Social Movement Rhetoric and the Social Psychology of Collective Action: A Case Study of...
-
Upload
nick-hopkins -
Category
Documents
-
view
214 -
download
1
Transcript of Social Movement Rhetoric and the Social Psychology of Collective Action: A Case Study of...
The Psychological Impact of Assessment in a
Development Center
Jan Francis-Sm yth e1,3 an d Philip M. Sm ith2
This paper extends previous work on the psychological impact on candidates of
se lection and asse ssme nt processes, presenting results from a study of 32
participants in an assessment center (AC) program of a major U.K. financial
services organization. Both quantitative and qualitative methodologies are used
to investigate : (a) the impact of the asse ssme nt outcome on cognitive and
affective reactions to the assessme nt process and candidates’ post-assessment
psychological states, job-related attitudes, and intentions, and (b) the role of
factors such as quality of feedback from the assessor, managerial support, and
care er line position on the impact process. Results show that the assessment
outcome decision does have an effect on post-assessment psychological states
and work-related attitudes but this effect works through a person’s perception
of the impact the assessment process will have on their career. This perception
of impact is well predicted by the candidate ’s perce ived quality of feedback and
current caree r position. Good feedback helps careers more than poor feedback
and the nearer the person’s current role is to the position for which they are
being assessed the more likely they are to benefit from the process.
KEY WORDS: impact validity, procedural justice, selection, assessment.
INTRODUCTION
Although early work in selection and assessment addressed candidate s’perceptions of the validity of the use of such technology, little attention
was give n to the psychological impact of these processes on candidate s until
the early 1980s.
Hum an Relations, Vol. 50, No. 2, 1997
149
0018¯7267/97/0200 ¯0149$12.50/1 Ó 1997 The Tavistock Institute
1Worce ster College of Highe r Education, He nwick Grove, Worceste r WR2 6AJ, UK.2Opus Consulting, Edgecombe Hall, Richmond Hill, Clifton, Bristol, BS8 1AT, UK3Requests for reprints should be sent to Jan Francis-Smythe, Worcester College of Higher
Education, Henwick Grove, Worcester WR2 6AJ, UK.
In 1982, Thornton and Byham sugge sted researchers should assess not
only candidate s’ ove rall perceptions of selection processes but also their
effects on various aspe cts of their psyche such as se lf-esteem, motivation,
and job satisfaction. At a more theoretical leve l, Dreher and Sacke tt (1983)
suggested the selection or assessment decision produce s first a cognitive
response , followed by an affective response and finally a behavioral re-
sponse (Fig. 1). In the case of a selection rejection or a fail at assessment
then the cognitive response might be : a) a belief that rejection is due to
the person not posse ssing the skills or attribute s needed for the job (an
inte rnal attribution) , or b) a belief that rejection is due to a decision error
on the part of the assessors (an external attribution) . Possible affective re-
sponse s are re duce d se lf-e ste e m and job motivatio n, and be haviora l
responses might be seeking additional training or education, seeking other
jobs not requiring the same skills or even a total withdrawal from the work-
force.
Empirical work on the impact of se lection processes on candidate s has
been carried out since the late 1980s. For example , Noe and Ste ffy (1987)
showed that reactions to assessment centers (ACs) were significantly re-
late d to job involve ment and career exploration behavior 1 year after the
AC. Iles and Robertson (1989) showed candidate s expe rienced no signifi-
cant change in organizational commitment and job involve ment, but were
more like ly to be thinking of changing the ir careers and having less-definite
career goals after attending an AC. Bray, Campbe ll, and Grant (1974; cited
in Noe & Ste ffy, 1987) showed that individuals who received less-favorable
evaluations after an AC reported less career salience and less upward mo-
bility motivation as measured by proje ctive tests than people who received
favorable evaluations. More recently, Smithe r, Reilly, Millsap, Pearlman,
and Stoffe y (1993) demonstrated a re lationship between perceptions of fair-
ness of the selection process and the recommendations applicants make to
the ir friends regarding the company. This alternative perspective of selec-
tion (i.e ., the extent to which a measuring instrument, or process such as
selection, has an effect on a subje ct’s psychological characteristics) has been
referred to as ‘‘impact validity’’ (Robertson & Smith, 1989), the social side
of se lection (Herriott, 1989) , and social validity (Schule r, 1993) .
One of the first important studies (Fletcher, 1991) monitored the long-
term e ffe cts of an AC by taking me asure s of nee d for achie ve ment,
Fig. 1. The basic impact model proposed by Dre her and Sackett (1983).
150 Fran cis-Sm ythe an d Smith
organizational commitment, job involve ment, and psychological well-be ing
before , immediate ly after and 6 months after the assessment. The results
showed that the time of measurement of the psychological characteristics
was crucial; in general, self-esteem and job involve ment rose immediate ly
after the center and then fe ll sometime later. There were, however, differ-
ences between successful and unsucce ssful candidate s. Those who were
unsucce ssful exhibite d significant reductions in self-esteem, the work ethic
component of need for achie vement and job involve ment, 6 months after
the center.
While this study had monitored the affective and behavioral response
to the assessment, Robertson, Iles, Gratton, and Sharple y (1991) were the
first to attempt to monitor the cognitive response originally propose d by
Dreher et al. (1983). They provide d evidence to sugge st that beliefs about
the ade quacy of the assessment process and its perceived impact on their
career acted as mediators in the impact process (i.e ., the effects were not
produced directly as a result of the assessment process but they were pro-
duced as a result of two diffe rent cognitive response s to the process: (a)
how ade quate the person be lieved the assessment process to be, and (b)
how much the outcome decision would have an effect on their career). For
example , if a person was rejected for a new position but be lieved the psy-
chometric tests used for the selection were inappropriate , then the impact
effect would be less than if they had be lieved the tests to be highly appro-
priate .
More recent work in organizational justice theory (e.g., Greenbe rg,
1990) propose s the use of the term procedural justice for the applicant ’sperceived fairne ss of the selection process. Gilliland (1993) identified ten
procedural rule s re lated to formal characteristics of the selection process,
information offered during the process, and inte rpersonal treatment, which
combined (in varying weights according to the circumstance s) to form an
ove rall evaluation of procedural justice . The cognitive response ‘‘belief
about the adequacy of the process’’ propose d by Robertson, Iles, Gratton,
and Sharple y (1991) , and used in this study, is a composite measure of
procedural justice .
Robertson et al. (1991) made comparisons between ‘‘pass’’ and ‘‘fail’’groups on psychological health, se lf-esteem, organizational commitment,
and career withdrawal cognitions. In one type of AC studied, there were
significant differences in organizational commitment and desire to leave a
career between ‘‘passes’’ and ‘‘fails.’’ It is, however, impossible to attribute
the cause of these diffe rences to the AC process itse lf without pre-assess-
ment measure s for comparison. Gilliland (1994) also provide s evidence that
selection procedure justice impacts on both self-efficiacy and work pe rform-
ance of hired employees.
Psych ological Im pact of Assessmen t 151
Other researchers have propose d that factors such as social support
and context (Robertson & Smith, 1989) , feedback and supervisory interest
(Ilgen, Fisher, & Taylor, 1979; Robertson et al., 1991) , and career line dis-
tance (Scholl, 1983) might act as moderators in the impact process. For
example , quality of feedback might act as a moderator between the effect
of the outcome decision and a candidate ’s self-esteem. A rejected candidate
with poor feedback might suffer a lowering of se lf-esteem, whereas a re-
jected candidate with good feedback might expe rience no reduction in
self-esteem. To date, as far as the present researchers are aware, none of
these moderators have been supporte d empirically, although timely feed-
back has been shown to enhance belief in the ade quacy of the selection
process (see Bies & Moag, 1986).
This study extends earlie r work on ‘‘impact validity’’ by using both
qualitative and quantitative methods to explore the effects of an assessment
process on middle managers in a large U.K. financial services organization.
The objective s were twofold: first, to establish more firmly the causal role
of the cognitive response variable s (be lie f about adequacy and perceived
career impact) in the impact process through incorporation of ‘‘pre’’ meas-
ures in the research design, and second, to explore the possible role of
moderators (quality of feedback after assessment, extent of managerial sup-
port, and current career position) in the impact process as depicted in Fig.
2. This give s rise to two specific hypothe ses:
Hypothesis 1. Beliefs about the ade quacy of the process and its per-
ce ived care e r impact will me diate the effe ct of outcome decision on
candidate s’ se lf-esteem, organizational commitment, job involve ment, and
career planning.
Hypothesis 2. Quality of feedback after assessment, extent of manage -
rial support, and current career position will moderate the relationship
between outcome decision and self-esteem, organizational commitment, job
involve ment, and career planning.
Fig. 2. Model of psychological impact explored in this study.
152 Fran cis-Sm ythe an d Smith
METHOD
Sam ple
The study was carried out in a large U.K. financial services organiza-
tion. At the time, 167 pe ople had atte nde d the Company ‘‘in-house ’’Assessment Center. Thirty-e ight attende es were selected as participants for
the study having satisfie d the following criteria: (a) attendee still worked
for the organization, (b) attende e was male (in acknowle dgment of possible
gende r effects (Alban-Me tcalfe, 1991) , (c) attende e had a male manager
(in acknowle dgment of possible sex bias effects), (d) attende e’s line man-
age r had between two and four subordinate s who had attended an AC
(management support is like ly to vary with numbe r of subordinate s attend-
ing an AC), and (e) the final sample to include participants drawn from
at least six diffe rent ACs (to eliminate bias from any one AC). Thirty-four
of the participants agreed to take part in the study although non-availablity
of two reduced the sample size to 32.
Procedu re
Pilot measures were trialled on one participant and three people in
middle /senior management positions in other organizations through role-
play scenarios. Each of the study participants were seen at their place of
work approximate ly 6 months after the AC. They first took part in a struc-
tured inte rview with the first author in which the ir perception of perceived
impact on their career, the adequacy of the assessment process, the quality
of the assessor’s feedback, and the degree of managerial support they re-
ceived were assessed. For each of these, candidate s were first asked to an-
swer pre-defined questions on a numerical rating scale and then encourage d
to talk freely about the topic in their own words. The researcher recorded
in writing, at the time of the interview, the responses to all questions. Each
interview laste d approximate ly 45 minute s. Transcripts of the interviews were
analyze d for participant statements which might serve to provide rich illus-
trations, either in support or refute of a particular quantitative finding or
from an alte rnative , previously-unconside red perspective , on the issues under
question. The researcher remaine d in the room while the candidate then
comple ted a bookle t of questionnaire measures (self-esteem, organizational
commitment, job involve ment, career planning, change in organizational
commitment (pre-AC-post-AC), change in job involve ment (pre-AC-post-
AC), and change in career planning (pre-AC-post-AC)).
Psych ological Im pact of Assessmen t 153
Measures
For full description of measure s see Appendix A.
Assessment outcom e (OUTCOME) was assessed on a 4-point scale :
A= Excellent, B= Good (meets standard) , C= Just Misse s Standard, and
D= Substantially Misse s Standard. This represented the final assessment
rating give n after conside ration of a participant ’s performance across six
exercises on six dimensions.
Participant ’s current career position (POSITION) represents the num-
ber of grade points he was be low the position for which he was be ing
assessed at the time of the AC.
Managerial support (SUPPO RT) was assessed as the degree of support
the participant believed he had from his current line manager with respect
to his job-re late d personal deve lopme nt.
Quality of feedback (FEEDBACK) was measure d as quality of feed-
back give n to the participan t by the assessor in terms of its perceived
usefulne ss.
The following previously-de veloped scales were used: (1) the partici-
pant’s beliefs about the adequ acy of the AC process (ADEQUACY) and
perceptions of the impact (IMPACT) the AC had had on his career (Robert-
son e t a l., 1991) ; ( 2) self-esteem (ESTEEM) (Rosenbe rg, 1965) ; (3)
organization al commitment (COMMITMENT) (Mowday, 1979) ; (4) job in-
volvement (INVOLVEMENT) (Lodahl, 1965) ; (5) The extent of future
career plan ning (PLANNING) (Gould, 1979) .
As an alternative to longitudinal studie s with self-report measure s,
Howard, Ralph, Gulanick, Maxwe ll, Nance , and Gerber (1979) propose d
the use of retrospective pretests. Five studies conducte d by Howard et al.
(1979) lend support to the fact that when actual pre¯post self-report mea-
sures are used, the results may well be confounde d by a response shift;
that is, the subje ct’s perception of the response scale would stretch to fit
the expe rience continuum. In the retrospe ctive pre¯post test, subje cts are
asked to rate themselve s after the inte rvention on both a posttest and a
retrospe ctive pretest (how they remember themselve s before the inte rven-
tion) . Postte st and retrospe ctive pretest diffe rence scores can then be
compared. Much research (e.g., Hoogstrate n, 1985; Pohl, 1982; Sprange rs
& Hoogstate n, 1989; Gutek & Winte r, 1992) has found differences of at-
titudes computed in this way to be more valid than those between actual
pre- and post-measures. Terborg and Davis (1982) found the retrospe ctive
technique was robust to proble ms of saliency, priming, and order effects
and the retrospective pretest ratings were not biased by memory distortion
or social desirability. Gutek and Winter (1992) recommend the inclusion
of both retrospe ctive and current evaluations in studies inve stigating atti-
154 Fran cis-Sm ythe an d Smith
tudinal consistency as a valid method of assessing and controlling for re-
sponse shift bias. The retrospective pretest method was utilized in this study
to measure change ( D ) (pre-AC¯post-AC) in organization al commitment, job
involvem ent, and extent of future career plann ing ( D COMMITMENT, D IN-
VOLVEMENT, D PLANNING).
RESULTS
Table I shows the means, standard deviations, and ranges of all the
variable s for the whole sample and separate outcome groups (N.B., no sub-
ject achieved Outcome A) and Table II shows the inte rcorrelations between
the variable s in the study.
Outcome groups differed significantly on IMPACT (F(2,29) = 4.7, p
< 0.05) ; participants with ove rall assessments suggesting ‘‘Good’’ or ‘‘Just
Misses Standard’’ (B or C) felt that the AC had a positive impact on their
career, whereas those in the group ‘‘Substantially Misse s Standard’’ (D)
perceived it to have had a negative impact. These quantitative differences
were supporte d by participants ’ comments at inte rview (eg. ‘‘it helped boost
my career to my current position of . . . much quicker,’’ Group B—Positive
comment; ‘‘they say the y only ke ep it for 18 months but I’m sure if I applie d
for a management job they’d look at it,’’ Group D—Negative comment).
For others, the effect of outcome decision appe ared less strong than the
quantitative results suggest and othe r variable s such as time and pre-AC
perceptions appeared to be important moderators (e.g., ‘‘A few months
Table I. Descriptive Statistics of Variables Involved in Study
Variable
Whole sample
(Bs, Cs, and Ds)
n = 32
Outcome group
B
n = 7
C
n = 18
D
n = 7
Mean SD Range Mean SD Mean SD Mean SD
Adequacy 2.88 0.86 1.3¯5 3.08 1.08 2.83 0.82 2.83 0.85
Impact 3.41 0.98 1.5¯5 3.86 1.18 3.58 0.71 2.53 0.98*
Support 2.82 1.25 1.0¯5 3.95 0.89 2.72 1.11 1.93 1.26*
Feedback 2.28 0.55 1.2¯4 2.50 0.90 2.28 0.42 2.06 0.49
Position 1.70 0.71 1¯4 1.57 0.54 1.80 0.77 1.57 0.79
Estee m 5.06 1.49 1¯6 4.71 1.25 5.39 1.29 5.14 1.57
Planning 2.30 0.72 1¯3.5 2.36 0.71 2.23 0.65 2.27 0.95
D Planning 0.94 1.58 2¯7 1.27 2.21 1.28 1.27 ¯0.01 1.34
Involvement 45.75 6.10 33¯58 43.43 5.44 46.22 6.20 45.14 5.27
D Involvement 0.73 1.48 ¯3¯4 1.00 2.00 0.78 1.21 0.43 1.81
Commitment 69.39 13.16 38¯97 68.00 19.28 69.44 12.94 71.14 8.31
D Commitme nt ¯0.10 1.87 ¯5¯5 0.27 2.98 ¯0.10 1.13 ¯0.43 2.44
*Indicates difference between outcome groups significant at alpha £ 0.05.
Psych ological Im pact of Assessmen t 155
Ta
ble
II.
Co
rre
lati
on
s A
mo
ng
Po
ten
tia
l M
ed
iati
ng
an
d M
od
era
tin
g V
ari
ab
les
an
d O
utc
om
e M
ea
sure
s
Outcome
Adequacy
Impact
Support
Feedback
Position
Esteem
Planning
DPlanning
Involvement
DInvolvement
Commitment
Ou
tco
me
Ad
eq
ua
cy
¯0.1
0
Imp
act
¯0.4
5*
0.4
4*
Su
pp
ort
0.5
3*
0.0
70
.43
*
Fe
ed
ba
ck¯0
.24
0.5
7*
0.5
3*
0.2
4
Po
siti
on
0.0
00
.09
0.3
5*
0.1
90
.17
Est
ee
m0
.11
0.1
5¯0
.04
0.0
90
.16
¯0.2
3
Pla
nn
ing
¯0.0
40
.11
¯0.1
6¯0
.48
*¯0
.19
0.1
7¯0
.31
D Pla
nn
ing
¯0.3
00
.40
0.4
0*
0.3
6*
0.6
3*
0.4
0*
¯0.0
6¯0
.24
Invo
lve
me
nt
0.1
0¯0
.09
¯0.3
0¯0
.43
*¯0
.13
¯0.2
40
.07
0.4
5*
¯0.2
4
D In
vo
lve
me
nt
¯0.1
30
.00
0.4
1*
0.1
80
.33
0.3
6*
0.0
1¯0
.41
*0
.47
*¯0
.36
*
Co
mm
itm
en
t0
.08
0.1
10
.34
*0
.48
*0
.12
0.2
90
.08
¯0.5
9*
0.2
9¯0
.58
*0
.33
D Co
mm
itm
en
t¯0
.13
0.3
9*
0.5
1*
0.3
30
.50
*0
.58
*¯0
.21
¯0.2
30
.49
*¯0
.18
0.4
9*
0.4
5*
*p
£ 0
.05
. ¯
ago, straight after the feedback I would have felt that the AC could have
pre judiced my career—now I’m not so sure ’’ and ‘‘some people see them
as a threat—if I mess it up it could harm me’’).
Outcome groups also diffe red on SUPPO RT (F(2,29) = 6.0, p < 0.05) ;
the better the AC performance, the greater the perceived managerial sup-
port. Again, these quantitative differences were supporte d by participants ’comments at inte rview (e.g., ‘‘I have a feedback session fortnightly with
respect to proje cts and needs and get a lot of instant feedback—I definite ly
feel one needs a manage r working with you 50/50 on your action plan,’’Group B—Positive comment; ‘‘my manage r actually phone d me up in the
middle of the assessment centre!’’ Group D—Negative comment).
Interview transcripts were checked for concordance with quantitative
findings. One major anomaly existed: although self-esteem did not differ
significantly between outcome groups when measure d approximate ly 6
months after the AC, comments made at inte rview suggest that immediate ly
after the AC this may not have been the case (e.g., ‘‘my own expe rience is
that it’s not stressful—post-AC blues is nonsense,’’ Group B—Positive ; ‘‘you
go with the ‘course mentality’ and then come away with a deflated ego—there’s the potential to undermine one’s confide nce’’; ‘‘I felt totally de-
stroyed by the report as did several other people ’’; ‘‘I felt very negative after
reading my report and thought it’s blown all my credibility’’; ‘‘the written
report was most demoralizing,’’ Group D—Negative ). To what extent the
other variable s may also be affected by time is unknown and the results
should therefore be seen in this context—as a ‘‘snapshot’’ in time. Further
research using a longitudinal design is required to explore this issue.
Robertson et al. (1991) suggest that IMPACT acts as a mediator in
the impact process, i.e ., the outcome decision affects the participant ’s per-
ception of the impact the process has had on his career which then in turn
affects his reactions to the process (i.e ., there will be no direct effect of
outcome decision on participants ’ reactions) . The patte rn of correlations
obtaine d in this study supports the idea that reactions to the outcome of
the AC are mediated by beliefs about its career impact. OUTCOME is
significantly related to IMPACT (r = ¯0.45, p < 0.05) and IMPACT is in
turn significantly relate d to D COMMITMENT (r = ¯0.51, p < 0.01), D IN-
VOLVEMENT (r = ¯0.41, p < 0.05) , and D PLANNING (r = ¯0.40, p <
0.05) . In order to examine further the pote ntial mediating role of the cog-
nitive re sponse (IMPACT), multiple re gression analysis was used, as
propose d by James and Brett (1984) . This analysis confirmed the role of
perceived career impact as a mediator between the outcome decision and
subse quent affective responses thus providing partial support for Hypothe -
sis 1. For full details of this analysis, see Appendix B.
Psych ological Im pact of Assessmen t 157
It was propose d by Robertson et al. (1991) that belief about ade quacy
might also act as a mediator in the impact process, i.e ., that an outcome
decision might evoke a cognitive response in terms of a belief about the
adequacy of the process and this would then in turn effect the participants ’affective reactions to the process. However, it was also noted that although,
in the Robertson et al. (1991) study, be liefs about ade quacy diffe red between
pass and fail groups in some of the ACs; in othe rs, which appeared to be
better designed, no differences existed. This suggests then that while ADE-
QUACY may be determined by outcome decision and hence act as a me-
diator in the impact process, it is perhaps more like ly that it is determined
by the procedure s used and hence does not act as a mediator. The results
of this study show that OUTCOME is not significantly related to ADE-
QUACY, and hence that ADEQUACY does not act as a mediator in the
impact process. Hypothesis 1 therefore remains only partially supported.
It is, however, conceivable that ADEQUACY might act as a modera-
tor in the impact process; typically, the e ffect of a ne gative outcome
decision on a participant ’s perceived career impact might vary, depende nt
on how adequate the participant perceives the process to be. If he thinks
the AC was a good one and he doe s badly at it then he is like ly to perceive
the AC as having a negative impact on his career, whereas if he thinks the
AC was a poor one and he does badly then he might perceive it to have
less of a negative impact on his career. While this hypothe sis was not sup-
porte d by statistical analysis, inte rview transcripts suggest it may well be
the case . Typically, one candidate who did poorly at the center (Outcome
D) thought the center was inade quate ‘‘too much onus is be ing put on
it—it’s not the be all and end all’’ and perceived it to have had very little
effect on his career: ‘‘I’ve actually been promote d to management in the
last 2 months despite the findings of the report! ’’ Interestingly, those who
did well perceived the center to be good (e.g., ‘‘it was an excellent test of
one ’s ability to act under pressure and time constraints—very much related
to the ‘‘real world’’ and it has he lped boost my care er to my current position
of ... much quicker’’). Several candidate s believed the assessment only pro-
vide d a small sample of behavior for assessors to monitor (e .g., ‘‘it’s a
snapshot’’; ‘‘picks up some things but doesn’t account for a bad day—tooth-
ache and so on’’; ‘‘if it’s not seen it’s not there!’’). Others be lieved it was
selecting people on the basis of the wrong criteria (e .g., ‘‘the AC is geared
toward extroverts, it funne ls out too many of the quieter type s—I be lieve
it’s funne ling more of the wrong type into management positions’’; ‘‘I think
anyone from my department would have been give n the same deve lopme nt
needs—we all have the same personality profile —all done the same de-
gree—all introve rts prepare d to work hard for exams—all the exercises
were geared to extrove rt sale speople ’’; ‘‘technical qualitie s are not given
158 Fran cis-Sm ythe an d Smith
enough importance ’’). Still others believed its ade quacy was very much de-
pendent on the individual concerned (e .g., ‘‘the center is as objective as
possible in a classroom environme nt but it’s an unfair way—it’s not suitable
or applicable to all individuals—it often brings out the worst in people—you’re in a false environme nt’’).
Regression analysis was used to test hypothe sis 2 and, hence, explore
the role of the pote ntial moderators (FEEDBACK, SUPPORT, POSI-
TION) of the relation between OUTCOME and IMPACT. The hypothe sis
was not supporte d as no statistically significant moderating effects were
found. For a full description of the analysis, see Appendix B.
Comments made at interview, however, supporte d the important roles
of FEEDBACK and POSITION in the impact process; those with good-
quality feedback felt the AC had he lped their career more than those with
poor-quality feedback, and those who were furthe r away from a middle -
management position fe lt it had he lped the ir career more than those who
were only one position away from middle manage ment. Typical positive
comments about feedback were: ‘‘I felt more positive after my feedback
chat than anything else’’; ‘‘my personal feedback was exce llent’’; and ‘‘ifyou know yourse lf well enough then the feedback shouldn’t surprise you.’’Negative comments include d: ‘‘vague , ambiguous, and not very contextual’’;‘‘long-winde d and deflating’’; and ‘‘feedback could have been much better.’’
Transcripts were analyze d for the numbe r of qualitative comments
made on each topic area. Reference was made to the ade quacy of the proc-
ess by 35% of candidate s, to managerial support by 29% , to the effects on
psychological state by 23% , and to the perceived impact on the ir career
and quality of feedback expe rience by 18% . Thus, while most comments
were made about the adequacy of the process, the least were made about
personal experience. It may be that this represents the relative importance
of these issues for candidate s or it may be that they simply fe lt more able
to expre ss their opinions on something tangible like the ade quacy of the
process than to discuss the ir own personal expe riences. This is an important
methodological point to note and supports the use of structure d closed
que stions when exploring areas in which some subje cts may feel unable to
expre ss themselves freely.
In summary, the results show that the perceived impact of the AC
process on one ’s career mediates between the effects of the AC outcome
decision and post-asse ssment states and work-relate d attitude s (partial sup-
port for H1), and that also the perceived quality of feedback and career
line distance are important predictors of perceived career impact. Hypothe -
sis 2 which states that they act as moderators was not supporte d. Good
feedback he lps attende es’ careers more than poor feedback and those at-
tende es who are 3 or 4 grade s below middle management are like ly to
Psych ological Im pact of Assessmen t 159
benefit more than those who are just 1 grade be low. A revised mode l of
psychological impact is propose d in Fig. 3.
DISCUSSION
This study provide s further insight into the dynamics of the psycho-
logical impact of assessment processes at work. Notably, be lie fs about the
ade quacy of the process do not mediate othe r responses in this study. This
is contrary to the findings of Robertson et al. (1991) and may be, as dis-
cusse d e arlie r, due to the fact that its role is more appropriate as a
moderator. Alternative ly, it may be due to the fact that the present study
was conducte d in more of a developmental context (where candidate s were
being selected for promotion but were also given guidance as to their de-
velopment needs) rather than a pure assessment context as in Robertson
et al.’s. Overall, one might expe ct the poignance of the assessment process
to be greater in selection/promotion contexts than in strictly deve lopme ntal
one s, thus entraining more comple x psychological processes and stronger
associations between selection decisions and response s. This hypothe sis is
furthe r supporte d by the Gilliland (1994) study where in se lecting stude nts
for temporary work, selected individuals saw greater fairness in the selec-
tion proce ss than re je cte d individ uals. Pe rhaps individuals se e king
promotion with the ir current employe r are more like ly to be obje ctive about
the ade quacy of the proce ss (and thus belief about ade quacy would be more
determined by the procedures used) than those seeking a job in a new
organization. The suggestion here then is that the conte xt of the asse ssment
may well moderate procedural justice effects. Typically, selection for a job
in a new organization may show greatest justice effects, followed by selec-
tion for promotion within an organization, followed by assessment for pure
developmental reasons. While previous work (see Gilliland, 1994) has con-
sidered procedural justice effects in different selection methods, the effect
of context of the selection/asse ssment appears to remain a matter for em-
pirical clarification.
Fig. 3. Revise d model of psychological impact.
160 Fran cis-Sm ythe an d Smith
While no evidence was obtaine d for the role of FEEDBACK, POSI-
TIO N, and SUPPORT as mode rator variable s in the impact proce ss,
FEEDBACK and POSITION were shown to be important variable s in the
prediction of IMPACT, while SUPPORT and ADEQUACY were not.
Closer examination of the data, however, revealed that SUPPO RT was sig-
nificantly and negative ly relate d to OUTCOME (r = ¯0.53, df = 30, p <
0.05) . There is thus a significant difference in managerial support between
OUTCOME groups; participants who have good managerial support are
more like ly to do well at the AC. The possibility of a causal re lation be-
tween these variable s needs to be explore d. It may simply be that people
who do well at the AC have more positive outlooks on their organizational
environme nts, and that this is reflected in their ratings of manage rial sup-
port. However, it is also possible that the measures of SUPPO RT taken
here indicate real diffe rences in the level of inte rest and involve ment shown
by managers for their subordinate s, and that this in turn is re late d to suc-
cess in assessment/de ve lopme nt contexts.
There are several methodological refinements that could be suggested
in further studies of this topic. First, notwithstanding the argume nts in favor
of retrospective pre¯post tests of attitude change in some areas, it would
be beneficial to gathe r genuine pre - and post-asse ssment measure s on many
of the variable s used in this study. Additionally, given that many of the
measures are derived through self-report, the issue of common method
variance rathe r than causal links explaining some of the observed relation-
ships should be raise d. In response , both inte rvie w and que stionnaire
technique s were employe d, and well-de ve lope d measure s with sound psy-
chometric prope rties were used, factors that should help to dispe l concerns
about the role of common method variance .
The size of the present sample is also small, and the generalizability
of these findings consequently limited. However, give n the limite d number
of employe es that had undertake n the AC and the limited number of these
who satisfied the selection criteria, it was decided at the outset that the
proble ms imposed by the small sample size were ade quate ly compensated
for by the exploratory nature of the study and the need for more qualitative
data. To the extent that our earlier conjectures about the relative ly muted
poignancy of assessment in a developmental as oppose d to a selection/pro-
motion context are valid, then the emergence of any statistically significant
results here is all the more impressive, especially give n the small number
of people studie d.
Although the effects of gende r were strictly controlle d in this study,
future research in this area should explicitly addre ss the possibility of sex-
relate d effects if possible . Candidate s’ gende r could possibly affect outcome
decisions made by assessors, the reactions of candidate s to those decisions,
Psych ological Im pact of Assessmen t 161
resultant effects on the candidate s, and the moderating effects of manage -
rial support and feedback.
The implicit message s for organizations engage d in or contemplating
using assessment or deve lopme nt center technology is clear: be aware that
the potential gains to individuals of assessment/de ve lopme nt opportunitie s
may be dependent on their career positions and ensure that high-quality
feedback is available to all. In order to minimize the potentially-ne gative
impact of such procedure s, careful feedback, counse ling, and deve lopme nt
recommendations are of vital importance . We should be seeking as best
we can to understand the ways in which assessment procedure s themselves
can enhance or unde rmine broade r organizational obje ctive s and individual
motivation.
ACKNOWLEDGMENTS
The authors wish to thank Professor Ivan Robertson for helpful com-
ments on an earlier version of this pape r.
APPENDIX A—DETAILS OF MEASURES
OUTCOME. Asse sse d on a 4-point scale : A= Excelle nt, B= Good
(mee ts standard) , C= Just Misse s Standard, or D= Substantial ly Misse s
Standard. This represented the final assessment rating give n after consid-
eration of a participant ’s performance across six exercises (presenting a
proposal, constructing a plan, running a hote l, in-tray exercise , developing
a work unit, and preparing and presenting a board report) . Each exercise
was assessed on each of six dimensions (energy and drive , results orienta-
tion, workstyle , analytical ability, emotional adjustment, and interpersonal
skills) using Grades A+ (high) through to D¯(low). A matrix of marks
across all exercises and dimensions was produced for each participant. The
final assessment rating was a subje ctive inte rpretation of ove rall perform-
ance based on this matrix of scores.
POSITION. Number of participant positions is away from the position
being assessed (maximum = 4, minimum = 1, mean = 1.7) .
SUPPORT. Three aspects of support were rated by participants in the
inte rview on 5-point scale s (5 = good support) : the participant ’s perception
of the extent of guidance give n, the extent to which an action plan was
constructed, and the extent to which the manager had facilitate d opportu-
n i t i e s for t he pa rt ic ip an t t o e ngage in oth e r work e xp e rie nc e .
Intercorre lations between the three aspects of manage ment support were
all significant at approximate ly r = 0.8 (df = 30, p < 0.05). The ir ave rage
was take n to give the measure of SUPPORT.
162 Fran cis-Sm ythe an d Smith
FEEDBACK. Quality of feedback give n to the participant by the as-
sessor in terms of its perceived usefulness. The measure was take n as the
ave rage of five aspects of assessor feedback rated during the inte rview: the
extent of guidance give n by the assessor in formulating an action plan, the
availability of relevant information in the AC report to formulate a plan,
the extent to which the feedback provide d a full synopsis of the participant ’sperformance , the extent to which it gave guidance as to how to develop
skills the participant already had, and those skills he was lacking. Items
were rated on 5-point scale s (5 = good feedback) . Item inte rcorrelations
were moderate but significant at approximate ly r = 0.5 (df = 30, p < 0.05)
and their average was take n to give the measure of FEEDBACK.
ADEQUACY. As per Robertson et al. (1991) through three items: ‘‘itaccurate ly measures important qualitie s,’’ ‘‘it is a fair way of assessing po-
tential,’’ and ‘‘it should be used in selection,’’ on 5-point scales ranging
from Strongly Disagree (1) to Strongly Agree (5). With a sample of 140
employe es going through a management development and tiering scheme
in a major U.K. company, the mean score previously obtaine d was 3.19
with SD = 1.08 and reliability (coefficient a ) of 0.86 (Robertson et al.,
1991) .
IMPACT. As per Robertson et al. (1991) through two items: ‘‘it has
helped my career,’’ and ‘‘taking it has not harmed my career,’’ again meas-
ure d on 5-point scale s (5 = positive impact) . With a sample of 140
employe es going through a management development and tiering scheme
in a major U.K. company, the mean score obtaine d was 3.39 (SD = 1.17)
and a reliability (coefficient a ) of 0.74 (Robertson et al., 1991).
ESTEEM. As per Rosenbe rg (1965) with a 10-ite m Guttman scale
(e.g., ‘‘On the whole , I am satisfie d with myself’’; ‘‘I take a positive attitude
toward myse lf’’) using a 4-point response scale from Strongly Disagree (1)
to Strongly Agree (4). Publishe d scale reliability is 0.92.
COMMITMENT. As per Mowday (1979) , through 15 items (e.g., ‘‘Iam proud to te ll othe rs I am part of this organization ’’) using a 7-point
scale (7 = high commitment). Publishe d reliability is 0.90 as measure d by
Cronbach’s alpha. Other studie s of manage ment trainees have found mean
value s of 4.7, 5.05, and 6.1 (Fletcher, 1991; Robertson et al., 1991; Mowday,
1979, respectively) .
INVOLVEMENT. As per Lodahl (1965) using a 20-ite m scale (e .g.,
‘‘I’ll stay overtime to finish a job, even if I’m not paid for it’’) with 4-point
response scale s ranging from Strongly Disagree (1) to Strongly Agree (4).
Publishe d split-half reliability coefficients are 0.8 and 0.67, respectively.
PLANNING. As per Gould (1979) using a six-ite m scale (e.g., ‘‘I have
a plan for my career’’) with a 5-point response scale ranging from Strongly
Psych ological Im pact of Assessmen t 163
Agree (1) to Strongly Disagre e (5). High scores indicate a low degree of
planning. Reliability (coefficient a ) of 0.8.
D COMMITMENT, D INVOLVEMENT, D PLANNING . They were
measure d by asking participants first to allocate a retrospe ctive score from
1 (low) to 10 (high) to indicate how they fe lt before (pre) the AC on the
job-re lated constructs, and then to allocate a score for how they felt now
(post) on the same construct using the same scale. The pre score was then
deducted from the post score to give the change ( D ) score. Correlations
were calculate d between the post-measures of the subjective change meas-
ures of commitment, involve ment, and planning and the que stionnaire
measure s of commitment, involve ment, and planning. The questionnaire
measure s had been include d mainly to give participants some indication of
the constructs on which they would be asked to make retrospe ctive com-
ment. The corre lations between the change measures and their respective
que stionnaire measure s were r = 0.6 (p < 0.01) , r = 0.3 (p < 0.05) , r =
0.65 (p < 0.01) . This was take n to signify some degree of validity for the
change measures.
APPENDIX B—DETAILS OF STATISTICAL ANALYSES
To Test for IMPACT as a Mediator in the OUTCOME-Affective
Respon se Process
The analysis is base d on the assumption that when the cognitive re-
sponse variable is used as a predictor of affective reactions, the addition
of variable s earlie r in the chain (i.e ., outcome decision) should not improve
significantly the prediction of affective reactions (i.e ., a statistically signifi-
cant increase in R2). Three separate regression analyse s were conducted
using the post-asse ssment difference measure s as dependent variable s. IM-
PACT was entered followe d by OUTCOME. The addition of OUTCOME
to the regression mode l did not significantly increase the value of R2 for
any of the response variable s, thus confirming the role of perceived career
impact as a mediator between the outcome decision and subse quent affec-
tive response s (see Table BI).
To Test for Moderating Effects of FEEDBACK, SUPPORT, an d
POSITION in the Impact Process
According to James and Brett (1984) , a variable z is a moderator if
the relationship between two or more other variable s, say x and y, is a
function of the leve l of z, thus indicating an x by z inte raction. The roles
of the variable s O UTCOME, ADEQUACY, FEEDBACK, SUPPO RT,
164 Fran cis-Sm ythe an d Smith
and POSITION, as predictors of IMPACT, were inve stigated through step-
wise multiple regression. OUTCOME, FEEDBACK, and POSITION were
found to be the only statistically significant predictors giving an R2 of 0.47
(p < 0.05) (see Table BII).
In orde r to explore the role of FEEDBACK and POSITION as mod-
e rators in the im pact proc e ss , in te rac t ion te rm s ( O U TC O ME *FEEDBACK and OUTCOME * POSITION) were built into the regression
mode l after the single variable s had been entered. The interaction terms
were not significant thus sugge sting that FEEDBACK and POSITION were
not acting as moderators.
Table BI. Regression Analyses with Post-Assessment States and Work-Related Attitudes as
Dependent Variablesa
Independent variables
D Commitme nt D Involvement D Planning
R2
B p R2
B p R2
B p
1. Impact 0.80 0.02 0.77 0.01 0.65 0.02
2. Outcome
R2 before Outcome 0.29 0.20 0.40
R2 after Outcome 0.30 0.21 0.42
Increase in R2
0.01 0.01 0.02
aStepwise procedures were used in the regression analysis and independent variable s entered
in the regre ssion equation only if they explained a significant amount of variance in the
dependent variable .
Table BII. Multiple Regression Analysis Testing the Roles of
De cision O utcome , B e lie f About Ade quacy, Q uality of
Fee dback, Leve l of Managerial Support, and Current Career
Poisition in the Impact Proce ssa
Step
Variable
entered R R2
R2
change F value
1 Feedback 0.53 0.28 11.6*
2 Outcome 0.63 0.39 0.07 9.3**
3 Grade 0.69 0.47 0.08 7.3**
aStepwise procedures were used in the regre ssion analysis and
independent variables entered in the regression equation only
if they explained a significant (p < 0.05) amount of variance
in the dependent variable.
*p < 0.01.
**p < 0.001.
Psych ological Im pact of Assessmen t 165
REFERENCES
ALBAN-ME TCALFE, B. Different Gende r—Different Rules? Assessment of Women in
Management. In P. Barre r and C. Cooper (Eds.), Managing organisations in 1992: Strategic
response. London: Routledge, 1991.
BIES, R. J., & MOAG, J. S. Interactional justice: Communications criteria of fairness. In R.
Le wicki, B . She ppard, and M. H. Baze rman (Eds.) , Research on negotiation s in
organisations . Gree nwich, CT: JAI Press, 1986.DREHER, G. F., & SACKETT, P. Perspectives on em ployee staffing and developm ent: Readings
and com m entary. Homewood, IL: Irwin, 1983.FLE TCHER , C. A. Candidates reactions to asse ssme nt centres and their outcomes: A
longitudinal study. Journal of Occupationa l Psychology, 1991, 64, 117-127.GILLILAND, S. W. The perceived fairness of selection systems: An organisational justice
perspective . Academ y of Managem ent Review, 1993, 18, 694-734.GILLILAND, S. W. Fairness from the applicant’s perspective: Reactions to employee selection
procedures. International Journal of Selection and Assessment, 1995, 3(1), 11-19.GILLILAND, S. W. Effects of procedural and distributive justice on reactions to a selection
system. Journal of Applied Psychology, 1994, 79(5), 691-701.GOULD, S. Characteristics of career planners in upwardly mobile occupations. Academ y of
Managem ent Journal , 1979, 22(3), 539-550.GUTEK, B. A., & WINTER, S. J. Consistency of job satisfaction across situations: Fact or
framing artifact? Journal of Vocational Behaviour, 1992, 41(1), 61-78.HERRIO TT, P. Selection as a social process. In M. Smith and I. Robertson (Eds.), Advances
in selection and assessm ent. Chichester: John Wiley and Sons, 1989.HO OGSTRATEN, J. Influence of objective measure s on se lf-reports in a re trospective
pretest-posttest design. Journal of Experimental Education, 1985, 53, 207-210.HOWARD, G. S., RALPH, K. M., GULANICK, N. A., MAXWELL, S. E., NANCE, D. W.,
& GERBER, S. K. Internal validity in pretest-posttest self-report evaluations and a
re-evaluation of re trospective pretests. Applied Psychological Measurem ent, 1979, 3(1),
1-23.
ILGEN, D. R., FISHER, C. D., & TAYLO R, S. M. Consequences of individual feedback on
behavior in organizations. Journal of Applied Psychology, 1979, 64, 349-371.
ILES, P. A., & RO BERTSO N, I. T. The impact of personnel se lection procedures on
candidates. In P. Herriot (Ed.), Assessment and selection in organisations. Chichester: John
Wiley and Sons, 1989.JAMES, L. P., & BRETT, J. M. Mediators, moderators and tests for mediation. Journal of
Applied Psychology, 1984, 69, 307-321.LO DAHL, T. M., & KEJNER, M. The definition and measure me nt of job involveme nt.
Journal of Applied Psychology, 1965, 49(1) , 24-33.MOWDAY, R. T., & STEERS, R. M. The measure me nt of organisational commitment.
Journal of Vocational Behavior, 1979, 14, 224-247.NOE, R. A., & STEFFY, B. D. The influence of individual characteristics and assessment
ce nter evaluation on care er exploration be havior and job involve me nt. Journ al of
Vocational Behavior, 1987, 30, 187-202.
POHL, N. F. Using retrospective pre -ratings to counteract response-shift confounding. Journal
of Experimental Education, 1982, 50, 211-214.
ROBERTSO N, I. T., & SMITH, J. M. Personnel selection me thods. In J. M. Smith and I.
T. Robertson (Eds.), Advances in selection and assessm ent. Chichester: John Wiley & Sons,
1989.ROBERTSO N, I., ILES, P. A., GRATTO N, L., & SHARPLEY, S. The impact of personnel
selection and assessme nt methods on candidates. Hum an Relations, 1991, 44(9), 963-982.ROSENBERG, M. Society and the adolescent self-image. Prince ton Unive rsity Press, 1965.
SCHULER, H. Social validity of selection situations: A concept and some empirical results.
In H. Schuler, J. L. Farr, and M. Smith (Eds.), Personn el selection an d assessm en t:
Individual and organisational perspectives. Hillsdale, NJ: Erlbaum., 1993.
166 Fran cis-Sm ythe an d Smith
SMITHER, J. W., REILLY, R. R., MILLSAP, R. E., PEARLMAN, K., & STOFFEY, R.
W. Applicant reactions to se lection procedures. Personnel Psychology, 1993, 46, 49-76.SPRANGERS, M., & HOOGSTRATEN, J. Pretesting effects in retrospective pre test¯posttest
design. Journal of Applied Psychology, 1989, 74(2), 265-272.TERBO RG, J. R., & DAVIS, G. A. Evaluation of a new method for assessing change to
planned job redesign as applied to Hackman and Oldham’s Job Characte ristics Model.Organisational Behaviour and Hum an Perform ance, 1982, 29, 112-128.
THO RNTO N, G. C., & BYHAM, W. C. Assessment cen ters and managerial perform ance. NewYork: Academic Press, 1982.
BIOGRAPHICAL NOTES
JAN FRANCIS-SMYTHE BSc (Hons), MSc is a Senior Lecturer in Psychology and BusinessManage ment at Worcester College of Higher Education. She is curre ntly carrying out research
for a PhD with Professor Ivan Robertson at UMIST into individual differences in attitudestoward time.
PHILIP M. SMITH BSc (Hons), PhD was, at the time the research was conducted, course
organizer for the MSc in Organisational Psychology at Bristol University. He now works forOpus Consulting in Bristol.
Psych ological Im pact of Assessmen t 167