were typically
not independent, the effect sizes were averaged and the mean effect size
was used in the meta-analysis.
Results
‘The literature search produced a total of 36 independent self—supervisor
correlations, 23 independent peer-supervisor correlations, and 11 indepen-
dent self-peer correlations (approximately 5 studies did not provide either
‘means or correlations). Table 1 shows the average correlations, reliabil
ties, and range restriction for the three rater combinations. The reliabilities
are based on intemal consistency (alpha) measures. It is noteworthy that
the average reliabilities are substantially higher than the .60 figure used by
‘Schmidt and his colleagues (e.g., see Pearlman, Schmit, & Hunter, 1980).
‘There are three reasons for using the more conservative figures found here.
First, Schmidt and his colleagues based their reliability estimates on the
correlations between ratings from different supervisors at different points
in time. In the present study, only intemal consistency measures. were
found. Second, the .60 figure used by other researchers assumes that one
source of ertor is the use of different raters. In the present case, at least
for self-ratings, the rater did not change. Hence, a reliability of .60 would
bbe an underestimate. Finally, there has been considerable debate over the
accuracy of .60 as an estimate of reliability (Sackett, Tenopyr, Schmitt, &
Kehoe, 1985; Schmitt & Schneider, 1983). Use of the higher, and hence
‘more conservative, figures therefore seemed in order.2 2 suo us usucanssou 30)
8 es 20) 5
Saison 2
Bones an pee
g o we Gro Oe OLE Doe wv APS
& wr eae WN Glew (SL x snostasodnss334
a s ww wise tne (Woe of aostniadns 196
Z punoq —punog si ° 7 ”
z ‘bdd) mop
2
suownja0.) 40f synsox s186}oUD-DI2y
vatavae, i
HARRIS AND SCHAUBROECK st
Because there was only one self-peer study reporting reliabilities, es-
timates of peer-raing reliability were obtained from the peer-supervisor
studies; likewise, estimates of self-rating reliability were taken from self
supervisor studies.
‘As shown in Table 1, all three rater combinations had lower 90% conti
dence intervals that were greater than zero, Hence, the correlation between
ratings from different raters is statistically significant. However, there were
marked differences between rater combinations: The peer supervisor cor.
relation (p = .62) was substantially higher than either the self-supervisor
(o = 38) or the selt-peer correlation (p = .36). Finally, all three rater
combinations contained more than 25% unexplained variance even after
sampling error, measurement error, and range restition were taken into
account; accordingly, moderator variables were examined.
Table 2 summarizes the moderator analysis for self-supervisor corre-
lations. Consistent withthe hypothesis, there was slightly more agreement
for dimensional (p = .36) than for global (p = .29) ratings. However,
there was no decrease in variance across the subgroups. In accord with
the rating-scale hypothesis, there was also a small difference between trait
and behavioral scales, wherein greater agreement occured for the latter
While the ratio of unexplained to total variance was less than 25% (15%)
for behavioral ratings, there was no decrease in sample-size weighted vari-
ance averaged across subgroups compared with the whole set. Thus, rating
sale and rating format were not significant moderators for self-supervisor
correlations. However, job type appeared to be a significant moderator.
‘When sampling error and other artifacts were taken into account, no vari
ance remained for managerial professional employees, Although substan
tial variance remained for blue-collar’ service workers, the average cor-
rected variance across the subgroups was lower than forthe set as @ whole.
Moreover, there was lower agreement for managerial professional employ.
ees (p = 21) than for blue-ollar/ service employees (p = .42). Hence,
jb type did seem to moderate self-supervisor comelations
Table 3 summarizes the results of the moderator analysis for peer
supervisor correlations. Contrary to expectations, slighty higher agreement
as observed for global than for dimensional ratings; nonetheless, there was
no decrease in variance across subsets, More surprising was the finding
that greater agreement was obtained for trait (p = .64) than for behavioral
ratings (p = .53). There was only a slight reduction in the sample-size
weighted average variance across subsets as compared with the variance in
the whole set (9 = 23 across subsets as compared with .24 for the whole
set), Finally, job type did not seem to affect the amount of agreement.
Due to the small number of studies (2) using either behaviorally based
rating seales or dimensional formats, these variables were not investigatedBY
PERSONNEL PSYCHOL
2
popu 2 Bue eo eu eur yo zou us
sere sasiod
WAP ila Seaqiun of ws Square Bus ys)
w 31 wr air =
Fea « (rome Me 0
os o wer os
ws a renee 87
se a sez
e e ast
wmnoq—punog one 7a ToREPOR
fsa Samo i aes
na s2quen, reo
amp) 106,
ouso.) s0spsodns- fas ‘sins>y sIsSjouy AOI,33
HAUBROECK
HARRIS AND
ojsod ye Joy pov roswaue
“Sompous See 5
‘won
co
SORApORY
sequins,
us ony pa
soil tod pau
SUOUI 209 s9>4- HES “sHMED HESPOUY soMDLpoW
patava
13 pu oun oa gus uns my
2 ads pauguo>,
inant fi sgn rs Bad yn Sas eae
du Sagan
7 on a ray st wer Sa roo
of ovr We (er 8 6
ox ow 0 ars s 0s
or we rns a at
os w we (os 90
oF se 6 (onus ree
‘onan a
ahr sve
satiny, aL
swoop Updns-12eq “Spey SIS fOHY s09D23po yy
eaTavLst PERSONNEL PSYCHOLOGY
fas moderators for self-peer correlations. Thus, only job type was consi
ered; the results are shown in Table 4
'As in the self-supervisor category, little or no variance remained for
either subgroup of self-peer correlations when sampling error and other
aifacts were taken into account. Accordingly. there was a reduction in
average variance across subsets as compared with the whole set, white the
average corelation was higher for blue-colla/ service workers (p = .40)
than for managerial/professional employees (p = .31)- Thus, job type did
‘meet the criteria for acceptance as a moderator.
{A total of 18 independent self-supervisor means, 7 independent peer
supervisor means, and 4 independent self-peer means were located. AS
predicted by the egocentrie-bias explanation, there was @ fairly large dif
ference between self- and supervisor ratings (d = 70). However, the
true variance was also large (07 = .25). Thus, the difference in means
between self- and supervisor ratings was not statistically significant, Simi-
larly, while on average self-ratings were somewhat higher than peer ratings
(d= .28), a substantial amount of variance remained even after correcting
for sampling exor (2? = .09). Hence, the difference in mean self=peer
ratings was nonsignificant. Finally, while peers provided somewhat higher
ratings than did supervisors (d = .28), the difference was not significant
given the large variance (0? = .11).*
Discussion
The first question of interest concemed the average correlation and
‘mean difference between rater pairs. Overall, the data suggest that after
correcting for measurement error and range restriction, moderate agree-
ment between self=peer (p = .36) and self-supervisor (p = .35) ratings
‘exists, There was much higher agreement between peers and supervisors
(o = .62), Nonetheless, the lower 90% confidence interval was greater
than zero for all three rater combinations. The results of this study then
are somewhat contrary to conclusions by Landy and Farr (1980). As noted
earlier, they concluded that different raters exhibit low to moderate agree
‘ment. The present results indicate that while this is true for self-ratings
‘compared with ratings by others, peer-supervisor ratings demonstrate a far
higher average correlation. This discrepancy between a meta-analytic-based
literature review and a traditional narrative review is not surprising, given
the differences between the two procedures. The present results indicate
the value of conducting quantitative reviews.
44, moderator analysis subgrouping by job type for self-supervisor means revealed no
sedition n variance,HARRIS AND SCHAUBROECK 3s
In terms of mean differences, on average, self-ratings were over a half
standard deviation higher than supervisor ratings and approximately one-
{quarter of a standard deviation higher than peer ratings. Nevertheless, given
the large variances, the lower-bound confidence intervals included zero in
both cases, However, examination of the effect sizes showed that in only
fone case were the self-ratings lower than supervisor ratings.
‘The second question concemed whether or not differences between
studies within each rater pair could be attributed to statistical artifacts,
‘The results indicated that in all cases substantial amounts of variance still
remained, even when the appropriate corrections were made. Accordingly,
‘moderator analyses were conducted,
Of the three moderators, only job type showed any meaningful ef-
fects. Specifically, self-supervisor and self-peer correlations were lower
for managerial/ professional employees than for blue-collar/ service employ:
ees, and no true variance existed for the former category. However, such
effects were not obtained for peer-supervisor correlations. This suggests
that while incumbents of managerial/professional jobs have very differ:
cent views of their performance than do others, observers exhibit far more
agreement. One possible explanation is that egocentric bias is more likely
to occur in ambiguous contexts (i.e., managerial/ professional jobs) than
in well-defined jobs (.e., blue-collar/ service). Whatever the explanation,
Job type does not appear to affect observer agreement; further research
is needed to test why this is the case. The one instance where rating
scale met the criteria for a moderator should be viewed cautiously for sev-
eral reasons. First, only @ small number of studies (5) using behaviorally
based rating scales were available, and the total sample size was small
(n = 506). Second, the decrease in amount of variance across subsets
was very small, Thus, rating scale was at best a relatively minor moder-
ator of peer-supervisor agreement. In answer to the third question then,
‘moderators did help account for some of the variance,
Recall earlier that three basic explanations were reviewed as to why
disagreement between raters may occur. The present findings are consistent
with two versions of the egocentric-bias theory. Specifically, in accord with
attribution theory (DeVader et al, 1986; Jones & Nisbett, 1972), observers
(c., peer-supervisor combinations) displayed higher agreement than did
“actors” with observers ([e., self supervisor and self—peer combinations).
‘The findings also mesh with the moderated defensiveness explanation: even
after the range restriction corrections, observer-observer ratings (i... Peer
supervisor combinations) demonstrated much higher agreement than self-
observer ratings.
Conversely, there was no support for the main-effect-of-defensiveness
explanation as correcting for range restriction did not reduce the difference
soumae we36 PERSONNEL PSYCHOLOGY
between self-peer/self-supervisor correlations and peer-supervisor corre-
lations, Nor was there evidence that observational differences accounted for
rater disagreement as there was a relatively high correlation between peer
supervisor ratings. Finally, there was no support for the organizational-level
hypotheses: an almost identical correlation was found between ratings by
individuals at the same level (sel-raters and peers) and individuals at dif:
ferent levels (self and supervisors). Moreover, the highest correlation was
found for raters at different levels (peer-supervisor). Nevertheless, further
research is needed to conduct more direct tests forall possible explanations
of rater disagreement
I is also noteworthy that, with one minor exception, neither rating
format nor rating scale were significant moderators. Other meta-analyses
Of performance ratings have come to similar conclusions. For example,
Kraiger and Ford (1985) found that rating scale did not moderate ratee
race effects; Heneman (1986) also reported a lack of support for rating
format as a moderator. Hence, despite what would seem 10 be critical
moderators. research has found rating instrumentation to be less useful
in identifying sources of variation than was previously thought to be the
‘case. Conversely, the findings regarding job type suggest that this may be
‘a moderator of importance that other researchers in performance appra
should examine. :
Although there are a number of potential reasons why a substantial
amount of variance between studies remained in most instances, two seem
‘most plausible. First, it seems likely that there are differences between
studies in regard to criterion contamination or rater independence. Specit
cally, studies may have differed as to how much raters knew about others"
ratings. In some studies, subtle forms of criterion contamination may have
existed, For example, itis possible that a few studies involving self-ratings
were conducted shortly after performance appraisal reviews, wherein ratees
had substantial information regarding supervisor ratings. In other instances,
cemployees may have received informal feedback, while in yet other studies,
cemployees may have had litle or no feedback of any sort. Unfortunately,
very few studies report enough detail about the amount or type of feedback
to lest whether this variable moderates rater agreement.
‘A second explanation for variance between studies may be differences
in raters’ opportunity to observe job performance. For example, workers
‘may have more interaction with peers in some cases than in others, thereby
leading to differential agreement with supervisors. Jobs with much task
interdependence may lead to greater agreement between self- and peer
ratings than jobs where workers perform independently.
‘Other potential moderators include such things as purpose of rating
(exg., Zedeck & Cascio, 1982), amount of rater training (¢.g., Smith, 1986),
and rater motivation (Bernardin & Beatty, 1984). However, a review of
alHARRIS AND SCHAUBROECK 37
ur studies showed that these variables were also rarely, if ever, reported.
Hence, no moderator analyses could be performed. Moreover, Melntyre,
Smith, and Hassett (1984) found that purpose of rating had little or no effect
‘on ratings, and many studies have failed to show that training increases the
accuracy of performance ratings, particularly when only a lecture approach
is used (Smith, 1986). Further theory development and empirical research
will be necessary to achieve greater knowledge about which conditions
affect rater agreement
Although one can always argue that a “file-drawer" problem may exist
such than inclusion of missing studies would change the results, this is
unlikely here. First, an extensive literature review was conducted. Second,
Rosenthal’s (1979) index suggests that it would take many studies to dra:
‘matically alter the results, given the size of the correlations and the total
sample size found here
Future research needs to address 1wo issues, First, a more direct test
of these competing hypotheses is needed. That is, a study should be done
wherein different raters (e.g., peers, supervisors, and self-raters) will assess
performance and provide information regarding moderators. For example,
attributions regarding cause of performance, self-esteem, amount of obser.
vational opportunity, as well as other possible factors would be measured.
Second, depending on the outcome of this research, further work on the
‘ways to ameliorate rater bias would be useful. For example, if differential
attributions are a source of conflict, the Performance Distribution Assess-
‘ment (PDA), described by Bernardin and Beatty (1984), might provide a ve.
hicle for increasing rater agreement. Specifically, the PDA attempts to take
into account situational constraints. Research should also focus on identify
ing possible moderators of self-other agreement, For instance, impression
management may be a variable that can account for self-supervisor and
self-peer disagreement (Zerbe & Paulhus, 1987), Clearly, more research
examining a variety of individual differences is in order.
In terms of practical implications, the results suggest that self-ratings
will generally show only moderate correlations with ratings by others
This is particularly true for managerial professional employees. Practi-
tioners considering the use of self-ratings should be aware that there is
liable to be much disagreement. Conversely, peer-supervisor ratings often
(but not always) reflect relatively high rater agreement, even for manage-
rial/ professional jobs. For legal purposes, then, peer supervisor ratings
may be useful, Using raters from different levels may also help to develop
consensus, eliminate biases, and perhaps in tur lead to greater acceptance
by ratees. Nevertheless, there is a plethora of other problems associated
with using peer ratings (Latham & Wexley, 1982).38 PERSONNEL PSYCHOLOGY
In conclusion, the present results show that peer-supervisor ratings
demonstrate a relatively high corelation, whereas slf-supervisor and self
peer ratings exhibit moderate correlations. Compared with the correla
tion between supervisor ratings and results-oriented outcomes. (produc
tion, sales, etc.) (Heneman, 1986), the average correlation obtained here
between supervisor and peer-ratings is substantially higher. Thus, from
both an absolute and relative point of view, there is high convergence be-
tween observer ratings. The moderator analysis showed that corelations
between self-peer and self-supervisor ratings are particularly low for man-
ageria/ professional workers, but not for peet-supervisor ratings. While
these findings are consistent with an egocentrie-bias explanation for rater
disagreement, future esearch should test the alternative explanations de-
scribed earlier in & more direct fashion than was possible here.
REFERENCES
Baird LS. (1977) Self and supervor ratings of performance: As related wo self-eeem and
satisfaction with supervision. Academy af Management Jounal, 20, 291-30,
Berardn HJ, Beaty RW: (1984). Performance appraisal” Assessing human behavior ot
work. Boston, MA: Keat Publishing Company.
Borman WC. (1974), The rating of individuals in oxgaiztions: An aerate approach,
‘Organisational Behavior a Hunn Performance, 12. 105-128
‘Campbell IP; Dunnette MD, Lawler EE, Weick KE. (1970). Managerial behavior, peor
‘mance, and efecrivenes. New York: McGraw Hil
‘Carol 8, Schneier CE. (1982). Performance appratel and review sates: The identi
tion, measurement, and development of performance n organations. Glenview, IL:
Scot, Fores
Cascio WE. (1987). Applied psychology m personne! management. Englewood Clif, NI
Prentice-Hall
| ‘Carnmings LL, Schwab DP. (1973). Performance n organizations: determinants and ap
| rasa. Glenview, TL: Seo Foresnan.
DeVader CL, Bateson AG, Lord RG, (1986). Atibuton theory: A metw-analysis of at
‘utonal hypotheses In Locke E (Ed), Generalizing from laboratory 1 feld sentings
(pp 63-81). Lexington, MA: Lexingion Books
Ferris GR, Yates VL, Gilmore DC, Rowand KM, (1985). The infuence of subocinate age
‘on performance rang’ and causal aibations. FeRSONNEL PSYCHOLOGY, 3, 545-857
| Henderson RI (1984). Performance apratal, Rest, VA: Reson Publishing Compan.
Heneman HG IL (1974). Comparisons of self and superior ratings of managerial perfor
‘ance. Journal of Applied Poychologs, $9, 638-682,
Heneman RL. (1986). The reationship between supervisory ratings and results-oriented mes
sures of performance: A meta analysis. ERSONNEL PSYCHOLOGY, 39, 811-826
olabach RL. (1978), Rater bis in perfomance ratings: Superior, sf, and pcr rings,
‘Journal of Applied Psychology 8%, 379-588,
Hunier J, Schmit FL, Jackson GB. (1982). Mewe-anasis: Cuming research ndings
across sues. Bevery Hills, CA: Sage Publications.
gen DR, Favero IL. (1988). Limits in generalization from paychoogial research operfor-
mance appraisal processes. Academy of Management Review, 10, 311-32
Soves EE, Nisbet RE. (1972). The ator andthe observer: Divergent percepions ofthe causes
‘of behavior, In ones EE, Kanouse DA. Kelly HH, Nisbett RE. Vans S, Weiner 8HARRIS AND SCHAUBROECK 9
(is), Anibution: Percesing the causes of behavior (pp.39-98). Moristowa, Ni
(General Leaming Press
‘Kay E, Meyer HH, French IRP. (1968). Ect of thet i a performance appraisal interview:
Lournalof Applied Prychlogy. 49, 311-317
Klimoski RJ, London M. (1974). Role of the rater in performance appraisal Journal of
‘Applied Psychology, 50, 445-451
Kraiger K. Ford JK. (1985), A met-analyis of exes race effects in perfomance ratings
oural of Applied Psychology, 70. 56-68,
Landy 1, Far JL, (1980). Performance rating. Payhologcal Bulletin, 87, 72-107
Landy 4, Far J, Stl PE, Freytag WR. (1976). Behavioral anchored scales fr rating the
Performance of police officers. Journal of Applied Psychology. 6, 458-857
{Latham GP, Wexley KN. (1982). Increasing productivity though performance appraisal
Reading. MA: Addison: Wesky,
{Lawler EE, (1967), The multitait-multater approach to measuring managerial job perfor.
mance. Journal of Applied Psychology, 51, 368-381
Mabe PA, West SG. (1982). Vly of el eafuation of ability: A review and meta-analysis.
ournal of Applied Psychology, 67, 280-296,
Melnyre RM, Smith DE, Hassett CE (1984), Accursey of performance ratings a alfected
‘by tater training and perecived purpose of rating. Jounal of Applied Psychology, 6
147-156,
Pearlman K, Schmidt FL, Hunter JE. (1980), Validity generalization results fr tess wsed 10
feet ob proficiency and ining success i lena acupatons. Journal af Applied
Psychology 8, 373-406,
Pym DLA, Auld HD. (1965). The self-rating as a measure of employee sstisfatorinss,
Journal of Occupational Paycotogy, 39, 03-113
Rosenthal 8 (1979). The “le dawer problem” and tolerance for nl esuks. Psychological
Burin, 86, 638-641
Suchet PR. Tenopyr ML. Schmit N, Kehoe J. (1985). Commentary on forty questions about
‘ality genetalzationa and meta-analysis, reasonnisrsvcHoLicn 3, 697.798,
Schmit. Schneider B. (1983), Curent ses in penonnel selection, Ta Rowland KW,
Farris GD (Eds), Research in personnel and human resourcer, Vol! (pp, 85-123)
Greenwich, CT JAI Press,
Smith DE. (1986). Training programs for performance appraisal: A review, Academy of
Managemen Review. 11, 22-40
Springer D. (1953). Ratings of candidates for promotion by co-workers an supervisors
Journal of Applied Psschlogy. 37. 47-381
Stele RP, Oalle NK. (1985). Self-appaial based! upon supervisory feedback. eKGONNEL
FSYCNOLOGY, 47, 657-68,
Thornton GC, (1980), Psychometric properties of selE appraisals of job performance. rex
SONNE. SYCHOLOGY, 3. 263-271
Tucker MF, Cline VB, Schmit JR. (1967). Prediction of creativity and other performance
‘measures from biographical information among pharmaceutical scents. Jounal of
Applied Psychology $1, 131-138
Wenley KN. Klimoski R. (1984), Perfomance appraisal: An updste. fn Rowland KM,
ems GD (ls). Resarch i personne a luman resoures, Vol. 2 tgp. 35-79),
Greenvich, CT: JAI Press,
Willams WE, Seiler DA. (1973. Relationship between measures of effort and perfomance,
Journal of Applied Paschlogy, 57.49.54
‘ammo RF. London M, Rowland KM. (1982), Organicaton end tate differences in per
formance appraisals. veRsonves rsvenotocy, 4, 643-688PERSONNEL PSYCHOLOGY
edeck 8, Baker HT. (1972), Nursing performance measured by behaviors expectations
scales: A multiaitmulirater analysis. Orgunisnional Behavior and Human Perfor
mance, 7, 457-46.
edeck S, Casco W. (1982), Peformance decision a a function of purpose of rating and
teining. Journal of Applied Paychlogy, 67, 752-158.
Zerbe Wi, Paulus DL (1987) Socllydesiable sponding in organizational behave: A
reconcepion. Academy of Management Review, 12, 250-264,
STUDIES INCLUDED IN THE META-ANALYSIS.
Baied LS. (1977). Self and superior ratings of performance: AS relae'd to seesteem and
satisfaction with supervision. Academy of Management Journal, 20, 281-300.
Blackburn RT, Clark MJ, (1975). An asessmen of faculty performance: Some comelates
between administrator, colleague. student, and sel rings, Sociology of Education,
48, 202-256,
Booker GS, Miller RW. (1966. A closer look at peer rating. Personnel 43, 42-47.
Borman WC. (1978), The rating of individuals im organizations: An akematve approach,
‘Organisational Behavior and Human Performance, 12, 105-12,
Boruch RF, Larking JD, Wolins L, MacKinney AC. (1970) Allemative methods of analysis:
Mule mukimetbod dat Educational and Psychological Measurement, 30, 833
Brit AP, Aldag RJ, Van Sel M. (1977), Moderators of the relationships between self and
superior evaluations of jab performance. Journal of Occupational Psychology, 50,
129-134
Burke RJ, Weivel W, Desaca G. (1982). Supervisors’ and subordinates’ communication
experiences in managing jo perfomance: Two blind men desrbing an elephant?
Paychologiea! Reports, 50, 527-834
Campbell DJ. (1985), SelFappraals from two perspectives: Estem versus consistency
influences. 28h Annuc! Proceedings ofthe Midwest Academy of Management
Cena JA. (19). Colleagues 35 raters of classroom instruction. Jounal of Higher Educ
tho, 46, 37-337.
Doyle KD, Crichton CL (1978). Student, peer and self evaluations of college instructors.
Journal of Eductional Psychology, 70, 815-825,
kpo-Ufot A. (1979). Selfperceivedtask-elevant abilities, rated job performance, and com
Plaining behavior of junior employees in a government mnitey. Journal of Applied
Psychology, 64, 429-434
Fes GR, Yates VL, Gilmore DC, Rowland KM, (1985). The influence of subordinate age
‘om performance raings and causal tributions. FERSONNEL PSVEHOLOOY, #4, S4S-SS7
Fuqua DR, Johnson AW, Newnan JL, Anderson MW, Gide EM. (1988). Varabliy eros
sources of performance ratings. Journal of Counseling Psychology, 31, 249-282
GGormly 1. (1983). Comespondence between personality trait rings and behavior evens
“ournal of Personality. 52, 220-282
CGrifihs RD. (1975). The accuracy and comelates of psychiatric patents self assessment of
theie work behavior. British Jownal of Socal and Clinica! Poychoogy. 14, 181-189
‘Gunderson EK, Nelson PD. (1966). Citron measures for exvemely isolated groups. Pak
SONNE PS¥EHOLOEY, 19, 67-80.
Heneman HG Il. (1974). Comparisons of seif- and superior ratings of managerat perfor.
mance. Journal of Applied Psychology, 59, 638-642,
Hortman EL, Rover JH. (984). An objecuve peer evaluation scale: Construction and
“ality, Educational and Pychological Measurement, 4, 382-34HARRIS AND SCHAUBROECK 6
Holzbach RL. (1978). Rater bias in peefomance ratings: Superior, self, and per ratings.
Journal of Applied Psychology. 63. $9588
Iigen DR, Peterson RB, Martin BA, Bocschen DA. (1981), Supervisor and subordinate
reactions t0 performance appraisal sessions, Oreaniational Behavior and Human
Performance, 28, 311-330
Kaufman GG, Johason JC. (1984), Sealing peer ratings: An examination ofthe diferent
valdites of postive and negative nominations. Journal of Applied Prchology, 39,
302-306,
Kay E, Meyer HH, Pench JRP. (196). Effects of teat ina perfomance appraisal interview.
Journal of Applied Psychology, $0, 311-317
‘King LM, Hur JE, Schmidt FL (1980). Halo in muli-dimensions forced-choice perfor.
mance evaluation scale. Journal of Applied Pechology, 68. 07-516
Kirchner WK. (1968) Relationships Berseen supervisory ad subordinate ratings fo technical
personnel, Journal of Industrial Peyehologs, 3, 37-60
Klimoski RJ, London M. (1974). Roe of the rater in performance appraisal. Journal of
‘Applied Pychology, 89, 485-451
Kraut Al. (1975), Prediction of managerial succes by peer and taining staff ratings. Journal
of Applied Psychology. 60, 389-394
Kulbany AJ. (1957), Use of sociomewc peer nominstions in medical education research,
Journal of Applied Psychology 41, 389-394,
{Lawler EE, (1967), The mutiait-multiater approach to measuring manager job perfor
‘mance. Jwnal of Applied Psychology, 1, 368-38,
Levine EL, Flory A. Ash RA, (1977). Seléasersment in personne selection. Journal of
Applied Peychology, 62, 428-435,
Love KG. (1981). Comparison of peer astesiment methods: Relabiliy, valid, friendship
bias, ad user reaction. Journal of Applied Psychology, 66, 451-5),
Maslow AH, Zimmerman W. (1956). College teaching ably, scholaly setvty and person-
ality. Journal of Educational Psychology, 47, 185-189,
Mial WL, Graumenz IL (198). An assessment ofthe accuracy of self asessment for career
decision making. Journal of Vocational Behavion. 25. 248-253
Mitchell TR, Albright DW. (1972). Expectancy theory predictions of the satisfaction, effort.
‘performance, and reiention of naval aviation officers. Organizational Behavior and
Human Performance, 8, 2.
Motowildo $i. (1982), Relationship tetween self-rated perfomance and pay saisaction
among sles representatives. Journal of Applied Paychology, 67, 208-213,
‘Mount MK. (1984, Supervisor, sl, and subordinate ratings of performance and satisfaction
with supervision. Journal of Management, 10, 308-320
Nealy SM, Owen TW. (1970). multiaitmulimethod analysis of predictors and criteria of
faursing performance. Organizational Behavior and Haman Performance, 3, 348.365
(Orpen: (1983. Etfects of race of tater and see on pecr-ratings of managerial potat)
Psychological Reports, 2, 07-510
Parker W, Tylor EK, Bamet RS, MarensL. (1950) Rating scale content I Relationship
between supervisor and seatings,PeRsowNe PS¥CHOURGY, 2, 89-63,
Porter LW, Lawler EE. (1968). Managerial anitudes and performance. Hetewood, tL
Irwin
Prien EP, Liske RE. (1962). Assessmens of higher-level personel: Il Rating criteria: A
‘comparative analysis of supervisor ratings and incumbent ratings of jab performance
PERSONNEL Psveun.00y, 15, 187-194
Pym DL, Auld HD. (1965). The selfsating asa meas of employe saisfetoriness. Journ
of Occupational Pychology, 39, 103-113.
Schmit N, Saar BB. (1978). Behsvor, situation, and rater variance in descriptions of leader
behaviors, Mulibariote Behavioral Research, 13, 483-498a PERSONNEL PSYCHOLOGY
Schneier CE, Beatty RW, (1978), The influence of role prescriptions om the performance
‘appraisal process. Academy of Management Journal, 21, 128-135,
Shapiro GL, Desser G. (1988). Are self-aprasals more realise among professionals oF
nonprofessionals in health care? Public Personnel Management 1, 28-290
Shore LF Thorton GC Ill (1986), Elects of gender on self and supervisory ratings.
Academy of Managemen Jour! 29, 15-12,
Shrauger JS, Lund AK. (1973). Self-evaluation and reactions to evaluations fom eter.
ournalof Personality, 43, 94-108.
Siegel L_ (1982). Pied comparison evaluations of manager effectiveness by peers and
supervisors. Feasonn rsvenotocy, #5, 43-852
Smircich L Caeser RU (1981), Supervisor” and subordinates’ perceptions of performance:
eyoredisareeman, Academy of Management Journal 24, 198-208,
Springer D. (1953). Ratings of candidates for promotion by co-workers and supervisors
Journal of Appled Psychology, 47. 47-351
Stel RP, Onalle NK. (984). SelFappraval based upon supervisory Feedback. exRsONNE,
ravenna, 47, 667-085,
“Thorton GC. 1968), The slationship between supervisory and scitappaisal of exceutve
Performance, PRSONNEL YSYCHOLOGY, 27, 81-188.
‘Tucker MF. Cline VB, Schmit JR, (1963), Prediction of creativity and ater perfomance
measures trom biographical information among pharmaceutical scents. Journal of
Applied Psychology. 51. 131-138
Webs WE, Nolan CY. 1955), Student, supervisor nd selfatings of intctional proficiency.
“oural of Educational Peycholoes, 48, 42-46
Wiliams WE, Seiler DA. (1973). Relationship beween measures of effon an job perfor:
mance. Journal of Applied Paychology, 57, 49-88