managerial performance development constructs and ... · each of the five factors showed a distinct...

24
Managerial Performance Development Constructs and Personality Correlates James M. Conway Department of Psychology Central Connecticut State University The goals of this study were (a) to identify managerial performance development con- structs through factor analysis, (b) to understand their motivational determinants us- ing personality correlates, and (c) to examine differences between rating sources. Factor analyses identified 5 developmental constructs: Interpersonal Effectiveness, Willingness to Handle Difficult Situations, Teamwork and Personal Adjustment, Adaptability, and Leadership and Development. Comparisons with Borman and Brush’s (1993) managerial performance megadimensions showed that the develop- mental constructs overlapped with but also added to the day-to-day performance do- main. Each of the five factors showed a distinct pattern of personality correlates. Per- sonality correlates supported hypotheses based on socioanalytic theory regarding the motive to get along with others (e.g., Interpersonal Effectiveness correlated with em- pathy and agreeableness) versus the motive to get ahead (e.g., Willingness to Handle Difficult Situations correlated with potency measures). Rating sources (supervisor, peer, subordinate, and self) showed some differences in their results. Several published works have addressed the constructs that make up the manage- rial job performance domain (e.g., Borman & Brush, 1993; McCauley, Lombardo, & Usher, 1989; Tornow & Pinto, 1976). These studies have tended to focus on job analysis-type data to identify the behaviors and activities managers engage in. Borman and Brush’s work in this area is probably the most comprehensive because they summarized and analyzed performance dimensions generated from critical in- cidents and task statements in a wide sample of organizations. They developed a set of 18 managerial “megadimensions” such as Training, Coaching, and Developing Subordinates. HUMAN PERFORMANCE, 13(1), 23–46 Copyright © 2000, Lawrence Erlbaum Associates, Inc. Requests for reprints should be sent to James M. Conway, Department of Psychology, Central Con- necticut State University, New Britain, CT 06050–4010.

Upload: vuongdien

Post on 30-Jul-2018

213 views

Category:

Documents


0 download

TRANSCRIPT

Managerial Performance DevelopmentConstructs and Personality Correlates

James M. ConwayDepartment of Psychology

Central Connecticut State University

The goals of this study were (a) to identify managerial performance development con-structs through factor analysis, (b) to understand their motivational determinants us-ing personality correlates, and (c) to examine differences between rating sources.Factor analyses identified 5 developmental constructs: Interpersonal Effectiveness,Willingness to Handle Difficult Situations, Teamwork and Personal Adjustment,Adaptability, and Leadership and Development. Comparisons with Borman andBrush’s (1993) managerial performance megadimensions showed that the develop-mental constructs overlapped with but also added to the day-to-day performance do-main. Each of the five factors showed a distinct pattern of personality correlates. Per-sonality correlates supported hypotheses based on socioanalytic theory regarding themotive to get along with others (e.g., Interpersonal Effectiveness correlated with em-pathy and agreeableness) versus the motive to get ahead (e.g., Willingness to HandleDifficult Situations correlated with potency measures). Rating sources (supervisor,peer, subordinate, and self) showed some differences in their results.

Several published works have addressed the constructs that make up the manage-rial job performance domain (e.g., Borman & Brush, 1993; McCauley, Lombardo,& Usher, 1989; Tornow & Pinto, 1976). These studies have tended to focus on jobanalysis-type data to identify the behaviors and activities managers engage in.Borman and Brush’s work in this area is probably the most comprehensive becausethey summarized and analyzed performance dimensions generated from critical in-cidents and task statements in a wide sample of organizations. They developed a setof 18 managerial “megadimensions” such as Training, Coaching, and DevelopingSubordinates.

HUMAN PERFORMANCE,13(1), 23–46Copyright © 2000, Lawrence Erlbaum Associates, Inc.

Requests for reprints should be sent to James M. Conway, Department of Psychology, Central Con-necticut State University, New Britain, CT 06050–4010.

RESEARCH ON HOW MANAGERS DEVELOP

McCauley et al. (1989) suggested that focusing specifically on how managers de-velop provides a different perspective as compared to focusing on what managersdo (Borman & Brush’s, 1993, focus). This perspective is potentially useful for tworeasons: (a) it is particularly relevant to personal development interventions such asmultisource feedback and (b) it can potentially identify important constructsmissed by analyses of managers’ behaviors and activities. This study used an in-strument called Benchmarks (McCauley et al., 1989), specifically developed basedon a critical incidents analysis of developmental experiences.

Benchmarks is based on a series of studies described by Lindsey, Homes, andMcCall (1987) and McCall, Lombardo, and Morrison (1988). In these studies ex-ecutives were asked to describe several “key events” (i.e., critical incidents) intheir careers—events that made a difference in their managerial approaches. Theexecutives were also asked to describe the lessons they learned from each event.This research effort was based on incidents from a wide variety of organizations.McCauley et al. (1989) described how the data were used to form the “Skills andPerspectives” section (the largest section) of Benchmarks. There are 16 Skills andPerspectives dimensions, listed and defined in Table 1 (Center for Creative Lead-ership, 1995; McCauley et al., 1989). McCauley et al. noted that Benchmarks in-cludes dimensions not found in other performance feedback instruments, such asStraightforwardness and Composure, and Acting With Flexibility (balancing op-posite qualities—e.g., being both tough and compassionate), which is an interper-sonal skill. The Benchmarks dimensions have good validity; McCauley et al.reported that supervisor ratings on Benchmarks for several organizations were sig-nificantly related to a variety of criteria, including a measure of organizational ad-vancement 24 to 30 months after Benchmarks was administered.

With regard to higher order developmental constructs underlying Benchmarks,McCauley et al. (1989) identified three using a post hoc conceptual analysis: Re-spect for Self and Others, Adaptability, and Molding a Team. Respect for Self andOthers included learning to show compassion, to be straightforward, and to putpeople at ease. McCauley et al. argued that Self-Awareness and Balance BetweenPersonal Life and Work belong in this category too because an appreciation of theself is important for dealing well with others. Adaptability is referred to in currentBenchmarks literature as “Meeting Job Challenges.” Meeting Job Challenges in-cluded developing resourcefulness, decisiveness, the drive to do what it takes toaccomplish goals, and the ability to learn quickly. Finally, Molding a Team, re-ferred to in current literature as “Leading People,” included learning to identify,recruit, and hire talented staff; to set a developmental climate; to confront problemsubordinates; and other team-related behaviors. McCauley et al. argued thatLeading People depends to some extent on the skills and perspectives inherent inthe other two constructs but also adds unique team-oriented components.

24 CONWAY

GOALS AND HYPOTHESES

Identifying Developmental Performance Constructs

My first goal in this study was to use factor analysis to determine whetherMcCauley et al.’s (1989) three constructs optimally describe the performance de-velopment domain. A related goal was to examine how consistent the constructswere across ratings by managers’ supervisors, peers, subordinates, and the manag-ers themselves. Meta-analytic evidence on between-source correlations finds su-pervisor and peer ratings to be similar (Conway & Huffcutt, 1997; Harris &

MANAGERIAL PERFORMANCE DEVELOPMENT 25

TABLE 1Benchmarks Skills and Perspective Dimensions

1. Resourcefulness(17 items): Can adapt to changing and often ambiguous circumstances, thinkstrategically, make good decisions under pressure, set up complex work systems, engage in flexi-ble problem-solving behavior, and work effectively with higher management.

2. Doing Whatever it Takes(14 items): Has perseverance and focus in the face of obstacles, e.g.,taking charge, facing difficult situations with guts and tenacity.

3. Being a Quick Study(4 items): Quickly masters new technical and business knowledge.4. Decisiveness(4 items): Prefers quick and approximate actions to slow and precise ones in many

management situations.5. Leading Employees(13 items): Delegates to employees effectively, broadens their opportunities,

and acts with fairness toward them.6. Setting a Developmental Climate(5 items): Provides a challenging climate to encourage

employees’ development.7. Confronting Problem Employees(4 items): Acts decisively and with fairness when dealing with

problem employees.8. Work Team Orientation(4 items): Accomplishes tasks through managing others.9. Hiring Talented Staff(3 items): Hires talented people for his or her team.10. Building and Mending Relationships(11 items): Knows how to build and maintain working

relationships with coworkers and external parties (e.g., finding common ground, negotiating,understanding others, and getting cooperation in non-authority relationships).

11. Compassion and Sensitivity(4 items): Shows genuine interest in others and sensitivity toemployees’ needs.

12. Straightforwardness and Composure(6 items): Is honorable and steadfast (e.g., relies onfact-based positions and doesn’t blame others for mistakes).

13. Balance Between Personal Life and Work(4 items): Balances work priorities with personal lifeso that neither is neglected.

14. Self-Awareness(4 items): Has an accurate picture of strengths and weaknesses and is willing toimprove.

15. Putting People at Ease(4 items): Displays warmth and a good sense of humor.16. Acting With Flexibility(5 items): Can behave in ways that are often seen as opposites (e.g.,

being both tough and compassionate).

Note. This table is based on Center for Creative Leadership (1995) and McCauley, Lombardo, &Usher (1989).

Schaubroeck, 1988). However, these two meta-analyses showed that self-ratingswere not highly related to ratings by other sources, and Conway and Huffcutt(1997) provided the same type of evidence for subordinate ratings. One study di-rectly compared factor structures across sources, finding no difference for peersand subordinates (Maurer, Raju, & Collins, 1998). Maurer et al. used a rating in-strument containing only a single factor, team-building, and unlike this study didnot include supervisor or self-ratings.

Understanding Motivational Determinants by ExaminingPersonality Correlates

The second goal of this study was to evaluate the motivational determinants of per-formance development constructs using correlations with personality constructs.Recent research has shown that there are clear patterns of performance–personalityrelations, and that these relationships can help understand the motivational basis ofperformance constructs (Hogan, 1998; Hogan, Rybicki, Motowidlo, & Borman,1998). The personality measures used in this study were the California Psychologi-cal Inventory (CPI; Gough, 1987) and the Myers–Briggs Type Indicator (MBTI;Myers & McCaulley, 1985).

A promising theoretical framework for understanding performance develop-ment constructs is the socioanalytic approach to personality (Hogan & Shelton,1998). According to socioanalytic theory, two primary motives driving behaviorare the desire to “get along” and the desire to “get ahead.” Getting along meansfeeling liked and supported, whereas getting ahead means gaining power and con-trol of resources (Hogan & Shelton, 1998). Interactions in work settings can beseen as attempts to achieve one or both of these goals (Hogan et al., 1998; Hogan &Shelton, 1998). Hogan and Shelton noted that some people are more motivated toget along, whereas others are more motivated to get ahead and therefore differentpeople use different behavioral strategies.

The first performance development construct isRespect for Self and Others.This highly interpersonal construct (including, e.g., Compassion and Sensitivity)should be influenced by characteristics important for getting along. Therefore, Re-spect for Self and Others should be positively correlated with CPI Empathy (a so-cial skill that should increase the ability to get along; Hogan, 1991) and MBTIThinking–Feeling (interpreted by McCrae & Costa, 1989, as a measure of agree-ableness). McCauley et al. (1989) linked knowledge of self to mental health,which is conceptually related to adjustment. Respect for Self and Others shouldtherefore be related to CPI Self-Control (Gough, 1987, provides evidence showingthat this dimension is correlated with measures of emotional stability).

The second performance development construct wasMeeting Job Challenges.This construct includes resourcefulness, decisiveness, the ability to learn quickly,

26 CONWAY

and the drive and attitudes to do these things. Taking charge, overcoming obsta-cles, and being decisive should be related to the characteristics necessary to getahead rather than those necessary to get along. In a management situation onecharacteristic that is probably important for getting ahead is potency. According toHough (1992), “Potency is defined as the degree of impact, influence, and energythat one displays” (p. 144). Meeting Job Challenges should therefore correlatepositively with CPI Dominance.

The third performance development construct wasLeading People,includinghiring, developing, and motivating subordinates. McCauley et al. (1989) sug-gested that Leading People depends partly on the first two constructs, Respect forSelf and Others and Meeting Job Challenges. Therefore, a manager’s standing onLeading People reflects the characteristics necessary to get along and those neces-sary to get ahead. This balancing act should be predicted by “getting-along” di-mensions such as CPI Empathy and “getting-ahead” dimensions such as CPIDominance.

Finally, differences between rating sources with respect to personality corre-lates were examined. These analyses were exploratory and there were no particularhypotheses.

METHOD

Participants

A total of 2,110 managers from a variety of industries and management levels par-ticipated in a leadership development seminar. Each manager was rated on Bench-marks by at least one supervisor, one peer, and one subordinate as well as by him-self or herself. Ratings were not used for any administrative decisions. Of the 2,110managers, 1,830 completed the CPI and 1,567 completed the MBTI. The CPI andMBTI were not used administratively. Because my goal was to factor analyze eachsource’s ratings and make comparisons across sources, I included only one memberof each source (when there were multiple members from a source I randomly choseone member).

Demographic data showed that participants represented many organizationaltypes (e.g., manufacturing, finance, public sector, nonprofit; 88% were from pri-vate for-profit organizations), functional areas (e.g., accounting, marketing), andlevels of management. Most managers were White (94%), male (70%), and had atleast a Bachelors degree (89%). The mean age was 42 years.

Instruments

Benchmarks. The Benchmarks Skills and Perspectives section contains 106items (statements such as “has personal warmth”) forming 16 dimensions as de-

MANAGERIAL PERFORMANCE DEVELOPMENT 27

scribed in Table 1. The development of Benchmarks was described in detail byMcCauley et al. (1989), who reported coefficient alphas for supervisor ratings of.75 or higher for each of the 16 dimensions. McCauley et al. also reported signifi-cant validity coefficients for predicting organizational advancement.

CPI. CPI Form 462 (Gough, 1987) includes 462 true–false items measuring20 “folk constructs” (concepts people use in everyday language to describe them-selves and others). Gough reported coefficient alphas for the CPI scales rangingfrom .52 to .79. Only 19 of the CPI constructs were used in this study (data for Femi-ninity/Masculinity were not available).

MBTI. The MBTI Form F (Myers & McCaulley, 1985) consists of 166 items,of which 95 are scored. Although the MBTI is intended to measure personalitytypes, McCrae and Costa (1989) presented evidence that the MBTI measures con-tinuous dimensions. Eight unipolar dimensions form four pairs (e.g., Extraversionand Introversion), and the two unipolar dimensions in each pair are almost com-pletely ipsative (McCrae & Costa, 1989). In this study I used four bipolar dimen-sions, each based on a combination of a pair of unipolar dimensions. McCrae andCosta suggested that these four bipolar dimensions represent four of the Big Fivepersonality factors: MBTI Extraversion–Introversion (EI) measures extraversion,MBTI Sensing–Intuition (SN) measures openness to experience, MBTIThinking–Feeling (TF) measures agreeableness, and MBTI Judging–Perceiving(JP) measures conscientiousness. Myers and McCaulley reported coefficientalphas ranging from .76 to .83.

Overview of Analyses

In this study I used a combination of exploratory factor analyses (EFA) and confir-matory factor analyses (CFA) to identify higher order constructs underlyingBenchmarks. CFA is appropriate when a strong theory exists about the commonfactor structure. Although McCauley et al. (1989) speculated on the higher orderconstructs of Benchmarks, this is not strong enough to justify CFA. CFA also usu-ally requires allowing a variable to load only on its hypothesized factor (Hurley etal., 1997), even though a variable might load on multiple factors includingunhypothesized ones. EFA, on the other hand, is well-suited to identifying multipleloadings.

I began with an EFA of the 16 supervisor-rated Benchmarks scales. I used theresulting parameter estimates to derive a hypothesized model for the other threesources. This hypothesized model was fit to each of the other sources’ ratings us-ing CFA. These CFAs tested whether the factor structures of peer, subordinate,and self-ratings were the same as that for supervisor ratings. Supervisors were usedas the “standard” because their ratings represent the most common and best ac-

28 CONWAY

cepted way to measure work performance (Murphy & Cleveland, 1995), and be-cause two meta-analyses have shown supervisor ratings to be more reliable thanpeer ratings (Conway & Huffcutt, 1997; Viswesvaran, Ones, & Schmidt, 1996) orsubordinate ratings (Conway & Huffcutt, 1997).

Finally, factor scores based on the factor analyses were correlated with CPI andMBTI dimensions.

RESULTS

Factor Analyses

EFA for supervisor ratings. Supervisor ratings on the 16 BenchmarksSkills and Perspectives scales (means, standard deviations, and correlations appearin Table 2) were subjected to a maximum likelihood EFA using squared multiplecorrelationsasprior communalityestimates.Ford,MacCallum,andTait (1986)andMacCallum (1998) advocated this type of common factor analysis over techniquessuch as principal components analysis. SAS Proc Factor software (Version 6; SASInstitute, 1989) was used and a Promax oblique rotation was conducted. Criteria fornumber of factors included the root mean square error of approximation (RMSEA;Browne & Cudeck, 1992), the Tucker–Lewis Index (TLI), freedom from improperestimates (communalitiesgreater than1,whichcanoccurwithmaximumlikelihoodestimation), and the interpretability of the solution. A TLI of at least .90 is usuallytaken to indicate acceptable fit. Browne and Cudeck suggested that RMSEA valuesless than .05 indicate a close fit between the model and the data, and RMSEA valuesless than .08 indicate reasonable fit. MacCallum (1998) further recommended ex-amining confidence intervals for RMSEA to assess the fit statistic’s accuracy. Dueto the large sample size in this study all 90% confidence intervals were small, ap-proximately .01 in width, and the intervals will not be reported further later.

The number of factors was determined by examining solutions with numbers offactors ranging from one to six. The six-factor solution contained an estimatedcommunality greater than one, which can indicate poorly defined factors (e.g.,only a single variable loading on a factor; McDonald, 1985).

One- and two-factor solutions showed large RMSEA values of .160 and .116,respectively, and low TLI values of .71 and .85, respectively, necessitating the re-jection of these models. The three-factor model showed a large RMSEA of .092and a barely acceptable TLI of .90, indicating this model should also be rejected. Itis worth noting, however, that the factor structure for this model was very consis-tent with McCauley et al.’s (1989) hypothesized constructs.

Four- and five-factor solutions had reasonable RMSEA values (i.e., less than.08) of .072 and .065, respectively, and relatively high TLI values of .94 and .95,respectively. The five-factor solution had the lower RMSEA value, the higher TLIvalue, and the more interpretable pattern of loadings (standardized regression co-

MANAGERIAL PERFORMANCE DEVELOPMENT 29

TA

BLE

2M

eans

,Sta

ndar

dD

evia

tions

,and

Cor

rela

tions

for

Sup

ervi

sor

Rat

ings

Dim

en

sio

nM

SD

12

34

56

78

91

01

11

21

31

41

51

6

1.R

esou

rcef

ulne

ss3.

66.5

31.

002.

Doi

ngW

hate

ver

itT

akes

3.82

.54

.78

1.00

3.B

eing

aQ

uick

Stu

dy4.

05.6

2.6

4.5

91.

004.

Dec

isiv

enes

s3.

64.7

7.4

6.6

5.3

91.

005.

Lead

ing

Em

ploy

ees

3.59

.57

.67

.58

.35

.25

1.00

6.S

ettin

ga

Dev

elop

men

tal

Clim

ate

3.77

.57

.65

.66

.39

.38

.80

1.00

7.C

onfr

ontin

gP

robl

emE

mpl

oyee

s3.

39.7

1.5

3.5

6.3

1.4

6.5

1.4

91.

00

8.W

ork

Tea

mO

rient

atio

n3.

60.7

3.3

8.2

9.1

2.1

4.6

3.4

9.3

41.

00

9.H

iring

Tal

ente

dS

taff

3.70

.64

.52

.53

.34

.31

.55

.58

.48

.41

1.00

10.

Bui

ldin

g/M

endi

ngR

elat

ions

hips

3.63

.65

.68

.51

.32

.17

.72

.59

.35

.43

.39

1.00

11.

Com

pass

ion

and

Sen

sitiv

ity3.

75.6

1.4

2.3

5.1

6.0

8.6

6.5

9.2

7.4

0.3

7.6

11.

00

12.

Str

aigh

tforw

ardn

ess

/Com

posu

re4.

14.6

1.4

5.3

3.2

4.0

8.4

7.3

7.2

5.3

0.2

5.5

1.3

51.

00

13.

Bal

ance

:Per

sona

lLi

fean

dW

ork

3.85

.75

.15

.06

.09

.03

.25

.15

.11

.32

.14

.21

.27

.26

1.00

14.

Sel

f-A

war

enes

s3.

62.7

0.5

9.5

1.3

2.2

2.6

1.5

3.3

7.3

6.3

8.6

5.5

1.4

8.1

91.

0015

.P

uttin

gP

eopl

eat

Eas

e3.

90.7

7.3

6.3

3.1

2.1

1.5

2.4

5.1

6.3

2.2

8.6

6.6

1.3

0.2

3.4

61.

00

16.

Act

ing

With

Fle

xibi

lity

3.64

.62

.69

.63

.36

.31

.76

.69

.49

.47

.48

.76

.61

.48

.24

.66

.57

1.00

efficients). I therefore accepted the five-factor solution, and the loadings anduniquenesses (a uniqueness was calculated as 1 minus the communality) areshown in Table 3. Factor correlations are shown in Table 4.

The fact that the RMSEA value was above .05 indicates that the model did notfit extremely well. However, the poor results for the six-factor model suggest thatfive factors is the most reasonable solution for these data.

CFA for peer, subordinate, and self-ratings. Next I applied the supervisorparameter estimates shown in Tables 3 and 4 to peer, subordinate, and self-ratingsusing CFA. Using LISREL 8 I fixed all model parameters at the supervisor values,and then separately assessed the fit of this model to each source’s data. This repre-sents the strongest possible form of CFA, in which no parameters are estimated(Hurley et al., 1997). Peer and subordinate ratings showed acceptable levels of fitwith RMSEA values of .055 and .074, respectively, and TLI values of .97 and .95,respectively. Self-ratings showed poor fit with a RMSEA of .086 and a TLI of .88.

I next estimated a less restricted model, in which the loadings and uniquenesseswere fixed but factor correlations were estimated. If a source showed better fit forthis model than for one with fixed factor correlations, then the factor correlationsfor that source’s ratings differed from those for the supervisors’ ratings. Widaman(1985) suggested that an increase in the TLI of .01 showed a better fit; no particularstandard for change in RMSEA is available.

Peers showed virtually no change in fit, indicating that the factor structure ofsupervisor correlations is reasonable for peers as well. Subordinates did showbetter fit for the model with estimated factor correlations, with a RMSEA of .065(decrease of almost .01) and a TLI of .96 (increase of .01). Examination of the fac-tor correlation estimates showed that they were significantly higher than the super-visor values. Supervisor factor correlations in Table 4 averaged .50, whereas thesubordinate factor correlations averaged .68, suggesting that the subordinates didnot distinguish as well between the different rating factors, resulting in the higherfactor correlations. Finally, self-ratings showed no improvement in fit when factorcorrelations were estimated. The supervisor model is a poor fit for the self-ratings.

EFA for self-ratings. Due to the poorly fitting CFA model, I conductedEFAs of the self-ratings. Results were surprisingly similar to those for supervisors.A five-factor solution was found to be the best, and the pattern of loadings lookedquite similar to the supervisor loadings. The major differences appeared to be thatmost self-loadings and self-factor correlations were lower than the correspondingsupervisor values.

Interpretation of the five-factor solution. The loadings in Table 3 show aninterpretable pattern. Factor labels and definitions appear in Table 5. The first fac-tor is clearly an Interpersonal Effectiveness factor and corresponds reasonably well

MANAGERIAL PERFORMANCE DEVELOPMENT 31

32 CONWAY

TA

BLE

3F

acto

rLo

adin

gsan

dU

niqu

enes

ses

for

Sup

ervi

sor

Ben

chm

arks

Rat

ings

Fa

cto

rL

oa

din

gs

Dim

en

sio

nIn

terp

ers

on

al

Will

ing

ne

ssT

ea

mw

ork

/Pe

rso

na

lA

dju

stm

en

tA

da

pta

bili

tyL

ea

de

rsh

ipU

niq

ue

ne

sse

s

1.R

esou

rcef

ulne

ss.0

5.1

5.1

0.7

5.0

2.1

22.

Doi

ngW

hate

ver

itT

akes

.14

.59

–.13

.35

.11

.13

3.B

eing

aQ

uick

Stu

dy–.

16.0

8–.

19.8

0.0

7.4

54.

Dec

isiv

enes

s.0

1.8

4–.

08–.

03–.

06.3

95.

Lead

ing

Em

ploy

ees

.17

–.03

.35

.12

.50

.11

6.S

ettin

ga

Dev

elop

men

talC

limat

e.1

5.1

2–.

01.1

2.6

7.1

87.

Con

fron

ting

Pro

blem

Em

ploy

ees

–.10

.56

.39

–.06

.06

.47

8.W

ork

Tea

mO

rient

atio

n–.

03.0

2.5

9–.

15.3

1.4

89.

Hiri

ngT

alen

ted

Sta

ff–.

03.2

4.1

7.0

9.3

4.5

710

.B

uild

ing/

Men

ding

Rel

atio

nshi

ps.6

4–.

07.2

1.2

9–.

07.1

711

.C

ompa

ssio

nan

dS

ensi

tivity

.59

–.10

.04

–.10

.38

.38

12.

Str

aigh

tforw

ardn

ess/

Com

posu

re.2

0–.

08.3

2.2

8–.

08.6

513

.B

alan

ce:P

erso

nalL

ifean

dW

ork

.12

–.02

.35

–.10

–.03

.87

14.

Sel

f-A

war

enes

s.4

0.0

9.2

2.2

1–.

03.4

615

.P

uttin

gP

eopl

eat

Eas

e.9

4.0

5–.

08–.

19.0

2.3

216

.A

ctin

gW

ithF

lexi

bilit

y.4

7.1

8.2

4.1

2.1

0.2

4

No

te.

All

load

ings

grea

ter

than

.30

are

italic

ized

.

to McCauley et al.’s (1989) Respect for Self and Others. This factor appears to re-flect the desire to get along, and was defined by Putting People at Ease with a load-ing of .94. Other high-loading scales (loadings in parentheses) were Building andMending Relationships (.64), Compassion and Sensitivity (.59), Acting With Flex-ibility (.47; e.g., “is tough and at the same time compassionate”), andSelf-Awareness (.40; e.g., “does an honest self-assessment”). Scales with slightlynegative loadings were Being a Quick Study (–.16) and Confronting Problem Em-ployees (–.10). The Interpersonal Effectiveness factor appears to tap effectivenessat getting along and making others feel comfortable and supported, with some fo-cus on using this behavior to achieve business goals (e.g., Building and MendingRelationships; see the definition in Table 1). The small negative loading for Con-fronting Problem Employees indicates that the interpersonal skill measured by thisfactor does not extend to difficult situations, however.

The self-oriented scales tended not to load on this factor; the exception wasSelf-Awareness. McCauley et al. (1989) suggested that Straightforwardness andComposure and Balance Between Personal Life and Work would also be a part ofthis construct, but they were not. The main focus of the factor is on relationshipswith others, with only a small focus on the self.

The second factor taps a Willingness to Handle Difficult Situations. The Will-ingness factor represents the volitional part of McCauley et al.’s (1989) MeetingJob Challenges, the part dealing with the drive and attitudes needed to cope withjob demands (the other part of McCauley et al.’s construct was represented by thisstudy’s fourth factor). This behavior should be driven by a desire to get ahead. De-cisiveness had the highest loading at .84. This scale taps a willingness to make de-cisions without hesitating. Doing Whatever it Takes (loading of .59) includesperseverance, showing tenacity in difficult situations, taking responsibility, andseizing opportunities. Confronting Problem Employees (loading of .56) includes,for example, being quick to confront people presenting problems, and being decisivewith tough decisions such as layoffs. This factor taps the willingness to make diffi-cult decisions and confront difficult situations without procrastination.

The third factor represents a sense of Teamwork and Personal Adjustment. Thehighest loading scale was Work-Team Orientation (.59). The items comprising

MANAGERIAL PERFORMANCE DEVELOPMENT 33

TABLE 4Factor Correlations for Supervisor Benchmarks Ratings

Factor 1 2 3 4 5

1. Interpersonal 1.002. Willingness .29 1.003. Teamwork/personal adjustment .59 .29 1.004. Adaptability .56 .62 .54 1.005. Leadership .54 .50 .57 .52 1.00

this scale show that the focus is on the manager working through the team to ac-complish goals rather than as an individual. Four scales had loadings ranging from.39 to .32: Confronting Problem Employees (.39), Leading Employees (.35), Bal-ance Between Personal Life and Work (.35), and Straightforwardness and Compo-sure (.32). Doing Whatever it Takes and Being a Quick Study had lower, negativeloadings (–.13 and –.19, respectively).

34 CONWAY

TABLE 5Factor Labels, Definitions, and Corresponding Borman and Brush (1993) Megadimensions

Factor 1: Interpersonal Effectiveness.Showing good social skill (e.g., tact, compassion, flexibility),making others feel comfortable, and influencing others.

Borman and Brush Megadimensions (numbers are from Borman and Brush’s numbering system)8. Maintaining good working relationships17. Selling/influencing

Factor 2:Willingness to Handle Difficult Situations.Showing courage and perseverance, and havingthe confidence and willingness to make decisions, confront problem employees, take charge, anddo whatever else is necessary in challenging situations.

Borman and Brush Megadimensions12. Persisting to reach goals13. Handling crises and stress

Factor 3:Teamwork and Personal Adjustment.Having an orientation toward working through theteam, and being well-adjusted (e.g., not obsessed with work; honest and not cynical or moody).

Borman and Brush Megadimensions9. Coordinating subordinates and other resources to get the job done

Factor 4:Adaptability.Showing the ability to learn quickly and apply learning to think strategically,work with executives, make good decisions, and solve problems.

Borman and Brush Megadimensions1. Planning and organizing6. Technical proficiency10. Decision making/problem solving15. Monitoring and controlling resources18. Collecting and interpreting data

Factor 5:Leadership and Development.Hiring competent people and effectively providing themwith the opportunity and motivation to develop skills (e.g., delegating, giving decision-makingresponsibility to subordinates).

Borman and Brush Megadimensions2. Guiding, directing, and motivating subordinates and providing feedback3. Training, coaching, and developing subordinates11. Staffing16. Delegating

Borman and Brush Megadimensions Not Corresponding to Benchmarks Factors4. Communicating effectively and keeping others informed5. Representing the organization to customers and the public7. Administration and paperwork14. Organizational commitment

High loadings for Work-Team Orientation and Leading Employees (e.g., dele-gating, providing opportunities for decision making) indicate a clear orientation to-ward the team. Confronting Problem Employees is also fairly consistent with thisorientation. Doing Whatever it Takes and Being a Quick Study tend to focus on themanager as an individual, so the small negative loadings are understandable. TheBalanceandStraightforwardnessscales thatMcCauleyetal. (1989)associatedwiththe Interpersonal Effectiveness factor loaded moderately highly on this factor in-stead. These scales seem to indicate personal adjustment. The Work-Team Orienta-tion scale can also be interpreted in terms of adjustment. It probably takes a certaindegree of self-control and perspective for a manager to allow a team to handleimportant tasks, rather than trying to do things himself or herself. So, better adjustedmanagers will probably find it easier to have a team orientation.

The fourth factor represents half of McCauley et al.’s (1989) Meeting JobChallenges construct (the other half was represented by the second factor, Will-ingness to Handle Difficult Situations). The two scales with very high loadingswere Being a Quick Study (.80) and Resourcefulness (.75). Doing Whatever itTakes had a loading of .35. Scales with small negative loadings wereWork-Team Orientation (–.15) and Putting People at Ease (–.19). McCauley etal.’s original label of Adaptability seems appropriate here. The highest loadingscales indicate that a manager scoring high on this factor is one who can assimi-late new information quickly and use it to think strategically, work with execu-tives, solve problems, and make good decisions. The small negative loadingsindicate that Adaptability is an individually focused (rather than team- orother-focused) factor. Although the Willingness factor was clearly volitional,this study’s Adaptability factor was more oriented toward mental ability, an in-tellectual orientation, and creativity as well as the ability to withstand stress(e.g., making decisions under pressure, part of the Resourcefulness scale).

The fifth factor is a Leadership and Development factor, similar to McCauleyet al.’s (1989) Leading People factor. As noted in the Introduction, this factorprobably requires both the desire to get along and the desire to get ahead. Thehighest loadings were for Setting a Developmental Climate (.67) and LeadingEmployees (.50), with moderately high loadings for Compassion and Sensitivity(.38), Hiring Talented Staff (.34), and Work-Team Orientation (.31). AlthoughFactor 3 (Teamwork and Personal Adjustment) focuses more on the orientationto work through the team rather than as an individual, the Leadership and Devel-opment factor focuses more on selecting, challenging, developing, and motivat-ing subordinates. One way to think of this distinction is in terms of the tendencyto do something versus the effectiveness in doing it. Teamwork and PersonalAdjustment is oriented toward the tendency to work through a team, and Leader-ship and Development is oriented toward the effectiveness at working throughthe team (e.g., motivating subordinates to take advantage of developmental op-portunities and achieve goals).

MANAGERIAL PERFORMANCE DEVELOPMENT 35

Personality Correlates

I calculated factor scores using scoring coefficients from SAS (estimated by re-gression; Kim & Mueller, 1978), and then computed correlations between thefactor scores and CPI and MBTI dimensions. This was done both with thesources kept separate and with the scores for each factor averaged across super-visors, peers, and subordinates. I excluded self-ratings for two reasons. First,self-ratings showed a different factor structure so the factor scores would not becomparable. Second, self-ratings showed considerably higher correlations withpersonality measures than did other sources’ ratings. This was probably becausethe personality measures were self-reports and the correlations were inflated byshared method variance.

I corrected the correlations for unreliability in both the performance ratings andthe personality dimensions to get estimates of correlations between constructs. Forpersonality measures I used coefficient alphas reported in Gough (1987) andMyers and McCaulley (1985). For performance factors I used interraterreliabilities. To correct the scores aggregated across sources I used coefficientalphas for each factor, computed by treating each source as an item. The resultingreliabilities ranged from .59 (Interpersonal Effectiveness) to .44 (Leadership andDevelopment). To correct each individual source’s correlations the appropriate re-liability would take into account consistency across ratings by members of thesame source. To calculate these reliabilities I used a somewhat larger database (N= 2,809 managers) of which the current data were a subset. In the larger datasetmany managers had multiple raters for at least one source; using all these raters Icomputed intraclass correlations (ICCs) separately by source for each factor. Su-pervisors had the highest ICCs, all above .40 except Leadership and Development(ICC = .30). Peer and subordinate ICCs were generally in the .30s. For bothsources Leadership and Development had the lowest ICC (.25 for both sources).

Personality correlates of factor scores aggregated across sources.For rating factors aggregated across sources, uncorrected and corrected correla-tions are shown in Table 6. Corrected correlations with absolute values of at least.15 are italicized in Table 6. The Interpersonal Effectiveness factor, representingthe motive to get along, correlated most strongly with CPI Empathy and Tolerance,and MBTI Thinking–Feeling (agreeableness; McCrae & Costa, 1989). These cor-relations were consistent with hypotheses for the “getting along” construct. Thisfactor also correlated with CPI Socialization and Self-Control, classified by Kampand Hough (1988) as measures of adjustment.

The Willingness factor, representing the motive to get ahead, correlated moststrongly with CPI Dominance, Independence, Self-acceptance, and Social Pres-ence. These dimensions were classified by Kamp and Hough (1988) as measuresof potency. Potency measures were hypothesized to reflect the “getting ahead”

36 CONWAY

construct, and these results support that hypothesis. There were negative correla-tions with MBTI Extraversion–Introversion (indicating more extraverted behaviorassociated with higher factor scores) and CPI Self-Control (indicating better ad-justed managers performed more poorly on this dimension).

Teamwork and Personal Adjustment correlated most highly and positively withadjustment-oriented personality constructs including CPI Socialization andSelf-Control. These results are consistent with Howard and Bray’s (1988) findingthat managers’ adjustment correlated negatively with impulsivity. The Teamworkfactor also correlated with CPI Tolerance.

MANAGERIAL PERFORMANCE DEVELOPMENT 37

TABLE 6Personality Correlates of Rating Factor Scores Aggregated Across Sources

Performance Development Factor

Dimension Interpersonal WillingnessTeamwork/Adjustment Adaptability Leadership

CPIDominance –.02 (–.04) .23 (.34) –.03 (–.05) .05 (.09) .07 (.12)Capacity for Status .06 (.10) .08 (.14) –.01 (–.01) .05 (.10) .06 (.12)Sociability .08 (.12) .09 (.14) –.01 (–.02) .01 (.01) .07 (.13)Social Presence .05 (.08) .14 (.22) .00 (.00) .04 (.06) .07 (.14)Self-Acceptance .05 (.10) .16 (.29) –.02 (–.03) .04 (.08) .08 (.17)Independence –.07 (–.10) .19 (.30) –.04 (–.06) .06 (.10) .05 (.08)Empathy .15 (.26) .05 (.09) .06 (.10) .05 (.09) .12 (.23)Responsibility .07 (.11) .03 (.04) .06 (.09) .12 (.21) .05 (.10)Socialization .12 (.18) –.05 (–.08) .14 (.22) .08 (.14) .07 (.12)Self-Control .10 (.15) –.10 (–.15) .14 (.22) .09 (.15) .04 (.07)Good Impression .07 (.10) –.02 (–.03) .07 (.11) .06 (.09) .03 (.06)Communality .02 (.03) –.05 (–.08) .03 (.05) .00 (.00) –.01 (–.02)Well-Being .05 (.07) .06 (.09) .09 (.14) .09 (.15) .06 (.11)Tolerance .11 (.17) .01 (.01) .10 (.16) .13 (.23) .08 (.14)Achievement viaConformance

.02 (.04) .00 (–.01) .07 (.10) .07 (.12) .02 (.04)

Achievement viaIndependence

.07 (.10) .01 (.01) .08 (.13) .12 (.22) .09 (.15)

Intellectual Efficiency .03 (.04) .04 (.06) .08 (.12) .12 (.20) .06 (.11)Psychological-mindedness .01 (.02) .01 (.01) .02 (.03) .10 (.17) .05 (.10)Flexibility .08 (.12) .01 (.01) .00 (.00) .05 (.08) .05 (.10)

MBTIExtraversion–Introversion –.07 (–.09) –.11 (–.17) .03 (.04) .03 (.05) –.06 (–.10)Sensing–Intuition .03 (.04) .10 (.15) –.02 (–.03) .09 (.15) .11 (.18)Thinking–Feeling .20 (.30) –.07 (–.11) .09 (.13) –.02 (–.04) .10 (.17)Judging–Perceiving .09 (.13) .06 (.08) –.03 (–.04) .02 (.04) .04 (.07)

Note. Values in parentheses are correlations corrected for unreliability in both performancedevelopment and personality constructs. Corrected correlations of at least .15 are italicized.

Adaptability correlated with personality constructs indicating an intellectual,creative orientation such as CPI Intellectual Efficiency and Achievement via Inde-pendence (according to Gough, 1987, both of these constructs were associatedwith adjective checklist ratings such as intelligent, insightful, and clear-thinking),and MBTI Sensing–Intuition (this dimension was interpreted by McCrae & Costa,1989, as a measure of openness to experience). This factor also correlated with de-pendability and adjustment-oriented constructs such as CPI Responsibility,Self-Control, Well-being, and Tolerance.

Leadership and Development, as expected, correlated with personality con-structs representing both the motive to get along (CPI Empathy and MBTIThinking–Feeling) and the motive to get ahead (CPI Self-Acceptance, a measureof potency). The predicted correlations with potency constructs were morestrongly confirmed for supervisor ratings and peer ratings separately, without sub-ordinate ratings (e.g., CPI Sociability and Social Presence in addition toSelf-Acceptance correlated with supervisors and peers; see Table 7).

Personality correlates by source. Table 7 shows factor score–personalitycorrelations (corrected for unreliability) separately for supervisors, peers, and sub-

38 CONWAY

TABLE 7Personality Correlates of Rating Factor Scores By Source

(Correlations All Corrected for Unreliability)

Source

Factor Supervisor Peer Subordinate

Interpersonal EffectivenessCPI Empathy .26 .27 .18CPI Socialization .21 .15 .13CPI Self-Control .16 .10 .14CPI Tolerance .21 .13 .13MBTI Thinking-Feeling .31 .27 .23MBTI Judging-Perceiving .15 .14 .06

Willingness to Handle Difficult SituationsCPI Dominance .31 .33 .28CPI Capacity for Status .13 .16 .08CPI Sociability .16 .18 .06CPI Social Presence .23 .25 .13CPI Self-Acceptance .31 .34 .13CPI Independence .29 .29 .22CPI Self-Control –.16 –.17 –.08MBTI Extraversion-Introversion –.18 –.18 –.10MBTI Sensing-Intuition .15 .11 .15

(Continued)

ordinates. Only personality constructs showing a corrected correlation of at least.15 with at least one source are shown. The personality constructs correlating witheach factor tended to be the same across sources. However, supervisors usuallyshowed the highest correlations and subordinates usually showed the lowest. Thelow subordinate correlations were especially noticeable for the Leadership andDevelopment factor.

There were some exceptions to the generally higher supervisor–personality cor-relations. Corrected peer–personality correlations equaled supervisor correlationsfor the Willingness factor. And for each of the five factors, peers showed higher cor-

MANAGERIAL PERFORMANCE DEVELOPMENT 39

TABLE 7(Continued)

Source

Factor Supervisor Peer Subordinate

Teamwork and Personal AdjustmentCPI Empathy .06 .17 .05CPI Socialization .25 .17 .17CPI Self-Control .24 .15 .19CPI Well-Being .18 .07 .12CPI Tolerance .21 .11 .10Adaptability

CPI Responsibility .20 .18 .15CPI Socialization .18 .06 .12CPI Self-Control .16 .10 .13CPI Well-Being .17 .09 .11CPI Tolerance .26 .17 .16CPI Achievement via Independence .20 .17 .18CPI Intellectual Efficiency .16 .17 .17CPI Psychological-Mindedness .18 .12 .14MBTI Sensing-Intuition .15 .08 .15

Leadership and DevelopmentCPI Capacity for Status .17 .10 .05CPI Sociability .15 .16 .05CPI Social Presence .16 .17 .05CPI Self-Acceptance .24 .25 –.01CPI Independence .16 .11 –.02CPI Empathy .24 .27 .13CPI Socialization .16 .06 .10CPI Well-Being .18 .03 .08CPI Tolerance .20 .12 .08CPI Achievement via Independence .20 .11 .11CPI Intellectual Efficiency .15 .07 .08MBTI Sensing-Intuition .19 .18 .13MBTI Thinking-Feeling .16 .17 .12

rected correlations with CPI Empathy than did either of the other sources. This sug-gests that peers are particularly attuned to and affected by empathic behaviors.

DISCUSSION

This study adds to a growing literature on the nature of job performance constructs(e.g., Borman & Brush, 1993; Campbell, McCloy, Oppler, & Sager, 1993) and pro-vides further evidence that when criterion constructs and personality constructs arecarefully matched, a clear, interpretable pattern of relations emerges (Hogan, 1998;Hough, 1998). The performance constructs examined in this study particularly con-cerned the development of managerial performance, rather than performance activ-ities contained in the normal workday (the concern of previous research; e.g.,Borman & Brush, 1993). Factor analyses revealed five managerial performance de-velopment factors, and correlations with personality measures suggest underlyingmotivational determinants.

Integration With Borman and Brush’s (1993) ManagerialPerformance Megadimensions

The developmental perspective can augment and expand our understanding of theconcept of managerial performance in general. To accomplish this it is important toexamine how the constructs identified here fit in with previous literature. I will dothis by interpreting Borman and Brush’s (1993) 18 managerial performancemegadimensions in terms of this study’s five factors.

Table 5 shows relations between the five performance development factors andBorman and Brush’s megadimensions. The Interpersonal Effectiveness factoroverlaps with two of the megadimensions: “Maintaining good working relation-ships” and “Selling/influencing.” For both taxonomies the emphasis is on estab-lishing smooth relationships allowing the manager to gain cooperation andachieve goals.

The Willingness factor also overlaps with two megadimensions: “Persisting toreach goals” and “Handling crises and stress.” The “persisting” megadimensionoverlaps with the “perseverance” aspect of Doing Whatever it Takes. The “han-dling crises” megadimension emphasizes quickly dealing with unexpected situa-tions and tight deadlines. This is similar to Decisiveness and also overlaps withDoing Whatever it Takes. This study’s Willingness factor has a unique focus onConfronting Problem Employees.

The Teamwork factor has only a relatively weak link, to the “Coordinating sub-ordinates and other resources to get the job done” megadimension. Workingthrough a team implies this type of coordination. Teamwork included a substantial

40 CONWAY

adjustment component (e.g., Balance Between Personal Life and Work andStraightforwardness and Composure) that did not overlap with themegadimensions. The lack of overlap is probably due to the fact that adjustmentper se is not an activity. However, the Management Progress Study results(Howard & Bray, 1988) suggested that, relative to poorly adjusted managers,better adjusted managers tended to experience positive changes over time. Thesechanges included increased dominance, achievement motivation, and optimism.They also found a positive, but low, relation between adjustment and organiza-tional advancement.

The Adaptability factor shows some degree of overlap with fivemegadimensions, all of which have to do with acquiring and applyingknowledge. The megadimensions are “Planning and organizing” (e.g., includ-ing “strategic planning; the Resourcefulness scale has a “thinking strategically”component), “Technical proficiency,” “Decision making/problem solving,”“Monitoring and controlling resources” (Resourcefulness includes items regard-ing building systems that are self-monitoring and can be managed by remotecontrol), and “Collecting and interpreting data” (the “problem-solving” compo-nent of Resourcefulness includes gathering information and analyzing situa-tions).

Finally, the Leadership and Development factor overlaps with fourmegadimensions: “Guiding, directing, and motivating subordinates and providingfeedback”; “Training, coaching, and developing subordinates”; “Staffing”; and“Delegating.”

There were four megadimensions that did not fit into one of this study’s factors:“Communicating effectively and keeping others informed,” “Representing the or-ganization to customers and the public,” “Administration and paperwork,” and“Organizational commitment.” McCauley et al. (1989) noted the lack of focus inBenchmarks on administrative and communication skills, and speculated thatthese skills are mastered early and are less important than other skills to manage-rial development. The “Representing” and “Organizational commitment”megadimensions are job-dedication-oriented behaviors that probably reflect ageneral tendency to be positive and conscientious. These tendencies are probablywell established before one attains a managerial position so they do not representcommon developmental experiences.

Each of the five performance development factors identified in this studyoverlapped with at least one of Borman and Brush’s megadimensions. TheTeamwork and Personal Adjustment factor had the smallest amount of overlap,with only one megadimension. This weak link suggests that the content of thisfactor is relatively specific to a developmental focus (especially the “PersonalAdjustment” part of the factor). Some megadimensions did not overlap with anyof this study’s factors, indicating areas more important for day-to-day activitiesthan for performance development.

MANAGERIAL PERFORMANCE DEVELOPMENT 41

One limitation of this discussion is that these results may be somewhat specificto the Benchmarks instrument. However, Benchmarks was carefully developedbased on extensive collection of information from managers in many organiza-tions. This broad sampling should tend to mitigate instrument-specific biases andpromote generalizability.

Why Personality is Related to Managers’ Behavior

Personality correlates generally supported hypotheses. The Interpersonal Effec-tiveness factor correlated with CPI Empathy and MBTI Thinking–Feeling (inter-preted here as agreeableness), providing evidence that this factor reveals the motiveto get along with others. The Willingness factor correlated with several po-tency-related CPI dimensions, suggesting that this factor reveals the motive to getahead. Interestingly, Willingness had a negative correlation with CPI Self-Control.The Leadership and Development factor correlated with empathy and agreeable-ness, and also showed some relations with potency variables. Leadership and De-velopment therefore shows both the motive to get along and the motive to getahead. The other two factors, Adaptability and Teamwork and Personal Adjust-ment, also showed distinct patterns of personality correlates.

Why are responses to personality instruments related to others’ ratings of man-agers? Multisource performance ratings reflect managers’ behaviors (and corre-sponding developmental needs). Hogan and Shelton (1998) suggested that aperson’s behaviors—what interactions a person enters into, and how the interac-tions are handled—are a function of identity, or the person’s hopes, dreams, fears,and self-image. According to this view, a person is motivated to engage in behav-iors consistent with his or her identity but not behaviors that are inconsistent. Forexample, if I see myself as someone who is well-liked by others but who is notcomfortable leading others and making difficult decisions, I will be motivated toengage in “getting along” behaviors such as showing empathy but I will not be mo-tivated to engage in “getting ahead” behaviors such as taking charge. I might there-fore be the kind of manager that Johnson, Schneider, and Oswald (1997) describedas an “amiable underachiever,” scoring high on performance ratings forInterpersonal Effectiveness but low on performance ratings for Willingness toHandle Difficult Situations.

Hogan and Shelton (1998) argued that responses to personality scales are an-other form of behavior that is governed by identity. The same identity that leadsto amiable underachievement would also therefore lead to high scores on per-sonality scales like empathy and low scores on scales like dominance. In otherwords, the personality–job performance relation exists because both personalityinventory responses and job behavior reflect the same underlying identities andmotives.

42 CONWAY

Practical Implications of Personality Correlates:Development and Selection

Personality measures, given a motivational interpretation, reflect why a managerdoes things. Understanding a manger’s personality and motivations can be help-ful in finding remedies for weaknesses. For example, the earlier section on per-sonality–job performance relations suggests that amiable underachievers aretypically motivated to get along but not to get ahead. However, Hogan andShelton (1998) noted that others’ evaluations of our behavior depend on socialskill in addition to motivation. It is possible that rather than lacking the motiva-tion to get ahead, an amiable underachiever is motivated to get ahead as well asto get along, but lacks the social skills necessary for taking charge and forleading others. Such a manager would be one whose idealized self-imageincludes ambition and dominance, but whose attempts to lead forcefully areineffective.

The amiable underachiever’s problem could therefore be due to personality(lack of motivation to get ahead), or due to lack of skill. This distinction isimportant because, although personality is fairly stable, social skills are moreeasily trained (Hogan & Shelton, 1998). Personality measures used in conjunc-tion with multisource feedback ratings can help to identify the source of a man-ager’s weakness. If the “amiably underachieving” manager scores low onpersonality scales indicating ambition and potency, then the source of theunderachievement would appear to be personality-related. If, on the other hand,scores on these personality scales are high, the manager would probably like tobe an achiever, but needs work on social skills to carry out getting-aheadbehaviors effectively.

Another example involves a manager receiving low ratings on Leadershipand Development (shown to correlate with both getting along and getting aheadpersonality constructs). The low ratings might be because of a lack of motiva-tion to get along, a lack of motivation to get ahead, or both. Scores on relevantpersonality constructs can help identify the underlying reason, leading to bettertargeted developmental experiences (e.g., improving interpersonal effectivenessvs. developing a more forceful motivational approach).

Personality measures are also useful for managerial selection. If the personalitymeasures are administered to managerial job candidates, the measures can be usedto predict which candidates will need more or less developmental work in a partic-ular area. For example, candidates scoring high on getting along personality di-mensions such as empathy are likely to need less work in the InterpersonalEffectiveness area than low-scoring candidates. Selecting candidates who scorehigh on relevant personality measures should reduce the need for extensivedevelopmental work.

MANAGERIAL PERFORMANCE DEVELOPMENT 43

Rating Source Differences

Differences between rating sources were found both in factor analysis results and inpersonality correlates. Factor analysis parameter estimates were generally verysimilar. One exception was managers’ self-ratings. Self-loadings and self-factorcorrelations were lower than those for other sources. This result is consistent withmeta-analyses showing self-ratings differed from those of other sources (Conway& Huffcutt, 1997; Harris & Schaubroeck, 1988) and adds to the literature showingthat workers see their own performance quite differently than it is seen by others.Another exception to the consistent results across sources involved subordinates,who showed higher factor correlations than other sources. This finding suggeststhat subordinates did not distinguish between dimensions as well as supervisorsand peers.

Correlations with personality measures were generally strongest for supervi-sors and weakest for subordinates. The low subordinate correlations were espe-cially evident for Leadership and Development. This is interesting because thisdimension is very subordinate-focused and should be more important to subordi-nates than to other sources.

Subordinate ratings are difficult to interpret based on these results. No person-ality scales correlated most highly with subordinates. It is possible that subordinateratings depend more on situational factors than do supervisor or peer ratings. Anumber of leadership theories have proposed that the most appropriate type ofleader behavior depends on the type of situation (e.g., Fielder & Garcia, 1987).Subordinates are probably better aware than supervisors or peers of the situation inwhich the manager’s behavior occurs, especially for Leadership and Develop-ment-related behavior. Subordinate ratings may better reflect the appropriatenessof the manager’s behavior, given the situation. For example, peers and supervisorsmay rate based on the general tendency to provide challenging assignments to em-ployees (a “getting ahead” behavior). Subordinates may be more sensitive to theappropriateness of the manager’s behavior, and rate not just based on the man-ager’s general tendency to try to get ahead, but based on how appropriate it is in thesituation at hand. Future research should explore the extent to which differences insources’ ratings reflect different, valid perspectives on performance as opposed tosource-specific error variance.

ACKNOWLEDGMENTS

This article was presented at the 1999 Annual Meeting of the Eastern PsychologicalAssociation.

Thanks to the Center for Creative Leadership for providing the data for this pro-ject, and to Joyce Hogan, Margaret A. McManus, and two anonymous reviewersfor their comments on an earlier version of this article.

44 CONWAY

REFERENCES

Borman, W. C., & Brush, D. H. (1993). More progress toward a taxonomy of managerial performancerequirements.Human Performance, 6,1–21.

Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fit.Sociological Methods &Research, 21,230–258.

Campbell, J. P., McCloy, R. A., Oppler, S. H., & Sager, C. E. (1993). A theory of performance. In N.Schmitt & W. C. Borman (Eds.),Personnel selection in organizations(pp. 35–70). San Francisco:Jossey-Bass.

Center for Creative Leadership. (1995).Benchmarks developmental learning guide.Greensboro, NC:Author.

Conway, J. M., & Huffcutt, A. I. (1997). Psychometric properties of multi-source performance ratings:A meta-analysis of subordinate, supervisor, peer, and self-ratings.Human Performance, 10,331–360.

Fiedler, F. E., & Garcia, J. E. (1987).New approaches to effective leadership: Cognitive resources andorganizational performance.New York: Wiley.

Ford, J. K., MacCallum, R. C., & Tait, M. (1986). The application of exploratory factor analysis in ap-plied psychology: A critical review and analysis.Personnel Psychology, 39,291–314.

Gough, H. G. (1987).CPI administrator’s guide.Palo Alto, CA: Consulting Psychologists Press.Harris, M. M., & Schaubroeck, J. (1988). A meta-analysis of self–supervisor, self–peer, and

peer–supervisor ratings.Personnel Psychology, 41,43–62.Hogan, J. (1998). Introduction: Personality and job performance.Human Performance, 11,125–127.Hogan, J., Rybicki, S. L., Motowidlo, S. J., & Borman, W. C. (1998). Relations between contextual per-

formance, personality, and organizational advancement.Human Performance, 11,189–207.Hogan, R. T. (1991). Personality and personality measurement. In M. D. Dunnette & L. M. Hough

(Eds.),Handbook of industrial and organizational psychology(2nd ed., pp. 873–919). Palo Alto,CA: Consulting Psychologists Press.

Hogan, R., & Shelton, D. (1998). A socioanalytic perspective on job performance.Human Perfor-mance, 11,129–144.

Hough, L. M. (1992). The “Big Five” personality variables–construct confusion: Description versusprediction.Human Performance, 5,139–155.

Hough, L. M. (1998, April). Job performance models & personality taxonomies. In J. C. Hogan (Chair),Meaning and models of the personality–job performance relation.Symposium conducted at the13th Annual Conference of the Society for Industrial and Organizational Psychology, Dallas, TX.

Howard, A., & Bray, D. W. (1988).Managerial lives in transition: Advancing age and changing times.New York: Guilford.

Hurley, A. E., Scandura, T. A., Schriesheim, C. A., Brannick, M. T., Seers, A., Vandenberg, R. J., &Williams, L. J. (1997). Exploratory and confirmatory factor analysis: Guidelines, issues, and alter-natives.Journal of Organizational Behavior, 18,667–683.

Johnson, J. W., Schneider, R. J., & Oswald, F. L. (1997). Toward a taxonomy of managerial perfor-mance profiles.Human Performance, 10,227–250.

Kamp, J. D., & Hough. L. M. (1988). Utility of temperament for predicting job performance. In L. M.Hough (Ed.),Utility of temperament, biodata, and interest assessment for predicting job perfor-mance(ARI Research Note 88–02; pp. 1–90). Alexandria, VA: U.S. Army Research Institute for theBehavioral and Social Sciences.

Kim, J.-O., & Mueller, C. W. (1978).Factor analysis: Statistical methods and practical issues.BeverlyHills: Sage.

Lindsey, E. H., Homes, V., & McCall, M. W., Jr. (1987).Key events in executives’ lives(Tech. Rep. No.32). Greensboro, NC: Center for Creative Leadership.

MANAGERIAL PERFORMANCE DEVELOPMENT 45

MacCallum, R. (1998). Commentary on quantitative methods in I–O research.The Industrial–Organi-zational Psychologist, 35,19–30.

Maurer, T. J., Raju, N. S., & Collins, W. C. (1998). Peer and subordinate performance appraisal mea-surement equivalence.Journal of Applied Psychology, 83,693–702.

McCall, M. W., Jr., Lombardo, M. M., & Morrison, A. M. (1988).The lessons of experience: How suc-cessful executives develop on the job.Lexington, MA: Lexington.

McCauley, C. D., Lombardo, M. M., & Usher, C. J. (1989). Diagnosing management developmentneeds: An instrument based on how managers develop.Journal of Management, 15,389–403.

McCrae, R. R., & Costa, P. T., Jr. (1989). Reinterpreting the Myers–Briggs Type Indicator from the per-spective of the five-factor model of personality.Journal of Personality, 57,17–40.

McDonald, R. P. (1985).Factor analysis and related methods.Hillsdale, NJ: Lawrence Erlbaum Asso-ciates, Inc.

Murphy, K. R., & Cleveland, J. N. (1995).Understanding performance appraisal: Social, organiza-tional, and goal-based perspectives.Thousand Oaks, CA: Sage.

Myers, I. B., & McCaulley, M. H. (1985).Manual: A guide to the development and use of theMyers–Briggs Type Indicator.Palo Alto, CA: Consulting Psychologists Press.

SAS Institute, Inc. (1989).SAS/STAT user’s guide, Version 6(4th ed., Vol. 1). Cary, NC: Author.Tornow, W. W., & Pinto, P. R. (1976). The development of a managerial job taxonomy: A system for de-

scribing, classifying, and evaluating executive positions.Journal of Applied Psychology, 61,410–418.

Viswesvaran, C., Ones, D. S., & Schmidt, F. L. (1996). Comparative analysis of the reliability of job per-formance ratings.Journal of Applied Psychology, 81,557–574.

Widaman, K. F. (1985). Hierarchically nested covariance structure models for multitrait–multimethoddata.Applied Psychological Measurement, 9,1–26.

46 CONWAY