tk_wagner

12
Journal of Personality and Social Psychology 1987. Vol 52. No. 6. 1:36-1247 Tacit Knowledge in Everyday Intelligent Behavior Richard K . Wagner Florida State University Two experiments were conducted to examine the scope and structure of tacit knowledge. In Experi- ment 1, three groups, whose members differed in level of professional advancement in the field of academic psychology, were presented with simulated work-related situations designed to measure their tacit knowledge. Between-group differences in tacit knowledge were found as a function of level of professional advancement, and strong within-group relations were between tacit knowledge and external criteria such as rate of citation. Th e scope of tacit knowledge was found to include (a) knowledge useful in managing oneself, others, an d one's tasks, (b) knowledge applicable to both short-term and long-term contexts, and (c) knowledge of ideal quality as well as practical reality. The results of testing four classes of alternative models of the structure of tacit knowledge supported a model characterized by a general factor, similar in form to Spearman's # for academic tasks. Experi- ment 2 replicated and extended these results to the domain of busines s management. Many professionals say that much, ifnot most, ofthe learning that matters to their careers took place after they completed their formal schooling. Furthermore, it is not uncommon for people who are highly successful in their occupations to have unremarkable academic records and, conversely, for people with histories of distinguished performance in school to be only moderately successful in their occupations. The lack o f a strong relation between actual occupational per- formance a n d scores o n either IQ tests o r employment tests sug- gests that these observa tions may not be completely unfounded. Although IQ test scores correlate moderately (.4-.7) with vari- ous measures of school achievement and performance in occu- pational tr aining programs, typic al correlations between IQ tests or employment tests and actual performance fall at the .2 level (Ghiselli, 1966; Wigdor & Garner, 1982). C orrelati ons of this magnitude indicate that performance on such tests ac- counts for only small amount s of variance (about 4%) in occ u- pational performance. Neisser (19761 has argued that the tasks found on IQ tests and in school settings measure academic intelligence, which is but one part of intelligence broadly defined. In contrast to aca- demic intelligence, Neisser denned "intelligent performance in natural settings" a s "res pondin g appropriately in terms o f one's long-range an d short-range goals, given the actual facts of the this research wa s supported in part by Contract MDA90385K.0305 from the Army Research Institute. This article is based on a doctoral dissertation submitted to the fac- ulty of Yale University. I gratefully acknowledge the contributions of my dissertation advisor. Robert J. Sternberg, and the dissertation com- mittee members, Wendell Gamer, Edmond Gordon, Richard Hack- man, and Roger Weissberg. I also acknowledge the contributions of members of the Sternberg research group. Correspondence concerning this article should be addressed to Richard K. Wagner, Department of Psychology, Florida State Univer- sity. Tallahassee, Florida 32306. situation as one discovers them" ( p . 137; s ee also Charlesworth. 1976; Wagner, 1986). Researchers have propos ed a number of traditional ap- proaches to measuring competencies required for success in real-world pursuits (see Sternb erg & Wagner. 1986). Recently, cogn itive psycho logists have taken a knowledge-based approach in their study of how "experts" and "novices" differ in their performance on tasks relevant to their domain of exper tise. (Se e Chi, Glaser, & Rees, 1982, for a comprehensive review of this literature.) Although the knowledge-based appr oach has been applied mostly to academic-intelligence-type tasks, Wagner and Sternberg (1985) used this approach in their investigation of practical tasks faced by business managers and academic psy- chologists. They proposed that a hallmark of practical intelli- gence is the acquisition and use of tad! knowledge —practical know-how that usually is not openly expressed or stated and which must be acquired in the absence of direct instruction. (SeePolanyi, 1976, and Schon, 1983 , for rel ated conceptions of practical know-how.) If the acquisition of tacit knowledge is indeed a hallmark of practical intelligence, what is the scope and structure of such knowledge? Scope o f Tacit Knowledge The scope of tacit knowledge refers to the range of situations to which tacit knowledge may be appl ied. To examine the scope of tacit knowledge I categorized it on the basis of (a) the content of a situation, that is, whether it primarily involves managing oneself, managing others, or managing one's tasks; (b) the con- text of a situation, that is, whether it is local (short range, self- contained) o r global (long range, related to the "big pi cture") in nature; and (c) orientation, that is, whether one's focus is ideal- istic or pragmatic. Three Kinds o f Content Tac it knowledge about managing self refers to knowledge about self-motivational and self-organizational aspects of per- 1236

Upload: renata-andra

Post on 05-Apr-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 1/12

Journal of Personality and Social Psychology

1987. Vol 52. No. 6. 1:36-1247

Tacit Knowledge in Everyday Intelligent Behavior

Richard K. Wagner

Florida State University

Twoexperimentswere conducted to examine the scope and structure of tacit knowledge. In Experi-

ment 1, three groups, whose members differed in level of professional advancement in the f ie ld of

academic psychology, were presented with simulated work-related situations designed to measure

their tacit knowledge. Between-groupdifferences in tacit knowledge were found as a fu n c t io n of level

of professional advancement, and strong within-group relations were found between tacit knowledge

and external criteria such as rate of citation. The scope of tacit knowledge was f o u n d to include (a)

knowledge useful in managing oneself, others, and one's tasks, (b) knowledge applicable to both

short-term and long-term contexts, and (c) knowledge of ideal quality aswel l aspractical reality. The

results of testing four classes of alternative models of the structure of tacit knowledge supported a

model characterized by a general factor, similar in form to Spearman's#for academic tasks. Experi-

ment 2 replicated and extended these results to the domain of business management.

Many professionals saythatmuch,if not most,of the learning

that matters to their careers took place after they completed

their formal schooling. Furthermore, it is not uncommon for

people who are highly successful in their occupations to have

unremarkable academic records and, conversely, for people

with historiesof distinguished performance in school to be only

moderately successful in their occupations.

The lack of astrongrelation between actual occupational per-

formance and scores on either IQtestsor employment tests sug-

gests that these observations may not be completely unfounded.

Although IQ test scores correlate moderately (.4-.7) with vari-

ous measures of school achievement and performance in occu-

pational training programs, typical correlations between IQ

tests or employment tests and actual performance fall at the .2

level (Ghiselli, 1966; Wigdor & Garner, 1982). Correlations of

this magnitude indicate that performance on such tests ac-

counts for only small amounts of variance (about 4%) in occu-

pational performance.

Neisser (19761has argued that the tasks found on IQtestsand

in school settings measure academic intelligence, which is but

one part of intelligence broadly defined. In contrast to aca-

demic intelligence, Neisser denned "intelligent performance in

natural settings" as"responding appropriately interms ofone's

long-range and short-range goals, given the actual facts of the

this research was supported in part by Contract MDA90385K.0305

from the Army Research Institute.

This article is based on a doctoral dissertation submitted to thefac-

ulty of Yale University. I gratefully acknowledge the contributions of

my dissertation advisor. Robert J. Sternberg, and the dissertationcom-

mittee members, Wendell Gamer, Edmond Gordon, Richard Hack-

man, and Roger Weissberg. I also acknowledge the contributions of

members of the Sternberg research group.

Correspondence concerning this article should be addressed to

Richard K. Wagner, Department of Psychology, Florida State Univer-

sity. Tallahassee, Florida 32306.

situation as onediscoversthem"(p. 137; seealso Charlesworth.

1976; Wagner, 1986).

Researchers have proposed a number of traditional ap-

proaches to measuring competencies required for success in

real-world pursuits (see Sternberg & Wagner. 1986). Recently,

cognitive psychologists have taken a knowledge-based approach

in their study of how "experts" and "novices" differ in their

performance on tasks relevant to their domain of expertise. (See

Chi, Glaser, & Rees, 1982, for a comprehensive review of this

literature.) Although the knowledge-based approach has been

applied mostly to academic-intelligence-type tasks, Wagner and

Sternberg (1985) used this approach in their investigation of

practical tasks faced by business managers and academic psy-chologists. They proposed that a hallmark of practical intelli-

gence is the acquisition and use of tad! knowledge—practical

know-how that usually is not openly expressed or stated and

which must be acquired in the absence of direct instruction.

(SeePolanyi, 1976, and Schon, 1983, for related conceptions of

practical know-how.)

If the acquisition of tacit knowledge is indeed a hallmark of

practical intelligence, what is the scope and structure of such

knowledge?

Scope of Tacit Knowledge

The scope of tacit knowledge refers to the range of situations

to which tacit knowledge may be applied. Toexamine the scope

of tacit knowledge I categorized it on the basisof (a) the content

of a situation, that is, whether it primarily involves managing

oneself, managing others, or managing one's tasks; (b) the con-

text of a situation, that is, whether it is local (short range, self-

contained) orglobal (long range, related to the "big picture") in

nature; and (c) orientation, that is, whether one's focus is ideal-

istic or pragmatic.

Three Kinds of Content

Tacit knowledge about managing self refers to knowledge

about self-motivational and self-organizational aspects of per-

1236

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 2/12

TACIT KNOWLEDGE 1237

formance in work-related situations. An example of tacit

knowledge about managing self is knowinghowbest to over-

come procrastination. Tacit knowledge about managingtasks

refers to knowledge of how to do specific work-related tasks

well. An example of tacit knowledge about managing tasks is

knowing the value of beginning a manuscript by telling the

reader whatmajor pointsone intends to make. Tacit knowledge

about managing others refers to knowledge about managing

one's subordinates and one's interactions with one's peers. An

example of tacit knowledge about managing others isknowing

how to reward subordinates so as to maximizeboth productiv-

ity and job satisfaction.

Although some situations primarily involve only one of the

three kindsoftacit knowledge (e.g., the problem ofprocrastina-

tion), most require each of the three kinds of tacitknowledge

to lesser or greater degrees. For example, chairing a grant review

study section is a task that requires considerable tacit knowl-

edgeabout managing oneself, others, and one'stasks.

Two ContextsA local context refers to a focus on the short-term accom-

plishment of the specific task at hand. No consideration isgiven

to one's reputation, career goals, or to the "big picture." A

global context refers to a focus on long-range objectives and on

how the present situation fits into the bigger picture.

Real-world accomplishment requires practical knowledge

that can be applied in both local and global contexts. It is not

enough to be knowledgeable about selecting tasks that will pay

off in the long term (global context) if one lacks the knowledge

about how to do well the tasks one has selected (local context).

Conversely, beingknowledgeableabout specific tasks (local con-

text) but not about howthey relate to one another (global con-

text) can result in doing verywell at what may turn out, in thelong run, to be the wrong tasks, or it can result in missing a

connection between a problem faced today and one faced last

month and thereby failing todiscover their root cause.

Two Orientations

The comments of respondents in earlier work suggested that

therearetimeswhenaperson's overriding orientation in agiven

situation is idealistic; that is, it focuses on howgood a solution

isideally. There areother times when aperson's overridingori-

entation ispragmatic; that is, it focuses on howworkable a po-

tential solutionis.

Ideal in the present context refers to a judgment about thequalityofsome courseofaction without regard to howpractical

or impractical it might be. For example, the strategy ofconsult-

ingwith eachofone's subordinates individuallyon all decisions

that mightaffect them isperhaps better ideally than it is inactu-

ality.

Effective performance requires knowledge relevant to both

orientations: Liabilities are associated with being overly prag-

matic aswell asoverly idealistic. Consider the task ofmaintain-

ing morale in the face ofpolicy changes that employees inyour

division of the companywill find threatening. Writing a memo

to these employees stating that you expect their morale to re-

ORIENTATION I D E A L I S T I C

\

» T i r

\\

\\\̂

CONTENT

T A S K S

L O CAL G L O B AL

C O N T E X T

Figure I . T he tacit knowledge framework.

main positive is a practical idea, but one that happens to be not

very good. Conversely, taking one's employees out to dinner

individually to convince them that the policy changeswill be to

everyone's benefit in the long run may be abetter idea than the

memojust described, but a highly impractical one if there are

a large number of employees in one's division.

Combining Contents, Contexts, and O rientation in aTacit Knowledge Framework

The framework that guided evaluation of the scope and

structure of tacit knowledge crosses content, context, and ori-

entation, yielding the 12 possible combinations illustrated in

Figure 1.

Structure of Tacit Knowledge

Consider the structure of traditional academic abilities. In-

vestigators generally agree that the structure is hierarchical,

with at least three levels. At the top is gor a general factor that

represents the positive manifold found among correlations for

almost all cognitive tasks (Jensen, in press; Spearman, 1927).

Positive manifold refers to the fact that the intercorrelations

among most cognitive tasks are positive. At the next level are

group factors, examples ofwhich include fluid and crystallized

intelligence (Cattell, 1963, 1971) and Thurstone's (1938) pri-

mary mental abilities of verbal comprehension, verbal fluency,

number, space, associative memory, perceptual speed, andgen-

eral reasoning. At the lowest level are task-specific factors.

These factors explain the portion of a task's variance that is

unique to the task, and thus they are of no value inaccounting

for shared variance among tasks.

In this study, I examine four classes of models of the struc-

ture of tacit knowledge. Thesemodels, which derive from anal-

yses of the structure of academic abilities, are presented in

Figure2.

Two-Factor Models

These models get their name from the fact that task perfor-

mance isexplainedin terms of ageneral factor and a task-spe-

cific factor for each of the tasks used. The specific factors are

unimportant because theydo not account for any shared vari-

ance. The interpretation of two-factor models when they are

applied to the tacit knowledge framework is that individual

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 3/12

1238 R I C H A R D K. WAGNE R

differences in tacit knowledgewill be general to each content,

context, and orientation.

Hierarchical Models

Hierarchical models share with the two-factor models the

idea that people who are good at one thing tend to be good at

everything, but added is the notion of group factors. Agroupfactor is a factor that accounts for shared variance among some

but not all tasks. In terms of the tacit knowledge framework,

shared variance among tacit knowledge scores would be ex-

plained in terms of a general factor, as before, aswell as in terms

ofgroup factors that might correspond to a local versus a global

context or to the contents of managing self, others, and tasks.

Primary Ability Models

According toprimary ability models, shared variance among

tasks is explained by a set of independent group factors. In

terms of the tacit knowledge framework, primary ability

models posit that people may be knowledgeableabout manag-

ing others, for instance, but not about managing themselves ortheir tasks.

Anarchy Model

This model posits that there is nocommon variance orgener-

ality in tacit knowledge. How knowledgeable one is depends en-

tirely on one's specific task. In terms of the tacit knowledge

framework, there would be no relations between performance

across work-related situations. This model is plausible if it

should turn out that what one leams from one's experience is

specific to the situation inwhich it is learned.

Overview of the Present Experiments

I conductedtwoexperiments, oneeach in thedomainsofaca-

demic psychology and business management. In Experiment 1,

a tacit knowledge measure for the domain ofacademic psychol-

ogy was administered to three groups of respondents whose

membersdiffered inamounts offormal trainingand experience

in the field of academic psychology. Tacit knowledge was com-

pared across groups, and the within-group relations between

tacit knowledge and career performance were examined. I as-

sessed the scope of tacit knowledge in terms of the tacit knowl-

edge framework and tested alternative models of the structure

of tacit knowledge. In Experiment 2, which served to extend the

results of Experiment 1 to a second domain, a tacitknowledge

measure for the domain of business management was adminis-tered to three groups of individuals whose members differed in

amounts of formal trainingandexperience in the professionof

business management.

Experiment 1

Method

Subjects

There were three groups of subjects, totaling212 individuals in all,

whose membersdiffered in amounts of formal training an d experience

in the field of academic psychology.

The faculty group consisted of91 faculty membersin 26 departments

of psychology. This group included 5 1 members of 11 departments

highly rated in terms of the scholarly quality of the departmental faculty

(Jones, Lindzey, &Coggeshall, 1982). The mean ratingof thescholarly

quality of the faculty in these 11 departments was 4.0 on a 5-point scale

(0 = low, 5 = high). Also included in the faculty group were39members

of 15 departments less highly rated in terms of the scholarlyquali ty of

I . T W O - F A C T O R M O D E L S ( S P E A RM A N, T H O M S O N )

SL OL TL SG OG TG

t T t t T t

II . H I E R A R C H I C A L M O D E L S ( C A T T E L L , V E R N O N )

SL OL TL SG OG TG

T t t T T 1

SL OL TL SO OG TG

TTTTTTI I I . P R I M A R Y A B I L I T Y M O D E L S ( T H U R S T O N E )

s—s

SL OL TL SG OG TG

t 1 t t 1 tIV . A N A R C H Y M O D E L

SL OL TL SG OG TG

t t t t t TFigure 2.Fourclassesofmodelsofthe structureoftacit knowledge(S

 

self; O = others; T = tasks; L = local;G = global; g = general factor).

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 4/12

TACIT KNOWLEDGE 1239

departmental faculty. Their departments received a mean rating of 1.0.

One person who did not choose to indicate his or her departmental

affiliation was also included in the faculty group. The sample contained

36 full professors, 33associate professors, and 22assistant professors.

The psychology graduate student group consistedof 61 students from

the same departments that were sampled to obtain the faculty group.

This group contained 36 graduate students from departments highly

rated in terms of the scholarly quality of the faculty and 25 graduate

students from departments less highly rated in scholarly quality of the

faculty. The sampleconsisted of I5 first-year students, 15 second-year

students, 16third-year students, 6 fourth-year students, and 9 fifth-year

and beyond students. The undergraduate group contained 60 Yale un-

dergraduates whowere enrolled in an introductory psychology classand

received course credit for their participation. The undergraduates had

assorted majors, and many had not yet selected their major area of

study.

Materials

The tacit knowledge measure for academic psychology consisted of

12work-related situations, each of which wasassociatedwith from 9 to

11 response items.Of the 12work-related situations, 4were constructed

to sample each of the three contents oftacit knowledge (managingself,

tasks, and others). Half of the 12 work-related situations were con-

structed to sample tacit knowledgewith a local context; the other half,

tacit knowledgewith a global context. Thus each work-related situation

was constructed to sample one of three contents of tacit knowledge

(managingself, tasks,or others) in one of two contexts (local or global).

There were 37, 39, and 40 itemson the managing self, tasks,and others

subscales, respectively, and 59 and 57 on items the local and global con-

text subscales, respectively, for a total of 116 items. An example of a

work-related situation sampling tacit knowledge about managing self

with aglobal context follows.1

It is your second year as an assistant professor in a prestigious psy-

chology department. This past year you published two unrelated

empirical articles in established journals. Youdon't, however, be-

lieve there is yet a research area that can be identified as your own.

You believeyourself to be about asproductive asothers. The feed-

back about your first yearof teaching has been generally good.You

have yet to serve on a university committee. There is one graduate

student who has chosen to work with you. Youhave no external

source offunding, nor have you applied for funding.

Your goals are to become one of the top people in your area of

the field and to gettenure inyour department.Youbelieve yourself

to be a hard workerbut find that you do not have enough time to

get the important things done. You believe that you have not given

enough thought to the relative importance of the tasks you find

yourself engaged in and therefore are developing an agenda of

things to do in the next 2 months that will increase the chances of

success in your career.

The following is a list of things you are considering doing in the

next 2months. Youobviously cannot dothem all. Rate the impor-

tance of each by its priority as a means of reaching yourgoal:

Actual Ideal

1. Improve the qualityof your teaching.

2. Writeagrant proposal.. . .

9. Begin several short-term research projects, each

of which may lead to an empirical article.

Subjects read a given work-related situation and then rated each re-

sponse alternative on a 7-point scale by either its quality (1 = extremely

bad, 4 = neither good norbad, and 7 = extremely good} or its impor-

tance (1 = extremely unimportant, 4 = somewhat important, and 7 =

extremely important), depending on the particular response item. Sub-

jects rated both the actual and ideal quality or importance of each re-

sponse item. In making the actual rating, subjects were told, "The actual

rating asks you to rate how important or how good the response alterna-

tiveactually isgiventhe realitiesof the academicworld as you know it."

In making the ideal rating, subjects were told, "The ideal rating asks

you to rate how important or how good the response alternative should

be; in other words, howimportant it would be in one's ideal academic

world." The actual and ideal ratings served to measure pragmatic and

idealistic orientations, respectively.

Design

Independent variables were the (a) three contents (managing self,

tasks, and others), (b) two contexts (local and global), and (c) two orien-

tations (pragmatic and idealistic) of tacit knowledge.Content, context,

and orientation were "crossed"with one another. The dependent vari-

ables consistedof the setsof actual and ideal ratings assigned to response

items.

Criterion reference measures for the faculty group included (a)

whether a subject's department was rated high in scholarly quality of

departmental faculty (Jonesetal., 1982), (b)numberofcitations for the

years 1982 and 1983 reported in the Social Sciences Citation Index

(1982,1983), and (c) numberofpublications in the years 1982 and 1983

reported in Psychological Abstracts(1982, 1983). Criterion reference

measures for the graduate student group included several measures of

research productivity and the rated scholarly quality of departmental

faculty. This latter measure wasincluded as a criterion reference mea-

sure for psychology graduate students on the assumption that the qual-

ity of a department's faculty is related to the quality of the graduate

students the department attracts. A criterion reference measure for the

undergraduate group wasverbal reasoning ability as measured byper-

formance on the Verbal Reasoning subtest of the Differential Aptitude

Tests, Form T (Bennett, Seashore, &Wesman, 1974), which wasused

to investigate the relation between tacit knowledgeand verbal ability.

Procedure

The tacit knowledge measure wasmailed to membersof the faculty

and graduate student groups, and it was administered in a small-group

setting to members of the undergraduate group. The undergraduate

groupwasgiven, incounterbalanced order, the tacit knowledgemeasure

for academic psychologists as well as the tacit knowledge measure for

business executives that wasused in Experiment 2.

Results and Discussion

Basic Statistics

Return rates. The psychology faculty group returned90tacit

knowledgemeasures in time to be included in the study, yield-

inga return rateof 18%. The difference between the number of

respondents from highly rated departments (51) and from the

less highly rated departments (39) was not reliable, x2U, N =

90) = 1.34,p >.05.1randomly selected a sample of 40 nonre-

spondents to compare the rates of publication and citationof

respondents and nonrespondents. Faculty whoresponded were

comparable to faculty who did not in termsofnumberofpubli-

cations (Afs = 0.8 and 0.5 forrespondentsand nonrespondents,

respectively), £(106) = -1.63,p >.05, and number ofcitations

1Additional examples ofwork-related situations used inExperiments

1 and 2 are availablefrom theauthor.

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 5/12

1240 RICHARD K. WAGNER

Table1

BasicStatistics or Reference Measures: Psychology Faculty,

GraduateStudents, and Undergraduates

Criterion reference

measure

No. citations (1982-

1983)

No.publications'

Rated scholarly

quality of

departmental

faculty

No. papers presented

%oftime spent in

Teaching

Research

Administrative

duties

Age

n

68

68

88

91

89

89

89

91

M

Faculty

44.9

0.8

2.7

2.1

26.7

31.7

19.4

44.9

Median

7

0

3.8

2

30

30

10

42

SD

79.6

1.2

1.5

2.2

15.1

18.6

18.1

11.1

Range

0-361

0̂ 1

0.4-4.4

0-10

0-60

0-80

0-97

29-77

Graduate students

No. publications

No. papers presented

Ratedscholarly

qualityof

departmental

faculty

No.years completed

%oftime spent in

Teaching

Research

Age

No.research projects

completed

61

61

61

61

60

60

61

61

1.1

1.1

2.8

2.8

10.7

37.8

27.2

4.2

0

0

3.9

3

7.5

38.5

26

4

1.7

1.6

1.6

1.5

14.3

24.2

4.2

3.0

0-10

0-6

0.4-4.4

1-6

0-75

0-99

22-39

0-15

Undergraduates

Verbal reasoning" 60 45.0 46 4.1 32-50

1This represents only theportion ofasubject's publications appearing

injournals covered byPsychological Abstracts.bThe theoretical rangeofthis variable was 0 to 50.

ject'spublications thatappeared in journals covered byPsycho-

logical Abstracts. The mean number of total publications for

the psychology faculty based on their self-report was3.1 for the

last academicyear.

Classification of materials by independent judges. Twoinde-

pendent judges, a first-yeargraduate student and a postdoctor-

ate student from the psychology department at Yale University,

were given definitions for each content (managing self, tasks,

and others) and context (local and global) of tacit knowledge

and then were asked to determine which content of tacit knowl-

edge and which context were being sampled by the 12 work-

related situations. For Judge A, agreement with intended con-

tentand context was 12 out of 12 forcontentof tacit knowledge

and 10 out of 12 forcontext. These levelsof agreement yielded

kappa values (Cohen, 1960),which is an estimate of the propor-

tion of nonchance agreement, of 1.00 (p < .01) and .75 (p <

.01), respectively. For Judge B, agreementwithintended content

and context was 10 out of 12 forcontentof tacit knowledgeand

12out of 12 forcontext, yieldingkappa valuesof .75(p < .01)

and 1.00(p < .01), respectively.

Characteristics of the raw ratings on the psychology tacit

knowledge measure. I found a moderate rangeon the raw rat-

ingsobtained fortheresponseitems. Noobvious ceilingor floor

effects were apparent.

One question of interest was whether differences in actual

and ideal ratings would diminish with increasing levelsof pro-

fessional advancement, owing either to a change of valuesover

the span of one's career or simply becoming more pragmatic

and less idealistic with age. Totaled across items, the actual-

ideal squared deviations averaged 339.4, 412.8, and 428.S for

the psychology faculty, graduate student, and undergraduate

student groups, respectively. This linear decreasing trend was

reliable, F(\, 207) = 4.67, p < .05, and a follow-up analysis

(Newman-Keuls) indicated that the means of the graduate and

undergraduatestudent groups differed from that of the faculty

group but not from each other. Thus, compared with the stu-

dent groups, there was more congruence between the psychol-

ogyfaculty's actual and ideal responses.

(Ms = 44.9 and 48.2 for respondents and nonrespondents, re-

spectively), «(90)=0.19,p >.05.Thepsychology graduate stu-

dent group returned 61 tacit knowledge measures in time to be

included in the study, yieldinga return rate of 31%. The differ-

ence betweenthe numberofgraduate student respondents from

highly rated departments (36) and from less highly rated de-

partments (25)was not reliable, x2

(l,AT- 61) = 1.64,p > .05.Sample characteristics: criterion reference measures. De-

scriptive statistics for measures of criterion performance are

presented inTable 1. The onlysurprising result inTable 1 is the

largemean numberofcitations for thepsychologyfaculty (M=

44.9). As is suggested by the much lower median number of

citations (median = 7), the distribution of citations washighly

skewed because the work of a small number of subjects was

widely cited. Consequently, I based subsequent analyses on a

logtransformation of the citations variable. At first glance, the

averagenumber ofpublications (.8) seemed low for a 2-yearpe-

riod. This figure, however, represents onlythe portion of asub-

Scope of Tacit Knowledge

Quantification of tacit knowledge. Tacit knowledge was

quantified by comparing a subject's response item ratings to a

prototype derived from the mean response-item ratings of an

expert group. The expert group consisted of the 11 faculty

members from psychology departments that received scholarly

quality ratings equal to or greater than 4.2 on a 5-point scale

and who were highly competent in their profession.2

I calcu-

lated tacit knowledge scores bysummingthe squared deviations

of a subject's ratings from the prototype (i.e., the mean ratings

oftheexpert groupforeach item)for response itemsassociated

with each of the 12 work-related situations. This yielded 12

scores. I obtained subscale scores for the three contentsoftacit

2Analyses were carried out using several alternative sets of criteria

for selecting an expert group. The pattern of results was remarkably

consistent regardless of the specific set of criteria used.

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 6/12

TACIT KNOWLEDGE 1241

knowledge (managingself, tasks, and others) and the two con-

texts (localandglobal)bysummingtheappropriate scores.Sep-

arate scoringwasdonefortheactualand theideal ratings, yield-

ingactual and ideal scale scores.

After conducting preliminary analyses, I made one minor

modification in the scoring procedures. It became apparent

that the tendency to use the entire scale when responding, as

subjects were instructed, rather than to useonly the intermedi-

ate ratings, was an individual difference that, although unre-

lated to criterion measures of performance, affected tacit

knowledge scores generated by the prototype method.3Conse-

quently, the ratings were transformed sothat thestandard devi-

ation of ratingsacross itemswasequal for everyone.Allsubse-

quent analyses are based on these transformed ratings, which

had astandard deviation of 1.5.

Expert-novice differences in tacit knowledge. If the tacit

knowledge instrument sampled tacit knowledge acquired as a

result of formal training and experience, there should be a de-

creasing trend in average scores across groups with increasing

levels of professional development. Note that scores are ex-

pected todecrease rather than increase with advancing level of

experience because these scores represent deviation from the

expertprototype. Thusthecloser thepattern of responsesto the

expert prototype, thelower the score.

The expected decrease was found in tacit knowledge scores

across groups with increasing amounts of trainingand experi-

ence in the field of academic psychology. For the actual ratings,

meantotal scoreswere215.9,243.8, and 311.6for thepsychol-

ogy faculty, graduate student, and undergraduate student

groups,respectively. This linear trend in the expected direction

was reliable, f[\, 196) = 51.1, p < .001. Follow-up analysis

(Newman-Keuls)showed that all meansdiffered reliably at the

.01 level. For the ideal ratings, mean total scores were 195.1,

205.7, and 294.7 for the psychologyfaculty, graduate student,

and undergraduate student groups, respectively. This linear

trend in the expected direction wasreliable, F(l, 196) = 53.4,

p < .001. The mean of the faculty group differed from that of

the undergraduate group at the .01 level, and the mean for the

graduate student groupdiffered from thatof theundergraduate

groupat the .01 level,but themeansfor thefaculty andgraduate

studentgroupswerenot reliably different.

Determining whether there were group differences for each

of the tacit knowledge subscales was accomplished by using

Timm's (1975) procedure. The effect of groupwas significant

beyond the .001 level for each content and context of tacit

knowledge, forboth the actual andideal ratings.

Tacit knowledge andcriterionperformance inacademicpsy-chology. I calculated internal consistency reliabilities (coeffi-

cient alphas) for the tacit knowledgescales separately for the

psychology faculty, graduate student, and undergraduate stu-

dent groups. Reliabilities for the total score ranged from .74 to

.90, with a median of .82. The reliabilities of the individual tacit

knowledge subscales were somewhat lower and more variable,

rangingfrom .48 to .90, witha median of .69.

Subscale scores were highly intercorrelated for both the ac-

tualandideal ratings.Thecorrelationsbetween local andglobal

context were .65(p < .001) for the actual ratings and .79 (p <

.001) for the ideal ratings. The median correlations among the

Table2

Correlation Coefficients Between Tacit Knowledge Scores and

CriterionReference Measures for Academic Psychology

Tacit knowledgescore

Criterion reference

measure

Rated scholarly quality

of departmental

faculty

No. citations

No. publications

%oftimespent in

Teaching

Research

Administrative duties

No. paperspresented

Age

n

Faculty

77

59

59

79

79

79

80

80

Actual

total

-.48***

-.44***

-.28*

.26*

-.41***

.19*

-.21*

.22

Ideal

total

-.42***

-.43***

-.03

.09

-.34**

.16

-.03

.06

Graduate students

Rated scholarly quality

of program faculty

No. publications

No.papers presented

%oftime spent in

Teaching

Research

No. research projects

completed

No.yearscompleted

61

59

80

79

79

61

61

-.46***

-.25*

-.12

.15

-.48***

-.24*

-.07

-.24*

-.16

-.11

.14

-.13

-.15

-.16

*p<.05.**p<.01.***p<.001.

managing self, tasks, and others contents of tacit knowledge

were .54 (p <.001) for the actual ratings and .67(p < .001) for

the ideal ratings.Finally, the correlation between total score for

the actual ratings and total score for the ideal ratings was .61

(p<.001).

Correlations between criterion measures and total tacit

knowledgescores for the actual and ideal ratings are presented

in Table 2 for the faculty and graduate student groups. Total

score, a measure of overall deviation from the expert prototype,

negatively correlated with the following criterion measures: (a)

rated scholarly quality of departmental faculty, (b) extent to

which a subject'swork iscited (often considered abetter mea-

sure of the quality of a person's work than sheer quantity of

3Individual differences in subjects' use of the entire scale when re-

sponding,a frequent finding in rating studies, would have affected scores

based on the prototype method because scoreswere based ondeviation

from a prototype, which would vary with the extent to which a subject

used the entire rating scale. This possibility was confirmed by the

strengthening of relations between tacit knowledge scores based on the

prototype methodwhen ratingswere transformed toholdconstant indi-

vidual differences in subjects' use of the entire scale when responding.

The transformation was done by standardizing the standard deviation

of ratings across response items for subjects to the common value of

1.5.

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 7/12

1242 RICHARD K. WAGNER

Table3

CorrelationsAmongthe Six Kinds of Tacit Knowledgefor Psychology Faculty: Observed

(Below Diagonal) andDisattenuated or Reliability (Above Diagonal)

Tacit knowledge type 1 2 3

1. Self-local

2. Self-global

3. Tasks-local

4. Tasks-global

5. Others-local

6. Others-global

.20*

.04

.25*

.12

.21*

.38***

—.37***

.28**

.30**

.36***

.07

.59***

—.30**

.33***

.27**

.47***

.49***

.48***

—.34***

.37***

.23*

.53***

.53*"

.77***

—.39***

.99

.99

.99

.99

.99

" Thedisattenuated correlationwas greater than 1.00.

*p<.05. **p<.01. ***;>< .001.

publications), (c) number of publications, (d) percentage of

time spent in research, and (e) number of conference papers

presented. Total score waspositively correlated with percentage

of time spent in teaching and percentage of time spent in ad-

ministrative duties. Total score did not vary reliably as a linear

function of academic rank. For the actual ratings, means of

218.2, 224.1, and 204.3were obtained for the assistant profes-

sor, associate professor, and full professor groups, respectively,

fU, 74) = 0.59,p > .05. For the ideal ratings, means of 212.9,

191.4, and 191.5wereobtained for the assistant professor,asso-

ciate professor, and full professor groups, respectively, yielding

anF(\,14)= 1.14,/».05.

For psychology graduate students, total score was strongly

and negatively related to criterion measures such as (a) rated

scholarlyqualityof departmental faculty, (b) number of publi-

cations, (c)numberof conference papers presented, (d)percent-

age of time spent in research, and (e) numberof research proj-

ects completed. Correlations between total score and percent-

age of time spent in teaching and numberof years of graduate

study did not differ reliably from zero, but most were in the

appropriate direction.

For undergraduates, the correlation between verbal apti-

tude and total score on the tacit knowledge inventory was —.30

(p<.05).

Correlations with criterion measures for the individual sub-

tests, which are not reported because of their high degree of

intercorrelation, were similar in magnitude and pattern to those

obtained for total score.

Structure of Tacit Knowledge

Alternative models of the structure of tacit knowledge werefitted to the matrix ofcorrelations between scores representing

the six combinations of content and context. These corre-

lations, as well as the same correlations after disattenuating for

unreliability, are presented inTable3.

Two kindsof analyses—principal-components analysis and

confirmatory factor analysis—were carried out, and each sup-

ported the two-factor model that explains shared variance in

terms of asingle general factor.

The first principal component accounted for 40% of the total

variance, a figure typical of that found for traditional academic

aptitude tests. After the first principal component was ex-

tracted, the residual matrix was not significant, x2('4, N =

90) = 3.66. For the disattenuated correlations, the first princi-

pal component accounted for 72% of the total variance.

Confirmatory factor analysis can be used to test the hypothe-

sisthat agivenmodel did not generate the observed correlations

or covariances. Essentially, this is a test of the significance of

the discrepancy between predicted and observed correlations or

covariances, and support for the model is found if the test is

nonsignificant. This test was not significant, x2(9, N = 90) =

5.40,p> .05. Support forthe modelwasalso provided by a root

mean squared residual of only .04.

Hierarchical models with correlated group factors, that is,

models proposing group factorsinaddition to thegeneral factor,

yielded similarly good fits, but the parameter estimate for the

correlations between group factors equaled or exceeded one.

This indicates that all the variance accounted for by the group

factors wasaccounted for by a single general factor. Hierarchi-

cal models with orthogonal group factors did not fit the ob-

tained correlations. Thus the results support models that posit

a single general factor.

To extend the results to another domain, I conducted a sec-

ond experiment to replicate Experiment 1 in the field of busi-

ness management.

Experiment 2

Method

Subjects

Therewere threegroupsofsubjects, totaling 149people in all, whose

members differed in amountsof formal training and experience in the

fieldofbusiness managementThe business professional group consisted of 64 managers (63 men,

1 woman) from a nationwidesample of 31 companies. Thisgroup con-

tained 26 managers whose companies rank among the top 40 on the

Fortune 500 list, 33 managers whose companiesare not on the Fortune

500 list, and 5 individuals who chose not to reveal their company affili-

ation.

The business graduate student group consisted of 25 graduate stu-

dents (20 men, 5 women) from seven business schools. There were 9

first-year students, 14 second-year students, and 2 third-year students.

The group contained 15graduate studentswhoseschoolsare among the

highest ranked in thenation(TheCartter Report, 1977)and lOstudents

whoseschoolsare not among thehigher rankedbusiness schools.

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 8/12

TACIT KNOWLEDGE 1243

The undergraduate group consisted of the same 60 Yale undergradu-

ates whoparticipated in Experiment 1.

Materials

Abusiness management tacit knowledge measure was constructed to

be theoretically isomorphic to the academic psychology tacit knowledge

measure and was identical in format. The content of the work-related

situations and response items, however, reflected the domain of business

management. There were 39, 40, and 41 items on the managing self,

tasks, and others subscales, respectively, and 60 and 60 on the local and

global orientation subscales, respectively, for a total of 120 items. An

example of a work-related situation that sampled managing others with

a global context follows:

You have been asked to take over another department. Youhavea

reputation not only for getting the most from your employees, but

for getting along with them aswell. Youwere asked to take onthis

new job because of rather serious personnel-related problems in

the new department. Morale in the new department is low. The

department is divided into those who are sorry the former head

wasasked to leave and those who are sorry the former head was not

asked to leave sooner. Performance ratings for the department have

been below expectations.

The problems have been around for some time, and you realize

that solving them won't happen overnight. You also believe this to

be a chance to show your superiors what you can do in a tough

situation, and you hope that by doing well you will improve your

opportunities foradvancement.

Rate the quality of the following actions you are considering tak-

ing in yournewrole:

Actual Ideal

1. Follow the advice of your new superior by an-

nouncing a major reorganization of the depart-

ment that includes getting rid of individuals

whom you believe to be "deadwood."

2. Give your superiors frequent progress reportson

thesituation.. . .

10. Be sure your superiors are awareof how bad the

situation really was so they will appreciate even

modest improvement.

Design

The design of this experiment was identical to that of Experiment

1 with the following exceptions. Criterion reference measures for the

business professional group included (a) salary, (b) years of manage-

ment experience, (c) level of company, which was an indication of

whether subject's company was among the top companies in the Fortune

500 list, (d) years of formal schooling beyond high school, and (e) age.

Criterion reference measures for the business graduate student groupincluded (a) years of business school completed, (b) whether a subject's

school was among the top-rated group of business schools, (c) years of

management experience, and (d) whether currently employed.

Procedure

The tacit knowledgemeasure was mailed to members of the business

professional and graduate student groups, and it was administered in a

small-group setting to undergraduate group members. Members of the

undergraduate group were administered, in counterbalanced order,

both the business and academic psychology tacit knowledge measures.

Table 4

BasicStatisticsfor Reference Measures: Business

Professionals, Graduate Students, and Undergraduates

Reference measure M Median SD Range

Professionals

Salary(n = 61)'Years management experience

(n = 62 )Level of company (« = 59)

b

Years of formal schooling

(post-high school) (n- 63)

Age(n = 61)

85-95

20.9

0.4

4.6

50.1

100+

22.5

0

4

52

20

8.9

0.5

2.1

9.4

20-100+

0-45

0-1

0-8

23-63

Graduate students (n = 25)

Level of school' 0.6 1Yearscompleted 1.7 2

Years management experience 2.6 2

Employed at present 0.5 0

Verbal reasoning

Undergraduates (n = 60)

45.0 46

0.5 0-1

0.6 1-3

2.6 0-9

0.5 0-1

4.1 32-50

*Numbersrepresent thousands ofdollars.

"Thisis an indication of whether subject's company is listed among the

top Fortune 500companies.cThis is an indication of whether subject's business school is among

those ranked highest in the nation by The Carrier Report (1977).

Results and Discussion

Basic Statistics

Return rates. The business professional group returned 64

tacit knowledge measures in time to be included in this study,

yielding areturn rateof 13%. The difference between the num-

ber of respondents from companieson theFortune500 list (26)

and those from companiesnot on the list (33) was not reliable,

X2(l, N = 59) = 0.61, p > .05. The business graduate student

group returned 25tacit knowledge measures in time to be in-

cluded in the study, yielding an identical return rate of 13%.

The difference betweenthe number of respondents from highly

ranked business schools(15) and those from unranked business

schools(10)was not reliable, x2( 1,N =25) =0.64, p > .05.

Sample characteristics: criterion reference measures. De-

scriptivestatistics for thecriterionmeasuresofperformance are

presentedforthebusinesssamples inTable4. The business pro-

fessional group wasboth highly experiencedand well compen-sated. The mediannumber of years of managementexperience

was22.5, and themediansalarywas$100,000or more. Slightly

fewer than half of the group wasemployed by companies high

on the Fortune 500 list. One difference between the business

graduate student group and the psychology graduate student

group inExperiment 1 wasthat the former group already had

some experience in their ultimate careers:The business gradu-

atestudentgroup had onaverage2 years of management experi-

ence, andabout halfwerecurrentlyemployed.

Classification of materials by independent judges For both

judges, agreement with intended content was 12 out of 12 and

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 9/12

1244 RICHARD K. WAGNER

agreement with intended contextwas 10 out of 12.These levels

of agreement yielded kappa values of 1.00 (p < .01) and .66

(p < .01), respectively.

Characteristicsof the raw ratings on the business tacit knowl-

edge measure. Examinationof the raw ratings suggested no

obviousceilingor floor effects.

1 compared average actual-ideal squared deviations across

groups for evidence of a linear decreasing trend similar to that

found for academic psychologists. Average total squared devia-

tionswere 243.6,340.1, and416.7for thebusiness professional,

business graduate student, and undergraduate student groups,

respectively. This trend was reliable, F( 1,145) = 21.2,p<.001,

as weredifferences between each of the means. Thus increasing

professional advancement was related to congruence in tacit

knowledgeabout actualand ideal quality.

Scope of Tacit Knowledge

Quantification of tacit knowledge. The expert group, whose

average ratings constituted the prototype to which the ratings

of others were compared, consisted of 13 executives who (a)

were employedby companies on the Fortune 500 list, (b) had

titles higher in status and responsibility than vice-president

(e.g., executive vice-president, chairman, president), and (c) re-

ported annual salaries of$100,000ormore.

Expert-novice differences in tacit knowledge. Mean total

scores on the actual ratingswere 254.2,296.8, and 363.6 for the

business professional, business graduate student, and under-

graduate student groups, respectively. This linear trend in the

expected directionwas reliable, P(l, 132) = 31.9, p < .001. A

Newman-Keuls follow-up showedalldifferences among means

to be significant at the .01 level. Mean total scores for the ideal

ratings were 305.6 346.5, and 438.8 for the business profes-

sional, business graduate student, and undergraduate student

groups, respectively. This linear trend in the expected direction

was reliable, F(l, 132) = 38.6, p < .001, and all differences

among means were significant at the .01 level. Also, for both

actual and ideal ratings, groupdifferences significant at the .001

level were found for each content and context of tacit knowl-

edge.

Tacit knowledge and career performance in business man-

agement. I calculated internal consistency reliabilities (coeffi-

cient alphas) for the tacit knowledge scores separately for the

business professional, graduate student, and undergraduate stu-

dent groups. Reliabilities fortotal score ranged from .79 to .89,

with a median of .83. The reliabilities of the individual tacit

knowledge subscales were somewhat lowerand more variable,

rangingfrom .48 to .83, witha median of .67.

Subscale scores weremoderately to highly intercorrelatedfor

both the actual and ideal ratings, with the magnitude of the in-

tercorrelationscomparable across the actual and ideal ratings.

The correlations between local and global context were .73(p <

.001) for both the actual and ideal ratings. The median corre-

lations among the contents of managing self, tasks, and others

were .63(p< .001) for the actual ratingsand .61 (p <.001) for

the ideal ratings. Finally, the correlation betweentotal score for

theactual andideal ratingswas .71 (p < .001).

Correlations between criterion measures and total tacit

Table 5

CorrelationCoefficients Between Tacit Knowledge Scores and

CriterionReference Measures for BusinessManagement

Tacit knowledge score

Criterion referencemeasure

Actual

total

Ideal

total

Professionals

Salary (n= 48)

Years management experience

(n = 49)

Level of company (n =46)'

Yeais schooling (post-high

school) (n = 50)

Age(n = 50)

-.30*

-.05

-.32*

-.27*

-.12

.12

-.12

Graduate students (n= 25)

Level of school"

Yearscompleted

Years management experience

Employed atpresent

-.34*

.09

-.14

-.24

-.43*

.32

.10

-.11

• This is an indication of whether subject's company is listedamong the

top Fortune 500companies.b This is an indication of whether subject's business school is among

those ranked highest in the nation by TTieCartterReport (1977).

*p<.05.

knowledge scores for the actual and ideal ratings appear in Ta-

ble 5. For the business professional group, overall tacit knowl-

edge was related to salary and to years of management experi-

ence.

For the business graduate student group, overalltacit knowl-

edge wasrelated to level of school. The only puzzlingcorrela-

tion was that between years of schooling completed and total

score for the ideal scale.The correlation was expected to be neg-

ative, indicating that increasing years of schoolingwere associ-

ated withdecreasing deviation from the expert prototype. The

observed correlation waspositive and would have been signifi-

cant had two-tailed tests been used, but it was in the opposite

direction predicted for the one-tailed tests. What this result

means, ifanything, isunclear.

For the undergraduates, tacit knowledgescores were not re-

lated toverbal reasoning ability(r= -. 12, p >.05).

Thecorrelationsbetween criterion measures and theindivid-

ual subtest scores were similar in magnitude and pattern to

those obtained for total score.

Structureof Tacit Knowledge

Alternative models of the structure of tacit knowledgewere

fitted to the matrix of correlations between scores representing

the six combinations of content and context. These corre-

lations, aswell as the same correlations after disattenuating for

unreliability, are presented in Table 6.

As in Experiment1, the results supported the classofmodels

positing a single general factor and no group factors. The first

principal component accounted for 44% of the totalvariance,

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 10/12

TACIT KNOWLEDGE 1245

Table6

CorrelationsAmongthe Six Kinds of Tad! Knowledge for BusinessProfessionals: Observed

(Below Diagonal) and Disattenuated for Reliability (Above Diagonal)

Tacit knowledge type 1 2 3 4

1. Self-local

2. Self-global

3. Tasks-local

4. Tasks-global

5. Others-local

6. Others-global

_

.38**

.14

.25*

.37"

.38**

.49***

—.24*

.17

.33**

.35"

.25*

.52"*

.52"*

.44"*

.21

.52***

.30*

.99'

.43***

.36"

.74*"

.55"*

.92*"

.75*"—

.33"

.99'

.90*"

.69***

.97"*

.83*"

• Thedisattenuatedcorrelation was greater than 1.00.

*/><.05.«p<.01.*"p<.001.

and the residual matrixwas not significant, x2('4, JV = 64) =

5.66 For the disattenuated correlations, the first principal com-

ponent accounted for 76% of the total variance.

Confirmatory factor analysis also supported a single general

factor. A model with a general factor and no group factors

yieldedagoodfit,x2(9,W=64) = 12.13,p>05.Therootmean

squared residual was .08. Hierarchical models with correlated

group factors yielded similarly good fits, but the parameter esti-

mates for the correlations between groupfactorsequaled or ex-

ceeded one. Hierarchical models with orthogonal group factors

did not fit the obtained correlations.

Between-domain generality of tacit knowledge. The under-

graduate group wasgivenboth the psychology and business tacit

knowledge measures in counterbalanced order. It was possible

to examine the generality of tacit knowledge, at least for the

undergraduate group, byexaminingcorrelations across the two

tacit knowledgemeasures. As a prerequisite to examining these

correlations, it was necessary to rule out order effects on the

covariance structure of the interrelations among question

scores. Finding order effects on the covariance structure of in-

terrelations among question scores would indicate that the na-

tureof whatwas being measured had changed upon administra-

tion of a second tacit knowledge instrument.(Althoughfinding

noorder effects onmean scoreswas not aprerequisite forexam-

ining the between-scales correlations, there were no order

effects on mean subscale scores.)

For the psychologytacit knowledge measure, aLISRELanaly-

sis of the fit to the data of a model that constrained the covari-

ances of the tacit knowledge measures to be equal across order

of administrationwas not significant, x2(21, N = 60) = 19.7,

p =.54. This suggestsno reliable order effect on the covariance

structure of the psychologytacit knowledge measure. For the

business tacit knowledgemeasure, analysis of the fit of the same

model also was not significant, x2(21, N =60) = 23.6 p = .32,

which,again, suggestsno reliable order effect on thecovariance

structure of the businesstacitknowledge measure.

Thegenerality oftacit knowledge was supported by the corre-

lations obtained between scores on the psychology and business

tacit knowledge measures: total score (r =.58, p <.001), local

orientation (r =.54,p < .001), global orientation (r =.50,p <

.001), managing self (r = .52, p < .001), managing tasks (r =

.47,p <.001), and managing others (r = .52,p < .001). Corre-

lations of this magnitude indicate that the two measures shared

approximately 25 to 35% of the total variance in tacit knowl-

edgescores.

General Discussion

The extent to which the results of the two present experi-

ments were similar to each other and to previous work was sur-

prising. The pattern of results wasnearly identical forExperi-

ments 1 and 2, which were carried out in different domains.

These results provide clear support for two previous findings

and partial support for a third.

First, inboth experiments, I found reliabledifferences intacit

knowledge across groups whose members differed in level of

professional advancement. The responses of academic psychol-

ogists were closer to the expert prototype than were those of

graduate students in psychology, whose responses were closer to

theexpert prototype than those ofthe undergraduates. Identical

findings were obtained for business professionals, business

graduatestudents,andundergraduates.

Second, in both experiments, within-group individual

differences in tacit knowledge were related to criterion mea-

sures of performance appropriate to each of the groups, with

some correlations of amagnitude (.3-.5) IVi to 2V 4 times that

ofcorrelationstypically found between performance on ability

tests and job performance. The resultant values of r2

(propor-

tion of variance accounted for) were2 to 5 times larger.

Third, in Experiment2, tacit knowledge was unrelated to

verbal aptitude. However, performance on the academic psy-

chology measure used in Experiment 1 was related to verbal

aptitude. Previously,no reliable relations were found forunder-

graduates between tacit knowledge and verbal aptitude as tradi-

tionallymeasured (Wagner &Steinberg, 1985). One difference

between these and previous experiments is that the present ex-

periments were more sensitive in detecting weak relations be-

tween tacit knowledge and verbal intelligence because of a larger

undergraduate sampleand more reliable tacit knowledge mea-

sures. Why there should be a relation with verbal aptitude for

the academic psychology measure and not the business measure

remains unclear. Thesedifferences cannot be due to differences

in subjects because I used the same group of undergraduates in

both experiments. An adequate determination of the true de-

gree of relation between tacit knowledge and verbal aptitude

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 11/12

1246 RICHARD K. WAGNER

will requiregivinga tacit knowledge measureand an IQtest to

largegroupsof professionals asopposed to undergraduates.

Scope of Tacit Knowledge

The most important finding to emerge from the present ex-

periments concerns the breadth of tacit knowledge. Tacit

knowledge is not simply "careerism,"that is, knowing how to

promote one's career, nor is it simply formal knowledge about

such things as the names of one's tools. Whereas Hunter (1983)

has reported relations between formal knowledge and perfor-

mance for occupations such as medical technician, the present

resultsimply that such relations between knowledgeand perfor-

mance extend to the informal knowledge of people in manage-

rial and professionalcareers.

The between-group differences in tacit-knowledge across

groups whose members differed in level of professional ad-

vancement and the within-groupcorrelationswithcriterion ref-

erence measureswere similar for tacit knowledge useful in the

short-term accomplishment of a variety of tasks, as well as in

tacit knowledge useful inattaining one's long-range objectives.

These results extended across work-related situations that re-

quired skill at managing oneself, one's tasks, and others. Fur-

thermore, these differences encompassed judgments about the

ideal quality as well as the practicality of possible responses to

work-related situations.

An interestingfinding forboth experiments wasthat increas-

ing level of professional advancement was associated with in-

creasing congruence between judged ideal quality (ideal ratings)

andpracticality (actual ratings). There are a number of possible

explanations forthis result. Perhaps because what counts in ac-

ademic psychology and business is actual accomplishment,

whether in terms of publications and student evaluations of

one's teaching or in terms ofgross earnings for one's division,

people in these career pursuits come to weigh practicality

heavily in their judgments about the quality of possible re-

sponses to work-related situations. Alternatively, it may be that

people generally become less idealisticwith age or that their val-

ueschange over time regardless of the reward structure of their

occupations.

Structureof Tacit Knowledge

The resultsofbothexperiments support a model ofthe struc-

ture of tacit knowledge characterized by a substantial general

factor. Thus, for the present, individual differences in tacit

knowledge are best described in terms of a general ability or

fund of knowledge, as opposed to a collection of independent

abilities or funds of knowledge. The various kinds of tacit

knowledgeshouldbeviewedasfacets of ageneral abilityor fund

of knowledge as opposed to psychological factors in their own

right. In practical terms, people who arehighlyknowledgeable

about managing themselves tend to be quite knowledgeable

about managing others and managing their tasks. Such people

can be expected to be knowledgeable about both short-term and

long-term aspects of work-related situations and about ideal

quality aswell as practicality.

Inaddition to the outcomeoftesting alternative modelsofthe

structureof tacit knowledge,the generality oftacit knowledgeis

supported by (a) the intercorrelations(.5-.8) among the scales

constructed to measure distinct contents and contexts of tacit

knowledge; (b) thecorrelations(.7-.8)between performancefor

the actual and ideal ratings; (c) the correlation (.6) found for

members of the undergraduate group between performance on

the tacit knowledge measures for academic psychology and

business management; and (d) the highly similar patterns of

correlations with external criterion measures acrossdomains,

across actual and ideal responses, and across the contents and

contexts of tacit knowledge.

The fact that a generalized ability accounts for muchof the

variance in performance on academic-type tasksmight make a

similar finding for more practical tasksless surprising. Butnote

that the evidence for a general ability in the domain of social

intelligence has, at best, been mixed (see, e.g., Ford & Tisak,

1983;Keating, 1978;Sternberg&Smith, 1985;Walker&Foley,

1973). There may,however, beproblemswithsocial intelligence

measures because except for Hogan's (1969) Empathy Scale

and Ford and Tisak's(1983) behaviorally derived measure, so-

cial intelligence measures have been unrelated to the measures

they should be related to, including other social intelligence

measures. The present results are consistent with two recent

large-scale studies of social competence. In a cross-cultural

study involving hundredsofAnglo,Chicano, andMexican chil-

dren, Mercer,Gomez-Palacio, and Padilla (1986) found inter-

correlations among subscales of a measure of adaptive behavior

to exceed those of the Wechsler Intelligence Scale for Chil-

dren—Revised, which wasgiven to the same samples. More im-

portant, the correlation between performance on the adaptive

behavior scale and performance on the IQ test was not signifi-

cantly different from zero. Ford and Tisak (1983) reported evi-

dence for a social competence factor that was general to several

measures of social competence yet distinct from measures of

academic competence.

There are two obvious sources of a general factor for tacit

knowledge, which not surprisingly are two possible sources of

the general factor for psychometric aptitude tests. The general

factor may be a manifestation of something inside the head,

such as a general ability to acquire tacit knowledge. Alterna-

tively, the general factor mayreflect thenature oflearning expe-

riences: Experience in a given situation may inform one's un-

derstanding of the management ofself, others, and tasks. Thus

the fact that a person is relatively knowledgeable about each

kind oftacit knowledge might merely reflectgreaterexperience

insituations that provide generalknowledge.

A number of limitations of the present studies qualify the

conclusion that the structure of tacit knowledge is character-

ized by a general factor. First, the results are based on studies

conducted in twoprofessional domains. Movingbeyondprofes-

sional and managerial domains, or perhaps moving to manage-

rial and professional domains other than academic psychology

and businessmanagement, mayyield a different picture of the

structure of tacit knowledge. Second, the results are specific to

the particular measures used in the present studies. The results

might have been different had I measured other kinds of tacit

knowledge. Third, the generality may reflect something the

measureshave in common in addition to tacit knowledge,such

8/2/2019 TK_Wagner

http://slidepdf.com/reader/full/tkwagner 12/12

TACIT KNOWLEDGE 1247

asresponse format. However, note that acommon responsefor-

mat has not been sufficient to yield a general factor in previous

studies ofsocial intelligence.

Obviously, it takes more than tacit knowledge to succeed in

an occupational setting. Motivation, academic ability, leader-

ship, chance, and other still undetermined elements undoubt-

edlyplay important roles.The importance of the present exper-

iments is to suggest that in addition to these other elements,

tacitknowledge playsa nontrivial role in work-related practical

judgment anddecision-making.

Inthe minds of laypersons and experts alike, intelligence in-

cludes more than the competencies required to succeed in the

circumscribed, although admittedly important, environment

of formal classroom settings (Sternberg, Conway, Ketron, &

Bernstein, 1981). Our understanding of intelligence will con-

tinue to be limited, at best, until the breadth of domains in

whichwe study intelligence becomes a closer approximationof

the breadth of domains in which intelligence is manifested.

References

Bennett, G. K., Seashore, H. G., &Wesman, A. G. (1974). The Differ-

ential Aptitude Tests (Form T). NewYork:The PsychologicalCorpo-

ration.

The Cartter report on the leading schools of education, law, and busi-

ness. (1977, February).Change, pp.44-48.

Cattell, R. B. (1963). Theory of fluid and crystalized intelligence: A

critical experiment.Journal of Educational Psychology, 54,1-22.

Cattell, R. B.Q97\).Abilities:Their structure, growth, andaction. Bos-

ton: Houghton-Mifllin.

Charlesworth, W R. (1976). Intelligence as adaptation: An ethological

approach. In L. Resnick (Ed.), The nature of intelligence (pp. 147-

168). Hillsdale, NJ:Erlbaum.

Chi, M. T.H.,Glaser, R., &Rees, E. (1982). Expertise in problem solv-

ing. In R. Sternberg(Ed.),Advances in thepsychology of human intel-ligence^. 1, pp.7-75).Hillsdale, NJ:Erlbaum.

Cohen, J. (1960). Acoefficient of agreement for nominal scales. Educa-

tionalandPsychological Measurement, 20,37-46

Ford, M. E., &Tisak, M. S. (1983). A further search for social intelli-

gence. Journal of Educational Psychology, 75, 197-206

Ghiselli, E. (1966). The validity of occupational aptitude tests. New

York:Wiley.

Hogan, R. (1969). Development of an empathy scale.Journal of Con-

sultingand Clinical Psychology, 33, 307-316

Hunter, J. E. (1983). A causal analysis of cognitive ability, job knowl-

edge, obperformance, and supervisor ratings. In F.Landy.S. Zedeck,

& J. Cleveland (Eds.), Performance measurement and theory (pp.

257-266). Hillsdale, NJ: Erlbaum.

Jensen, A. R. (in press). The g beyond factor analysis. In B. Plake, J.

Win, &R. Ronning (Eds.), The influence of cognitivepsychology on

testingandmeasurement. Hillsdale, NJ: Erlbaum.

Jones, L. V, Lindzey,G., &Coggeshall, T. E. (Eds.). (1982).An assess-

ment of research-doctorate programs in the United States:Social and

behavioral sciences. Washington, DC: National Academy Press.

Keating,D. P.(1978).Asearchforsocialintelligence. Journal of Educa-

tionalPsychology, 70,218-223.

Mercer, J. R., Gomez-Palacio, M., &Padilla, E. (1986). The develop-

ment of practical intelligence in cross-cultural perspective. In R.

Sternberg & R. Wagner (Eds.), Practical intelligence:Natureand ori-

gins of competence in the everyday world (pp.307-337). Hillsdale,

NJ: Erlbaum.

Neisser, U. (1976). General, academic, and artificial intelligence. In L.

Resnick (Ed.), The nature of intelligence (pp. 135-144). Hillsdale,

NJ:Erlbaum.

Polanyi, M. (1976). Tacit knowing. In M. Marx & F.Goodson (Eds.),

Theories in contemporary psychology. New York: Macmillan.

PsychologicalAbstracts. (1982).Arlington, VA: American Psychological

Association.

PsychologicalAbstracts. (1983). Arlington, VA: American Psychological

Association.

Schon, D. A. (1983). The reflective practitioner.NewYork:Basic Books.

Social Sciences Citation Index. (1982). Philadelphia: Institute forSci-

entific Information.

Social Sciences Citation Index. (1983). Philadelphia: Institute for Sci-

entific Information.

Spearman, C. (1927). Theabilitiesof man. NewYork: Macmillan.

Sternberg, R. J., Conway, B. E., Ketron, J. L., &Bernstein, M. (1981).

Peoplesconceptions ofintelligence.Journal of Personality and Social

Psychology, 41. 37-55.

Sternberg, R. J.,&Smith, C. (1985). Social intelligence and decoding

skills in nonverbal communication. SocialCognition, 3, 168-192.

Steraberg, R. J., &Wagner, R. K. (1986). Practical intelligence:Nature

and origins of competence in the everyday world. New York:Cam-

bridge University Press.

Thurstone, I. L. (1938). Primary mental abilities. Psychometric Mono-

graphs (No. 1). Chicago: University of Chicago Press.

Timm, N. H. (1975). Multivariateanalysis: With applications ineduca-

tion andpsychology. Monterey, CA:Brooks/Cole.

Wagner, R.K. (1986). Intraterrestrial intelligence. In R. Sternberg & R.

Wagner (Eds.), Practical intelligence: Nature and origins of compe-

tence in the everyday world (pp. 361-378). New York: Cambridge

University Press.

Wagner, R. K., &Sternberg, R. J. (1985). Practical intelligence in real-

world pursuits: The role of tacit knowledge. Journal of Personality

andSocialPsychology, 48,436-458.

Walker, R. E., & Foley, J. M. (1973). Social intelligence: Its history and

measurement. PsychologicalReports, 33,839-864.

Wigdor, A. K., & Gamer, W. R. (Eds.). (1982). Ability testing: Uses,

consequences, and controversies. Washington, DC: National Acad-

emy Press.

ReceivedMay 13, 1986

Revision received September 15, 1986

Accepted December 12, 1986 •