ML20238A440

Intervenor Exhibit I-SC-53,consisting of 741209 Article Entitled, Training Managers to Minimize Rating Errors in Observation of Behavior, from Journal of Applied Psychology,Vol 60,Number 5

ML20238A440
Person / Time
Site:	Shoreham File:Long Island Lighting Company icon.png
Issue date:	04/30/1987
From:	Latham G, Pursell E, Wexley K AKRON, UNIV. OF, AKRON, OH, WEYERHAEUSER CO.
To:
References
OL-5-I-SC-053, OL-5-I-SC-53, NUDOCS 8708310079
Download: ML20238A440 (7)

Text

3ff

- - ~ -n

~n

._,.,,-.___m--

r= m - w 5 m = =_. r c _- - -~ ~ --

-m m-u

..a p

gro -3 2 2 -065 g o/ty g{

• %{> $ 3 a

l;ii"v'i'd.Wsif5n!' 1~ Sd - C

'87 AUG 24 A9 :42 Training Managers to Minimize Rating Errors in the 4

Observation of Behavior f,

o Gary P. Latham Kenneth N. Wexley E

f Weyerhaeuser Company University of Akron

,f r

Tacoma, Washington I

Elliott D. Pursell 4

T Weycrhaeuser Company 3

y Plymouth, North Carolina A

1 3

b Sixty managers in a large corpoddon-were-randomly assigned to a workshope

~

a group discussion, or a control group. The workshop and group discussion p'

involved training directed toward the elimination of rating errors that occur in performance appraisal and selection interviews, namely, contrast effects, s

hi halo effect, similarity, and first impressions. Sin months after the trainlog, the d

F managers rated hypothetical candidates who were observed on videotape. The I 7t results showed that (a) trainees in the control group committed similarity, 4

h contrast, and halo errors; (b) trainees in the group discussion committed im-g J[

press!an errors; and (c) trainees in'the workshop committed none of the j

errcrs. The importance of observer training for minimizing the " criterion problem" in industrial psychology is discussed.

l $

r A major problem in the selection and/or observing and evaluating present or potential S h

development of criteria as evaluative indices It would seem logical that to solve th{e employees.

ti of human performance is the lack of reliabil-S ity in the observation of behavior (Ronan & probletn of rating errors, the observer must?

i Prien,1966,1971). This unreliability can be be trained. For example, Levine and Butleri

-d largely attributed to well-known rating errors (1952) worked with supervisors who were'"

?

such as first impressions and the halo effect. overrating employees in the higher job grades' 3

Rating errors are errors in judgment that can and underrating those in the lower grades.%

i occur when one individual observes another. The supervisors were randomly assigned to a4 0

Research designed to overcome this problem control, a lecture, or a discussion group.~

has been restricted primarily to manipulating Supervisors in the lecture group were given a

' {d' the format of the ratings (Dellshire & High-detailed lecture on the technique and theory land,1953; Blanz & Ghiselli,1972; Smith & of performance rating. They were also given Kendall,1963); examining the ratings given background information on wcge administra-by observers in different organizational roles tion and job evaluation. The lecturer ex-(e.g., Kavanagh, hiscKinney, & Wolins, plained to the supervisors the problem caused 1971; Klimoski & London,1974; Lawler, by their previous ratings and what each i

1967; Ronan & Latham, 1974); and/or supervisor needeel to do to correct the prois-5 studying the kind of content or stimuli that lem. After the lectare, questions were encour-are rated, namely, personality versus per-aged and answered by the lecturer. In the formance variables (Kavanagh,1971). In discussion group, the supervisors met together, spite of this work, managers and supervisors to discuss the nature of the problem and i

have continued to make rating errors when how it could be solvei.. The discussion leader ii The authors are grateful to Terence R. Mitchell merely acted as a tnoderator and avoided n:

and Gary A. Yukt for their helpful comments in injecting his own opinions. After expressing d,

writing this article. A shorter version of this paper a number of ideas, the group then arrived at was presented at the ennual meeting of the Ca.

One decision' acceptable to all members. The

'[

nar!Jan Psychological Association, Quebec City,1975.

4 Requests for reprints should be sent to Gary P.

results showed that no observable behavior L I Latham, Human Resources Research, Weyerhaeuser change occurred on the part of the super-

((

Company, Tacoma, Washington 98402.

visors in the control group. Similarly, the i

550 hu 8708310079 070430 PDR ADOCK 05000322 s

G PDR g

mc tt.f '

~

J-natlAAlt Mamalget casussugu

- sedet n SD-372 06-Of 'naica hh. an.-/. 53 si m..enn ee b t W % d o c h er 8) A h. s -

answe' /

J su ao m neuno

/

notcne intmena 1

- ca.r, w, Q 87 i

c..

Other gita,es Reporter TO IMON$

T,-

)

i 3

i 1

1 i

l 1

.i l

l n

a-l;

l..

1,

___._J

_m F"*

4..E._E5'd_.r.Es.M_6L_ 1.. JE!~E43524&.7% _RCCT"".1. rj-.M. 4.C~~~Wi. ME.

~". n 3,

.e -_

_~

gem __.

g.,.. s.. 3 ~,;

,- ;~ c. v w

r-x.y w.

TaArnmo MANAGERS To MINntIZE RATINo Ennons 551

~

lecture had no influence on changing the Procedure ng Errors m. the supervisors' method of rating. This finding knowled e of these errors

" "..' " " ' " ' " " " *.*P""*"P''

indicated that 8

Or alone will not lead raters to take effective who were to act as trainers were given 3 days of training. Three were randomly assigned to receive t

steps to counteract them. Only the group training in conducting the workshop and the re.

'enneth N. Wexley h

discussion method was effective in modifying maining three were given training in conducting a raioersif!/ of Akron M

the supervisors' rating behavior.

group discussion. Examples of each rating err.>r, s

Wexley, Sanders, and Yuki (1973) showed made in a performance appraisal, a selection inter.

view, and in an off-the-job setting (e.g., making y

that despite attempts to reduce contrast ef-Judgments about neighbors), were demonstrated 1

l fects by means of a warning (lecture) and/or and/or discussed with each group of trainers.

an anchoring procedure (comparing appli.

The trainees were brought to one of three loca-signed to a workshop cants to a preset standard or anchor), sub-tions i r training. At each location, one trainer and group discussion M

jects contmued to make this error. Only an conducted the workshop while another conducted the group discussion.

ting errors that occur

{

intensive training workshop led to a sigm,n-Workshop. The workshop -included videotapes of mely, contrast effects, cant behavior change. Specifically, the work. hypothet!:al job candidates being appraised by a after the training, the i

ved on videotape. The shop gave subjects a chance to practice ob. manager. The development of the videotapes was cotamitted sLnilarity, A

serving and rating actual videotaped cand1-Identical to the procedure discussed by Wexley, eussion committed im.

dates. In addition, the subjects were given Yuki, Kovacs, and Sanders (1972). The trainees mmitted none of the

n immediate feedback regarding the accuracy of gave a rating on a,uoint. scale as to how they thought the manager in the videotape evaluated the imizing the " criterion "j

their ratings.

candidate, and how they themselves would rate the I1, A possible limitation of this study may candidate. Gr up discussi n f 11 wed as to the I evaluating present or potential have been the fact that it was conducted in reasons for each trainee's rating of both the mana.

a laboratory setting with undergraduate psy-ger's evaluatlon and the candidate. In this way, the trainees had an opportunity to observe videotaped cem logical that to solve the q chology students. Thus, the generalizability managers making observational errors, to actively ating errors, the observer must ^i of the results to experienced individuals in participate in discovering the degree to which they indust was questionable. Moreover* this themselves were prone to making such errors, to or example, Levine and Butler y study, as well as Levine and Butler's (1952), foral observations, and to practice job-related ta 4

receive knowledge of results as to their own behav-ed with supervisors who were f, only eliminated one rating error, and the ployees in the higher job grades,a ing those in the lower grades. [

eHects of the training were not assessed over to reduce any errors they were committing. f The first exercise focused on the similarity, or f

time

1. i"'3"'-'8'"' effect. This error is a tendeucy on the 3rs were randomly assigned to a $

The Purpose o.! the present research was to part of a rater to judge more favorably those he 1;

cture, or a gscussion group. E-evaluate the workshop approach (Wexley et perceives as similar to himself (Rand & Wexley*

g i the lecture group were given a. :

al.,1973) and the group-discussion method closely an assessee resembles the rater in attitudes i

1975; Wexley & Nemeroff,1974). That is, the more L

re on the technique and theory (Levine & Butler,1952) in training employ. or background, t.he stronger the tendency of the e rating. They were also given '

C iformation on wage administra* C ces in a large corporation to minimize the rater to judge that individual higher.

followin8 our ratin errors: contrast effects, list of the job regm.e given a job description and a f

evaluats'on, The lecturer ex.

supervisors the problem caused halo effect, similarity, and fust impressions.

rements for a loan officers posi.

Data were collected 6 months af ter the train-tion. They then observed a videotape in which there vious ratings and what each was a strong attitudinal and biographical similarity ded to do to correct the prob.

ing was completed to assess the training between the mariager and a below average candi-effects over time.

date. Relatively httk job-related information was

. lecture, questions were encour. g clicited by the manager from this candidate. At e

wered by the lecturer. In the f up, the supervisors met together,'

MErnoD the end of the videotape, a colleague of the manager

)I e nature of the problem and '

Subjects requested an evaluation of the candidate. At this I

I i

point the videotape was stopped. Each trainee was i

asked to give two ratings: (a) How would you Je solved. The discussion leader Sixty managers in a large corporation were ran.

rate the candidate? and (b) How do you think as a inoderator and avoided >g d mly assigned to one of three conditions: a work.

the manager rated the randidate? They then dis-own opinions. After expressing shop, a group discussion, or a control group that cussed their ratings among themselves m relation to jj received no training. Each group contained 20 indi.

the similarity error. Finally, the trainees were shown deas, the group then arrived at )

viduals. An undenying assumption of this study the remainder of the videotape. The tape showed tcceptr ble to all members. The f was that all managers need this type of training to the manager extolling the virtues of the candidate to I that no observable behavior }

some extent. The only criterica for participation ed on the part of the super. +

was that the trainee's duties require frequent per.

his colleague and thus committing the similar-to-rne

{

f formance appraisals of subordinates and/or the

error, j

control group. Similarly, the x selection of potential employees.

The second exercise used the same job descrip-tion and requirements but focused on the Aalo

,?

r

^

4;. + ' # 4*,

n

.--..g,-. Z

MQ" ~ " Q;u -,M_;hina;;;m.

--9'--

.$ha y C3hg-ggg gg d {si g5T Q*y{' Q gg s: 4,n

~:

h

.;f-

?

552 G. P. LATHAM, K. N. Wrxin, so E. D. PunsELL f

h j.

effect. This error refers te the rater's exaggeration in the workshop group. Thus, the primary dif.

of the homogeneity of an individual's character-ference between the two training programs was the

' I' istics or traits. The trainees watched a videotape in method used to eliminate the rating errors.

which a manager was so impressed by a candidate's Festint The individuals in the workshop, group.,

S nonjob-related background that he was also im-discussion, and control groups were tested togethe -

i E pressed with the candidate's job-related qualiLca-6 months after the training program. Two fo R.

tions even though the latter were highly dubious. A and B, were used for testing each rating errorJ f

The trainees were asked to rate the candidate as The control, group-discussion, and workshop train.

Sie fL they thought the manager would rate him and to ces were each randomly divided so that half of

• ?

indicate how accurate they thought the manager each grcup (i.e.,10 individuals) received Form A ",

Fir:

was in his rating by supplying their own rating of and half received Form B.

the applicant. The trainees then discussed their Before seeing the Erst videotape, all subjects werg Cor ratings as well as solutions for redudng this error. given the following instructions:

f The third exercise dealt with controsi effects 7

(Wexley et al.,1972). The trainees were given a You are going to rate some Individuals. Please job description and a list of the job requirements rate each person in terms of the degree to which for an accountant's job. They were then given the g g g, g ;,

gg resum6 of a highly qualified apph,eant and were job. There are no tricks. For example, if you see"

.Na$

r asked to rate him. The procedure was then repeated someone who obviously does not fulSil the re ~

• 8 "*

{-

with a second highly qualined applicant and then quirements of the job, don't say to yourself that g

9

,,g 3

with an average applicant. The fact that evalua-tions of job applicants can be affected by the suit-acceptable. If he is unacceptable, rate him as un ~

e ot' that subordinates are often evaluated in comparison

, acce table. Similarly, if you see someone who is -

ability of immediately preceding apphcants and g,

ee h

h w% um mWn % $

g y

to other subordinates rather than against an estab-lished standard was discussed. The necessity of

g g a

basing ratings a;;ainst a predetermined anchor or During testing, the subjects were given detailed jot;,t ben c,i

/..

standard was emphasized. The trainees were then descriptions and lists of the requirements for e.ch j

g trained in accordance with the principles and pro-job in question. None of the videotapes had been.

ndeota S

J' cedures described in Wexley et al. (1973).

i seen previously by any of the subjects during train ~

The final exercise concerned first impressions ing. In testing for similarity error, the subjects who "

'd d: -

error. This error is committed when an observer received Form A saw a candidate who fulfilled the' evaluates som' cone on the basis of judgments made requirements of the job, but whose biographical N'*" '

1 primarily after an initial meeting. The trainees background and attitudes were those of a culturally r

were given a job description and a listing of tie disadvantaged individual; those subjects receiving' h

specihc job requirements for an insurance rater.

Form B saw the same individual, but in this in-E y( -

They were then shown a videotape of an interview stance his background and attitudes were those of

.in which the applicant presented a poor impression a typical middle-class, job candidate. If a given

["k by her answers, actions, and appearance. The re-training program were effective, the trainees t'.hould mainder of the videotape showed that the applicant have given this individual approximately the aame i

was acceptable for the job; however, the manager rating on the 9-point hiring-recc.nmendation scale I

continued to act on the basis of his initial impres-regardless of the tape they saw.

i sion. Again, the trainees were requested to give two First impressions were tested by changing the^

f types of ratings; one based on their own opinion of order of favorable anii unfavorable responses given =

the candidate and the other based on their opinion by an individual. Form A showed a woman who I'

of what the manager thought of the applicant. The presented an unfavorable Erst impresrion but later j

trainees discussed their individual ratings as wsll as in the interview presented a favorable impression.

1. • T y,

ways to reduce this observation error.

Form B presented the same individual with a

,Q Gromp discunion. The format for the group dis-favorable-unfavorable sequence of responses. The jcy cussion required the trainer to first define the four actual responses given by the woman were identical.

i rating errors. An example of each error was giwn in both forras, but the order of the questions and benefici

h for three situations: the performance appraisal, the answers in the interview were reversed. Ratings "

selection interview, and an off-the. job situation. The were agai. ma5e on a 9-point biting-recommenda.

i y#

trainees then divided into subgroups of three or four tion scale.

people and generated personal examples that they llalo was established by presenting trainees with The could remember experiencing in each of the three an individual's folder containing an application -

awe e l['

situations. These examples were subsequently shared blank, refetences, interview information, and work

{

between subgroups luring a general group discus-experience. Form A contained extremely unfavor, reactio sion. The trainees again returned to their respective able information and Form B contained highly '

mitted. 4

.s;g subgroups and generated solutions to each of the favorable information. Ilowever, no information measur!

,' :M rating problems. The solutions were also shared was given on the individual's educational back- :

g3p _ !

{

between subgroups during an overall group meeting, ground. The trainees were subsequently abown a l

i, b[

These solutions were identical to those decided upon videotape of this individual being questioned about don tr;

?

tt

- -. : z T. ;..,. Y

%K Y 5 f5.s.; d 5 5ur^%

= m~ Q ((#f l =f =~=k & {A$ $ 5 j.- hh h ".?*=~.[ Q hh&;i&

g~.e w.

q-g.g;

.a 4.a

?

D. PrJRSELL TRAINING hlANAGERs To hi!NIht!ZE RATINo ERReas 553

[

9 Sroup, Thus, the primary di' TN1 i

the two training progranis was the MEAN RATINGS Mb STANDARD DEVIATIONS FOa Tux TRWUNG MD CortraOI, Gaoers eliminate the rating errors.

p Individuals in the workshop, group _7 control group:, were tested together f Wnrkshop Group discussion Control the traming program. Two fortz.

Response error Form A

SD X

SD 8

SD used for testing each rating error.

a T

sup-discussion, and workshop train. '

Simikr-to-me A

3.7 2.21 4.8 2.78 3.4 2.46 th hall B

4.7 2.41 4.4 1.90 5.9 1.37 1

0 divid 1s) re elved Form 9

First impressions A

7.5

.85 7.0 1.89 5.8 2.04 B

7.0 1.15 4.7 2.45 5.6 2.59 h

t videotape, all subjects were i Contract effects A

5.8 2.20 3.9 2.33 3.0 2.45 3

B 5.9 1.73 4.6 2.63 6.5 1.18 S.

M E 3D5P"Ctl0"8 Ifalo A

8.1

.99 8.2 1.23 6.8 1.48 f

g to rate some individuals. Please~

B 8.5

.71 8.4

.70 8.4

.97 en in terms of the degree to which 1 he or she is acceptable for the ggg',Qc" " T{r tion of those invohnns the halo e6ect were made on 9-point scales of the deeinion to u,. AP ratinga. with the enc e no tricks. For example, if you see f i c b'I '8"' '*"" "e snade on a 9 Pmnt scale Di the appbcant's education. with 9 indicaung obviously does not fulfill the re-f l

' the job, don't say to yourself that 2 usly a trick so I will rate him as i her education. Ratings were made on education and of 9 indicated the greatest benefit. The dif-he is unacceptable, rate him as un.

the other job-related factors. If the trainees who imilarly, if you see someone who is)j were exposed to the unfavorable information gave ference in the ratings between the two train-gific, don't fool yourself by saytr.g this individual approximately the same rating on g

.t e something wrong :;omewhere. Rate I education as did the trainees who were exposed to 5.80, p < 001).

al the way you actually see him.

the favorable information, halo error would haie The means and standard deviations of the the subjects were given detailed job ~ been considered climinated.

~

ratings given to the individuals on the video-Testing for contrast effects involved use of the

! lis*s of the requirements for each

. None of the vidcotapes had been videotapes from the Wexley et al. (1973) study.

' apes are presented in Table 1. Comparisons U

~

3 Form A presented a highly qualified applicant sol-among means were made between Forms A by any of the subjects during train-h lowed by an average applicant. Form B presented and B for each of the three groups using

[or similarity error, the subjects who y a low applicant followed by the same average aP-two-tailed i tests. The results indicated that A saw a candidate who fulnlled the

( the job, but whose biographical plicant. Ratings were made on, the 9-point hiring-the control group committed the ic11owing 1

1 attitudes were those of a culturally recommendation scale. The ratmgs of the average individual; those subjects receiving.

applicant were analyzed to determine the extent of ratm.g errors: similar-to-me (t = 2.81, p <

i be same individual, but in this in*'

.05); halo (t = 2.87, p <.05); and contrast contrast effects.

ground and attitudes were those of For all test exercises except the one for halo, the effects (t = 4.07, p <.01). The m' analyses I

t rating scales were benchmarked as follows:

9-would recommend strongly that an offer be made; (Hays,1973) revealed that similarity, halo i

lle-class, job candidate. If a given a were effective, the trainees should M 5-would recommend with reservations that an offer.and contrast errors accounted for 267o,27%,'

Individual approximately the same 2 be made; and 1-would recommend that no offer and 44% of the variance in the ratings, re.

6 9-point hiring-recommendation scale be made. The halo-exercise benchmarks were modi-spectively. The participants in the group

)e tape they saw, h 6ed as follows: 9--applicant has excellent educa-tional quali5 cations for the job; 5-applicant has discussion exhibited only one error. Rather dons were 'ested by changing the 3

eak educational qualifications for the job; and than committing a first-impressions error' ible and unfavorable responses given -

.al. Form A showed a woman who these a. dm.. duals displayed a sizable recency nfavorable first impression but later 1-apphcant has unsatisfactory educational qualiS-n w presented a favorable impression. " cations for the job.

or "last-Impressions" error (t = 2.35, p <

Finally, subjects completed a reaction consisting of a 9-point scab on which they rated

.05, w' = 18%). The m' values for similarity measure inted the same individual with i give.n by the wornan were identical the extent to which they felt the program had been halo, and contrasi errors were found to be corable sequence of response' but the order of the questions and y beneficial to them on their jobs during the past 6 close to zero. The trainees in the workshop 7

months.

e interview were reversed. Ratings :

did not commit any of the rating errors; all de on a 9-point hiring-recommends-t tests were nonsignificant and all w Rrstn.Ts esti-l' 8

rnt.tes were negligible (i.e., 0%-1%).

Lablished by presenting trainees with,

The results of the two training programs i folder containing an application irere evaluated on the basis of the trainee SWssa t$1n d eiy unfavo reaction measures and the rating errors com-The difficulty in developing adequate mea.

tre

$n and Form B contained highly cutted. The mean rating on the reaction sures of an individual's job performance is rmation. However, no informauon the individual's educational back,g measure for the workshop trainets was 8.08 generally referred to as the " criterion prob-g3D =.92), while that for the group-discus-lem" by industrial psychologists. Lifson l

NiiviYual $ing que tion d si n trainees was 6.25 (SD = 1.52); a rating (1953) showed that up to one third of per-out :

.rg

E d $? i;k N N $.5r-d N N I b = h EkbN5bN c_-

_-_ _ o w 554 G. P. LAmAM, K. N. WrxLrv, AND E. D. PURSELI.

i i

formance-measurement variance was due to The finding that the group-discu '

rater differences.* As Ronan and Prien trainees rather than the control group "

abe gr p

(1971) have pointed out, the fact that this mitted the error of first impressions is pug bad m k

f nding was obtained "Io a controlled experi. zling. Anecdotal evidence from the traing '

co"*a..

mental situation makes it conceivable that indicated that the trainees in this group --

indivic Ji even more such unwanted variance occurs in sidered discussion of the error irrelevant to,se F

field studies having less stringent attempts at cause "the error is so obvious that no c5P',

c ;

t.ontrol" (p.151),

rienced manager commits it." Consequen issk 11 The significance of the present study is there was relatively little group discussion cu N. r

,4 that it has shown that this aspect of the cri.

it. Inspection of Table 1 indicates that d.

then i

terion problem can be reduced by training testing, these trainees actually made a andth observers to minimize rating errors. In addi-cency error. It appears that in the process job se(

J:

tion, the study provides external validity proving their point they " leaned over Fina (Campbell & Stanley,1966) for the previous wards."

the sm p[j; research by Levine and Butler (1952) and That the workshop appears to be slightly and t1 Wexley et al. (1973). Specifically, these re.

more effective than the group discussion in have I g

suits show that training programs that carc. eliminating rating errors could be a result

"' ' l fully apply basic principles of learning in one or more combinations of factors. Fi -

secrdi f,

accordance with the task being taught can the trainees in the group discussion were becaus.

L eliminate more than onelating error com. to obtain knowledge of results from that11

?

mitted by subjects from a different popula. trainer and from each other as to their pei what t tion. Levine and Butler (1952) trained first-sonal understanding of the errors and theli dV'".'

,Y line foremen, Wexley et al. (1973) trained solutions, but unlike the trainees in the work Il*S y

college students, and the present authors shop, they did not receive feedback as -

this su y

trained managers in a corporate setting. their own specific behavior in committing prgrat Moreover, the present study showed that the error.

y adratet results of this training are sustained over Second, the trainees in the workshop "

c stly 1

time, ported that the highly structured format

'*P'"S' f

A possible limitation of this study is that their program made them feel that the ting reting 7.

the testing was a simulation rather than an required away from their job for this train" Proacht 6

actual measure of the trainee's on-the-job ing was being used wisely. This was noi the nu worksh behavior. It would have been practically in.

always the case with trainees in the' group f.

feasible to assess the degree to which these discussion. Although this group was given Pared i managers actually make all of these errors in greater freedom to participate in the strue.

ers, be.

pk a real performance appraisal. The present turing of the training content, they wanted Pmgrar

.i -

approach, however, is consistent with Hemp. more feedback from the trainer than frog light o j

hill's claim made in 1964 (cited in Wal. each other. Consequently, many of them,

th". 8" lace,1965) that the only hope for evalu. expressed a lack of interest in the program.,

N5'l2" a:ing managers is to establish simulated cri. This finding is consistent with the trainee erkshi teria in which their performance can be com. reaction data.

ocar to pared to known standards uuder controlled, Third, the workshop training lasted 9 rating.

i standardized conditions. Moreover, research hours, while the group discussion lasted s "Sl ar B

cited by Lifson (1953) has shown that filmed hours. This difference was due to the time trainm{

h performances are rated the same as live per-required in the workshop to set up and USS formances.

the various videotapes. Although this dc difference could have explained the superi-'

g

• Lihon (1953) was concerned with the nature of ority of the workshop trainees if testing had till be y

errors that are involved in time. study pace ratings. Occurred immediately after training, it g orcums It is intercating to note that his raters were similar

.e to those used in our control group, in that both difficult to believe that this difference by it.

} l sets of individuals had considerable experience in self could explain test results obtained 6 krkshin i observing and evaluating workers, but had not months later.

perfon I

ypi received training in ways to reduce rating errors.

Fourth, it is possible that the trainees In; mson h

O

[

u

wc==55NbkN== :Mh$

~5 YEE WN Y-

,,_m A. e n. s =wn:.:

~'

,g

'Y Y"?

y-

~

nw

_.,q -..,.. sw u...

E. D. PURSEI.I' s

\\

ding that the group-discussi TRAinrxc hfANAGERS To hf1 NIM 12E RATIxo ERRoss w

her than the control group corn [..

t e group discussion and the control group Blanz, F.,

& Ghiselli, E. E. The talted standard 555

\\

error of first impressions is puz.

ad more anxiety during testing due to their scale: A new rating system. Personnd Psychotory, dotal evidence from the tramerg :

Comparative lack of experience in observing1972, 25,1ssa sat the trainees m. th;s group c h-individuals on videotapes. This explanation, Campbell, D. T., & Stanley, J. C. Experi ussion of the error irrelevant be.goo, seems doubtful in that the subjects were quasi-experimental designs for research. Chicago:

error is so obvious that no expe.^

experienced manage,rs who were engaging in a Rand McNally,1966.

Hays, W L. S888i3 tics for the social sciences. New

'2ager commits it." Ca2sequentlyg tssk that was similar to their day-to-day ac-York: IIolt, Rinehart & Winston,1973.

"latively little group diicussion ogi Uvities. Moreover, the managers knew that avanas, M. J. m content issue in performance j

3 n of Table 1 indicates that durint' their ratings durm.g testm.g were anonymous appraisal:

se trainees actually made a re. - and thus could not have any impact on their 24, 653-668.A review. Personnel Psychology,1971, It appears that in the process et l0b security.

Kavanagh, M. J., MacKinney, A. C., & Wolins, L.1 Iasues in

!)

ir point they " leaned over back.

Finally, the findings may be an artifact of managerial performance: Aiuttitrait-8"It'8'"> 1971 75> 34-49-muheth d analyses of ra w

g" the sm..larity between the workshop tralm.

u workshop appears to be slightly - and the testing. That is, the subjects may glimoskj, R. J., & London, M. Role of ng ve than the group div;ussion lu have been able to recognize when a given (Adory,1974, 59, 445-451. performance appraisa i

[a rmjs error was being tested and thus responded u

a esu rst>

accordingly. This explanation was ruled out Lawler, E. E. The multitrait-multirater approach to jj77,g7sychol y, l967f52 36N8t.

in the group discussion were able -

ecause postexperimental interviews indicated anage l job p d (nowledge of results from thehthat the trainees had no idea whatsoeverLevine, J., & Butler, J. lecture versus group dis-of from each other as to their per.

what error was actually being tested on a cussion in changing behavior. Journal of Applied Psychdogy, 1952, 36, 29-33.

.tanding of the errors and their -

'"en videotape' Lifs n, K. A. Errors In time-study judgments of

~-t unlike the trainees in the work.

It is difficult to conclude on the basis of Industrial work pace. Psychological JIonographs, did not receive feedback as t this study that only one of the two training s

1953, 67(5, Whole No. 355).

ecific behavior m committing an -

programs should be used. The primary dis-Rand, T, M., & Wexley, K. N. A demonstration of the Byrne similarity hypothesis in simulated em-advantage of the workshop was that it was

.s pl yment interviews. Psychological Reports,1975,

.e trainees in the workshop rp' c s@ and dmoconsuming to develop. The 36,535-544.

the his;hly structured format erPense of preparing the videotapes and the Ronan, W. W., & Latham, c. P. The reliability 2 mad'e them feel that the tu. renting and/or purchasmg of equipment ap.and validity of the critical incident technique:

ne Jt y from their job for this train.

closer look. Studies in Personnel Psychotory,1974, proached $1000. On the other hand, once A

6,53-64.

L 2g used wnsely. Thn,s was not.

the materials were developed, training the Ronan, W. Wo & Prien, E. P. 7'oward a cruerion ase with trainees in the groY'

• E* P W.ers was relatively easy conl-theory: A redew of research and opinwn. creens.

.ithough this group was given pared tn tralmng the group-discussion lead-boro, N.C.:

Richardson Foundation,1966. Creativity Research In

]m to partics,pate m the strue.

ers, because the major part of the training

' tralmng content, they wanted. program was on the use of the videotapes. InRonan, W. W., & Prig E. P. Perspectives on the Appleton century crofts,1971.mumtment of human k from the trainer than frorn'd light of the one rating error committed b t

Consequent;y, many og gem the group-discussion tramees and the moreSmith, P., & Kendati, L. M. Retranslation of expee.

Y

$ck of interest m the program.J ; positive reaction to the workshop by the tations: An approach to the construction of un.

as consistent with the tralDee workshop trainees, the workshop would ap.ArpWd Psychdogy,1963, n,149-155. ambig 3 ating errors. On the other hand, when theDear to be slightly more e aHace, S. R. Criterla for what ? American Psy.

workshop training lasted 9 chologist, 1965, 20, 411-417,

'the gtoup discussion lasted 6w mst and time for developmg a structured Wexley, K. N., & Nemeroff, W. F. Effects of racial

.J ifference was due to the timei l raining program is prohibitive, the group prejudice, race of applicant, and biographical simi.

larity on Interviewer evaluations of job appli.

e workshop to set up and shor uiscussion method appears to be a highly.1974, 20, 66-78. cants. /ound of Social and Behavioral Sci

-ideotapes. Although this thne benefcial alternative. Thus, the chotce be.

Id have explained the super!6 men the two methods, m most instances, Wexley, K. N., Sanders, R.

E., & Yuki, G. A orkshop tram.ees if testing had e till be co:.tm, gent upon a given organization s Training interviewers to etiminate contrast efrects ediately after tralmng, it arcumstances, in employment Interviews. Journd of Applied psychology, jeve that this difference bY 1973, 57, 233-236.

+:

REFERENCES Wexiey, K. N., Yuki, G. A., Kovacs, S.

Z.,

!i plain test results obtained k lerkshire, J. R., & Highland R W Fo Sanders, R. E. Importance of contrast effects in h

rced-choice s possible that the traineesy "performancerse a Psycholoxy.1953, o, 355-378.

"'"# "I O#N# D rating-A methodological study.

8' I

<neceived necemde,,,1974)

?

)

$N

..