douglas mcgregor's theory x and y: toward a construct ... · ries found that theory x/y was...

JOURNAL OF MANAGERIAL ISSUES Vol. XX Number 2 Summer 2008: 255-271

Douglas McGregor's Theory X and Y: Toward a Construct-valid Measure*

Richard E. Kopelman Professor of Management

Baruch College

David J. Prottas Assistant Professor of Management

Adelphi University

Anne L. Davis Colonel

Tooele Army Depot

Douglas McGregor's landmark book, The Human Side of Enterprise (1960), changed the path of manage-ment thinking and practice. Ques-tioning some of the fundamental as-sumptions about human behavior in organizations, he outlined a new role for managers: rather than command-ing and controlling subordinates, managers should assist them in reach-ing their full potential. At the foun-dation of McGregor's Theory Y are the assumptions that employees are: (1) not inherently lazy, (2) capable of self-direction and self-control, and (3) capable of providing important

ideas/suggestions that will improve organizational effectiveness. Thus, with appropriate management prac-tices, such as providing objectives and rewards and the opportunity to par-ticipate in decision making, personal and organizational goals can simul-taneously be realized. In contrast to Theory Y, McGregor posited that con-ventional managerial assumptions (which he called Theory X) reflect es-sentially an opposite and negative view—viz., that employees are lazy, are incapable of self-direction and au-tonomous work behavior, and have little to offer in terms of organiza-

• We gratefully acknowledge the helpful comments of our colleagues, Abe Korman, Allen Kraut, Hannah Rothstein, and Donald Vredenburgh as well as an anonymous reviewer. Earlier versions of this article were presented at the 112th Meeting of the American Psychological Association (2004) and the 22nd Annual Conference of the Society for Industrial and Organizational Psychology (2007).

JOURNAL OF MANAGERIAL ISSUES Vol. )0( Number 2 Summer 2008

(255)

256 THEORY X AND THEORY Y

tional problem solving. Hereafter, we refer to McGregor's theorizing as Theory X/Y.

Indicative of McGregor's impact, Miner's (2003) review of 73 estab-lished organizational behavior theo-ries found that Theory X/Y was tied for second in terms of recognition and in 33rd place with respect to im-portance. By the time The Human Side of Enterprise was republished in 1985, it had become a classic with the book jacket reading like a Who's Who in Management. Drucker hailed it as "ever more relevant, more timely, and more important." Townsend called it "the most powerful and use-ful book about people I've ever read." Kanter claimed it contained "profound and timeless truths." Wa-terman declared it "a classic text that is a fundamental touchstone for any-one in management and organiza-tional development." Bennis wrote ‘`. . . this book, more than any other book on management, changed an entire concept of organizational man and replaced it with a new paradigm that stressed human potentials, em-phasized human growth, and ele-vated the human role in industrial so-ciety" (McGregor, 1985: iv).

However, as Miner noted in his comprehensive (2002) text on organ-izational behavior theories and re-search, " W here are very few direct tests of McGregor's formulation in the literature. . . . Furthermore, McGregor himself conducted no re-search related to his formulations, nor did he attempt to make his vari-ables operational in any kind of meas-urement procedures" (2002: 261). In our view, McGregor's theorizing about the effects of individual differ-ences in managerial assumptions has remained virtually unexamined due to the absence of prior construct val-

idation research. Clearly, it is not pos-sible to test McGregor's theory if the central construct—the assumptive world (or cosmology) of the focal manager—lacks a published, con-struct-valid measure. In light of this long overdue undertaking, the pres-ent research reports on the develop-ment and construct validation of a measure of Theory X and Theory Y assumptions/attitudes.

McGregor identified a number of management practices that he thought were consonant with Theory Y assumptions (such as participative leadership, delegation, job enlarge-ment and performance appraisals). Consequently—and unfortunately in our view—tests of the efficacy of these management practices were often in-terpreted as a proxy for assessing the validity of McGregor's theorizing. Successful implementation of partic-ipative leadership, for example, is at best only tangentially related to McGregor's theorizing. Moreover, McGregor recognized that imple-mentation of these practices with a Theory X mindset will be limitedly successful, with employees seeing such techniques as disingenuous ma-nipulations (Heil et al., 2000; Mc-Gregor, 1966, 1967).

At the heart of McGregor's argu-ment is the notion that managers' as-sumptions/attitudes represent, po-tentially, self-fulfilling prophecies. The manager who believes that peo-ple are inherently lazy and untrust-worthy will treat employees in a man-ner that reflects these attitudes. Employees, sensing that there is little in the job to spur their involvement, will exhibit little interest and motiva-tion. Consequently, and ironically, the manager with low expectations will lament that "you can't get good help nowadays," oblivious as to the

JOURNAL OF MANAGERIAL ISSUES Vol. XX Number 2 Summer 2008

KOPELMAN, PROTTAS AND DAVIS 257

actual nature of cause and effect. Closing the self-reinforcing cycle, the manager feels vindicated; that is, his/ her low expectations were warranted. Conversely, the manager who be-lieves that employees are generally trustworthy and desirous of growth will facilitate their achievement. McGregor's explanation was that the manager had created conditions that enabled "the individual to achieve his [her] own goals (including those of self-actualization) best by directing his [her] efforts toward organiza-tional goals" (1967: 78). Subse-quently, numerous, more intricate, psychological and social-psychologi-cal mechanisms have been invoked to explain this phenomenon (e.g., Ban-dura and Locke, 2003; Eden, 1990; Heil et al., 2000; McNatt and Judge, 2004).

McGregor (1957, 1967) noted that some businesses were adopting prac-tices that could be expected to yield superior results, such as decentrali-zation and delegation, job enlarge-ment, participative/consultative management, and performance ap-praisal. However, he also observed that these programs often were un-successful due to the way they were implemented. When those executing the programs did so with Theory X attitudes or within organizations with Theory X climates, the programs would be likely to fail—perhaps an-other self-fulfilling prophecy.

Eden (1990) reported on numer-ous field experiments demonstrating that when managers were led to have high expectations of some subordi-nates (based on fictitious informa-tion), the subordinates outper-formed their peers. Attempting to apply this finding to leadership train-ing—but without using deception—Eden et al. found weak results in seven

field experiments, results they char-acterized as "a disheartening basis for practical application" (2000: 195). Indeed, Eden et al went on to say that leadership training, in general, may be unrealistic; some managers "have it naturally and some do not, and those that do not cannot be trained, coaxed, or coached to have it" (2000: 204; emphasis added). However, nei-ther the early nor the latter studies by Eden and his colleagues speak to McGregor's theorizing; in all of Eden et al.'s research, expectations were ar-tificially manufactured. In contrast, McGregor's theory relates to organic differences in managers' assumptive worlds (or cosmologies). To wit per-haps the Theory Y managers "have it."

To our knowledge, only one field investigation (Fiman, 1973) has been conducted that speaks directly to the posited effects of Theory Y manage-rial attitudes. In Fiman's study of fe-male clerical employees and their su-pervisors in one corporation, a perceived Theory Y managerial ori-entation was positively related to job satisfaction but unrelated to job per-formance. Fiman's X/Y attitude items were never published and the only construct validity information Fiman (1973) reported was split-half relia-bility coefficients.

Although McGregor's Theory X/Y may be characterized as representing a dispositional mindset suggestive of a one-best-way to manage, McGregor recognized that a Theory Y manage-rial style will not be appropriate in all situations (Heil et al., 2000; Mc-Gregor, 1967). In any event, before McGregor's theory and numerous theoretically-related propositions can be researched (see the Discussion section), it is necessary to develop a construct-valid measure of the central



concept. Accordingly, we see the present endeavor as a critical first step in assessing the substantive valid-ity of McGregor's theorizing.

METHOD

Sample

We distributed surveys to under-graduate and graduate students in business at two east coast colleges from 2002 to 2006. Participation was voluntary and anonymous. The final sample consisted of 512 participants with a mean age of 28 years (using midpoints of categories and age 55 for 50+) and was 56% female. Nearly 80 percent of respondents were cur-rently or recently employed, with 68% working in the private sector. The largest categories of employment were financial services (24%), health care (20%) and miscellaneous pro-fessional services (16%). Respon-dents tended to work for either very large organizations with over 1,000 employees (44%) or small organiza-tions with fewer than 100 (32%), with 24% working for organizations of in-termediate size. Their mean annual salary was $55,800.

Measures

The survey consisted initially (N 159) of four principal sections: 17 items measuring Theory X and The-ory Y attitudes (drawn from two sources described below); 19 items measuring Theory X and Theory Y behaviors (drawn from two prior works); five items measuring faith in people; and five items measuring fast food opinions. A fifth section consist-ing of three items relating to leisure time activities was added to the later version of the survey (N = 353). With regard to our theorized nomological

network, we reasoned that Theory X/ Y attitudes and assumptions would be closely related to Theory X/Y behav-iors and that Theory X/Y attitudes and behaviors would be positively but distally related to generalized faith in people. However, we could see no reason why opinions about fast food meals and leisure-time activity pref-erences would be related to Theory X/Y attitudes or behaviors.

17seory X/Y Attitudes and Assumptions were assessed by 17 items (see Appen-dix) drawn from two sources. We in-corporated ten items from a scale en-titled "McGregor's Theory X-Y Test" (Swenson, n.d.) and seven items were selected from the "Theory Y/Theory X Leadership Assumption Test" (Scanlon Leadership Network, n.d.). The latter source consisted initially of ten items, but three were dropped be-cause they mirrored items in the first scale. All 17 items were scored on a five-point Likert scale with end-points ranging from "strongly disagree" to "strongly agree." Scanlon's Leader-ship Assumption Test is a product of the Scanlon Leadership Network and the measure appeared on their web-site (www.scanlonleader.org ). Mc-Gregor's Theory Y aligns with Scan-lon's belief that organizations can be more effective if information is shared between managers and em-ployees, and the latter are involved in problem solving. It might be noted that in both of McGregor's books (1960/1985, 1966) an entire chapter was devoted to the Scanlon Plan. There is no available evidence sup-porting the reliability and validity of the scores on either the Scanlon or the Swenson measures. Cronbach al-pha for these 17 items (hereafter, the "17-item X/Y attitude scale") was .78. It should be noted that the au-thors only became aware of Fiman's


KOPELMAN, PRO1TAS AND DAVIS 259

(1973) research after the present in-vestigation was well underway. Like-wise, we have recently discovered a few more scales published in organi-zational behavior textbooks, some combining attitudes and behaviors; none with psychometric data or con-struct validity evidence. Two non-public-domain instruments exist, one only available commercially (Teleo-metrics International, 1995). A list of all known X/Y scales, including prop-erties and construct validity evidence is provided in Table 1.

Timmy X/Y Behaviors were assessed by 19 items drawn from two sources. We incorporated 15 statements adapted from Costley and Todd's (1987) list of managerial actions that result from Theory X and Theory Y beliefs and we also used the four-item measure developed by Miles (1964). Costley and Todd (1987) listed seven actions that result from Theory X be-liefs and six actions that result from Theory Y beliefs. We modified these items slightly to improve clarity. A sample item is: "The amount of re-sponsibility given to employees should be limited and controlled." Miles' (1964) scale was originally de-veloped to measure managers' atti-tudes about participative leadership policies. Miles' items were slightly modified (e.g., changing "subordi-nates" to "employees"). Response options used the above described five-point Likert scale. Cronbach alpha for these 19 items (hereafter, the "19-item X/Y behavior scale") was .71.

Faith in People was assessed by five items (two forced choice items and three agree-disagree statements) from Rosenberg (1957). Positive re-sponses indicate an absence of faith in people; we reverse coded re-sponses, with scores ranging from 1 (low faith) to 6 (high faith). Validity

evidence might be adduced from the occupational choices of the 4,585 na-tionwide college students who com-pleted the instrument years ago (cf. Robinson and Shaver, 1973). Stu-dents with a high faith-in-people score selected people-oriented occu-pations such as social work, human resource management, and teaching. Both men and women with low scores tended to select occupations such as sales, finance, and advertising. Cron-bach alpha in the present study was .55.

Fast Food Opinion Scale consisted of five items developed by the authors to measure opinions about fast food meals. A sample item is: "On the whole, I would say that a meal con-sisting of a McDonald's hamburger, fries and soda is an ideal meal." Re-sponse options used the same five-point Likert scale. Cronbach alpha for this study was .76.

Leisure-Time Activity items were also developed by the authors to serve as unrelated measures. A sample item is: "Roughly how many hours per week do you spend watching television?" Re-sponse alternatives were 0-4 hours, 5-10 hours, and more than 10 hours, with corresponding scores of 1 to 3, respectively. The other two leisure items related to number of movies at-tended annually and hours per week spent reading for pleasure. Because each of the leisure time items en-tailed the expenditure of time, we found moderate levels of intercorre-lations: rs of .27, .30, and .34.

RESULTS

It has long been recognized that an assessment of the construct validity of a measure should precede substan-tive research (e.g., Schwab, 1980). Ac-cordingly, we conducted a number of


8008

IQU

IUM

S g la

CIU

MN

Table 1 Listing of Extant Theory X/Y Attitude and Behavior Scales

Source Type Items

Published Public

Domain Description Psychometric/Construct

Validity Evidence

1. Swenson (n. d.) Attitude Yes Yes 10 items, 5-point ratingscale None published

2. Scanlon Institute (n. d.) Attitude Yes Yes 10 items 4-point ratin gscale None published

3. Costley and Todd (1987) Behavior Yes Yes 10 items, no scale provided None published

4. Miles (1964) Behavior Yes Yes Four items, 5-point rating scale

Correlations and comparisons of group means

5. Fiman (1973) Attitude and Behavior No Yes

29 attitudinal items and 12 behavioral items, undefined rating scales

Split-half reliabilities

6 ' Chapman (in Borkowski, 2005) Attitude Yes No 15 items, 5-point rating

scale None published

7. Greenberg (1999) Behavior Yes Yes 8 pairs of forced choicestatements None published

8 ' Osland, Kolb and Rubin (2001) Behavior Yes Yes 10 pairs of forced choice

statements None published

9. Baron and Paulus (1991) Attitude Yes Yes 7 pairs of forced choicestatements None published

10. Gordon (1999) Attitude and Behavior Yes Yes 12 pairs of forced choice

statements None published

11 . Teleometrics International (1995) Attitude Yes No 36 items, 7-point rating

scale Self-published reliabilities and correlations.

KOPELMAN, PROTTAS AND DAVIS

261

empirical analyses to develop a con-struct-valid measure of Theory X/Y attitudes. Table 2 presents descriptive statistics, reliabilities, and intercorre-lations among the variables utilized in these analyses.

Factor Analysis and Creation of an Abbreviated Scale

We performed CFA using struc-tural equation modeling with LISREL (Version 8.53; Joreskog and Sorbom, 2002). We specified a model using three "parcels" for each of the X/Y attitude and X/Y behavior scales and two "parcels" for each of the fast food opinion and trust in people scales. More specifically, the X/Y at-titude and X/Y behavior scale parcels were comprised of 6, 6, and 5 items and 6, 6, and 7 items, respectively, constituting all items of the two scales. The two parcels related to each of the fast food and trust in peo-ple scales consisted of three and two items. The CFA model's fit statistics (x2 = 42.10, df = 29, p = .055, x2/df = 1.45; GFI = .98; AGFI = .97; CFI, .99; NFI = .98; RMSEA = .03) showed that the measurement model had very good fit. The ratio of x2/df was below the recommended value of 3 and the fit statistics values were at or above the recommended thresholds of 0.9 for NFL above 0.8 for AGFI, and the RMSEA value was below the recommended value of 0.10 (Hair et al., 1998). All X/Y attitude items loaded significantly on their assigned latent constructs, although one of the parcels for X/Y behaviors had a co-efficient of .64, below the recom-mended threshold of .70, and the two parcels for fast food loaded at .68 and .41. This model had a better fit than alternative models where the three X/Y attitude and three X/Y behavior

"parcels" were loaded onto a single latent variable (x2 = 274.73, df = 32, p = .00, x2/df = 8.59; GFI = .90; AGFI = .83; CFI, .90; NFI = .89; RMSEA = .12) and where all ten par-cels were loaded onto a single latent variable (x2 = 597.51, df = 35, p = .00, x2/df = 17.08; GFI = .81; AGFI = .70; CFI, .69; NFI = .68; RMSEA = .18).

Given the high degree of concep-tual overlap among the 17 Theory X/ Y attitude items, we sought to deter-mine if a shorter scale might be de-veloped. We conducted a principal axis exploratory factor analysis (EFA) with Varimax rotation using meth-odology similar to that employed by Kelly and Lee (2002) and Sato (2003). (Principal axis factoring is the preferred exploratory-descriptive method of factor extraction when an-alyzing common variance. Orthogo-nal (i.e., varimax) rotation yields fac-tors that are maximally independent.) There are varying opinions concerning the sample size required to perform factor analysis, but it is generally accepted that 10 re-spondents per item is sufficient (Tin-sley and Tinsley, 1987), and our sam-ple provided more than 20 cases per item. We first conducted EFA on the 17 items, suppressing coefficient val-ues (factor loadings) that were less than .50. This analysis yielded four factors with eigenvalues > 1.0. The first factor, with an eigenvalue of 4.4, accounted for 26.1% of the variance. The second factor had an eigenvalue of 1.30 and accounted for 7.6% of variance, and the remaining two fac-tors together accounted for 14.0%. Five items loaded above .50 on three factors. We then conducted a second iteration of EFA, again suppressing coefficients less than .50. Four items were retained, loading on a single fac-

JOURNAL OF MANAGERIAL ISSUES Vol. 30C Number 2 Summer 2008

l'0171.-01'Ai > d*** !PTIM-0 NO 'TO' > d** tPoPei-oml > d* . sasamuand u! srpuoRullp uo maddr arras uoua ioJ stuldp yaequoij •areas maj!

-100J apnume Aix r.fi 2turtpxo ,afgos oPtIIPW A/X suloll agl jo pasyduroo s! glum wal!-E1 apnlply •S.1()!Autva = •tioa !sapul91y = .(01mu = Z tgIumgJ = xas :giqugun tuppogaTeD . sasicigur zoj pasn (OS R- JOJ SS pun) auez Jo smociPlul glIm (05 < `617-5b `bb-Ob '6E- 5E

'17E-OE `6z-gz `bz-Oz) a8ue1 a sr POSSOAX0 geM a8y '(ZgE = N) SUM! ampt oqi puu `(z617 = N) ginugl `(0017= N) ARM Jo.; idooxo I 1 c o1 90S = N *RioN

(IL') ***lc' ***RE' ***917' ***lir LO . 170' 90' **9T • 80' OT ' 80' - **ZI' 9E' LtiE 111011-6I 'CT 1.10E1 A/X

(ZL') ***99' ***E8' ***5Z' 90' 170' - 170' - 80' 170' **9I' *I I' - **ZI' 89' 08'E mat? 'ZI -17 'DV A/X

(L9') ***L6' ***ZZ' 90' 00' - 80' - 90' *I I' 60' **El' *II' 5.17' 5T'£ 1.1-1711 'II 0 - ET 1W A/X

Z ua (8L') ***17Z' 90' ZO.- LO' - LO' *01' *Z1' **£T' - **Z1 . Lb' OE'E Dual? '01 F. - LT 11y A/X

X 7, g 0 44

(cc') 00' - *II'

***PC

*171 .-

**/2'

*OE'

L0'

co' -

LO'

***LT'

LO'

ZO'

**bir

LO' -

90' -

£0'

10'

ZO'

***EZ'

10'

***OZ .-

6Z'

6L'

I8'

b17'

tiZ . I

68'1

aidold u! q uud RupeaN ams!al sa!now ainsIal

'6

•8

. L

Z - *ff - 00' ZO' - 60' *ZI .- £8' 95'1 Al onisIal '9 (9 L') 10' - 170' 90' - ZO . 19' £1717 pool Jsrd 'S

***K . 90' ***817' Z9 - Z 0 I' E 0.1111.12j, '17

**Pt' Mgt. LZ . 8Z 6C55 (000$) 'E /Cirrus

£0.- 05' ITV I Xn 7 - T9 .9 OE'8Z 05V 'I

ET ZI TT OT 6 8 L 9 as JAI apiemuA

suopeia.uoa pue ‘sjuaipwaoa SppciegaH ‘sapspels anudipasaq z mei

Num

ber 2 Su

mm

er 2

008


Table 3 Factor Analysis of Theory X and Theory Y Attitude Items

# Description of Items Loading 1. Most employees can't be trusted. .72 2. Most employees will not exercise self-control and

self-motivation — managers must do this for them. .61

3. Most people are lazy and don't want to work. .60 4. Most employees have little ambition. .57

Eigenvalue 2.16 Percent of explained variance 54.10

Note. These four items resulted from two consecutive principal axis factor analyses with suppression of coefficients less than .50.

tor. The items and loadings are shown in Table 3. Replicating this analysis with an oblique (Oblimin) rather than a varimax rotation re-sulted in the same four items loading on a single factor.

We used these four items to form a shortened measure of managers' un-derlying assumptions about their em-ployees. We next examined internal consistency reliability estimates for all the measures in the present research, and reviewed evidence pertinent to the convergent, substantive, and dis-criminant validity of response scores using the new four-item Theory X/Y measure.

Reliability

Table 2 presents internal consis-tency reliability estimates (Cronbach a) on the diagonal. Alpha reliabilities ranged from .55 to .78. The 17-item X/Y attitude scale, the Fast Food Opinion Scale, the new four-item X/ Y scale, the 19-item X/Y behavior scale, and the 13-item X/Y attitude scale (the 17 X/Y attitudinal items ex-

cluding those in the new four-item X/Y scale) had the highest reliability estimates (coefficient a = .78, .76, .72, .71, and .67, respectively). Four scales showed internal consistency es-timates that exceeded Nunnally's (1978) .70 benchmark. The (five-item) Faith in People scale had rela-tively low internal consistency relia-bility (a = .55), but adjusting for scale length (per the Spearman-Brown prophecy formula), alpha would have been .74.

Convergent, Substantive, and Discriminant Validity

Correlations among the two meas-ures of Theory X/Theory Y atti-tudes—viz., the new four-item X/Y scale and the remaining 13 X/Y atti-tude items—and the other key varia-bles comprising the theorized nom-ological network are summarized in Table 4. The four-item X/Y scale was seen as conceptually identical to the 13-item X/Y scale, as closely related to the 19-item X/Y behavior scale, as distally related to the more generic



Table 4 Correlations among Closely and Distally Related Constructs

Four-item 13-item X/Y X/Y

Variables Attitudes Attitudes Same Construce .66*** Closely Related Construct (Behaviors) b .51*** .38*** Distally Related Construct (Faith) e .25*** .22*** Unrelated Constructsd -.01 -.01 Note. aFour-item X/Y correlated with 13-item X/Y-13; bfour-item

X/Y-4 and 13-item X/Y correlated with X/Y Behaviors; ccoffelated with Faith in People; dmean of correlations with the fast food opinion scale and the three leisure time pursuit items (r-to-z transformations). ***p < .001, two-tailed.

Faith in People scale, and unrelated to the Fast Food Opinion scale and the three leisure pursuit items.

A strong relationship was found be-tween the new four-item measure and the 13-item X/Y attitude scale (r = .66), and this association exceeded the correlation between the new four-item measure and the closely related construct of Theory X/Y behaviors (r = .51). Notably, the four-item X/Y at-titude measure was more highly re-lated to the 19-item X/Y behavior scale than was the 13-item X/Y atti-tude scale ( r = .38). Both the four-item and the 13-item X/Y measures were moderately related to the dis-tally related construct of Faith in Peo-ple (rs = .25 and .22, respectively). As anticipated, mean correlations be-tween the four- and 13-item X/Y at-titude measures with the four concep-tually unrelated measures (fast food attitude, movies attended, hours spent watching television, and hours spend reading for pleasure) were

quite low at r = -.01 and r = -.01, re-spectively. Further, neither the four-nor 13-item X/Y attitude measure was sizably related to the four biographic variables measured: age, sex, salary, and tenure. Examining correlations using absolute numbers (because the coding of sex was arbitrary), rs ranged from .04 to .16 with the four-item scale and from .09 to .13 with the 13-item pool. Overall, therefore, the pattern of associations is supportive of the theorized nomological net-work.

We systematically controlled for the potential confounding effects of biographic variables by performing a hierarchical regression. Entering age, sex, salary, and tenure in Step 1, we regressed in Step 2 the four-item X/ Y measure on the six dependent var-iables comprising the nomological network (Model 1). We also per-formed this analysis by regressing the 13-item X/Y measure in Step 2 on the same dependent variables (Model 2).

JOURNAL OF MANAGERIAL ISSUES Vol. )0C Number 2 Summer 2008


As shown in Table 5, the Beta coeffi-cient for the four-item X/Y scale when regressed on the 13-item X/Y scales was .65 (almost identical to the simple bivariate correlation of .66). Similarly, Beta coefficients for associ-ations between the four- and 13-item X/Y measures and the other six vari-ables in the nomological net differed from correlations on average by about .03. Further, the significance of Step 2 (with the four-item measure as the independent variable) paralleled the theorized network of relation-ships: F = 60.25 (same construct), F = 25.04 (closely related construct), F = 8.19 (distally-related construct), and for the four unrelated constructs, F ranged from .94 to 2.65.

We also performed a post hoc anal-ysis to see if X/Y attitude scales were related to industry, along the lines of the prior research on occupation and scores on the Faith in People scale. We compared 216 participants work-ing in more caring industries (health-care, travel, non-profit, public utility, and government) to 233 who worked in financial services, retail, and mis-cellaneous professional services (which included accountants, con-sultants, and attorneys). We found no significant difference in X/Y atti-tudes (for 17-item X/Y attitude scale ( t(447) = .30, p = . 77, d = .03); how-ever, we did replicate the earlier find-ing that the former had greater faith in people (t(446) = 2.84, p < .01, d = .27) .

Further, we sought to examine the generalizability of our results by per-forming our EFA analyses separately for respondents with less than three years of tenure on their current job and those with three or more years of job tenure. Using a cut-point of three years is meaningful because the US Bureau of Labor Statistics (US De-

partment of Labor, 2006) reports that the median tenure of wage and salary workers in the private sector in 2006 was 3.6 years. The EFA results for the two subgroups yielded very similar four-item scales, and they would have been identical except one item loaded at .49 after the first iteration and was dropped from the second EFA analysis, with the minimum load-ing for retention being .50.

The present evidence suggests that the new four-item measure of Theory X/Y assumptions/attitudes is psycho-metrically sound and reasonably con-struct valid. The scale taps most of the central concepts pertinent to Theory X/Y—viz., whether employees are lazy, are trustworthy, are capable of self-control and self-motivation, and have ambition. Accordingly, it would seem appropriate to use this measure in the conduct of substantive re-search regarding relationships be-tween individual differences in X/Y assumptions/attitudes and variables related to human behavior in organ-izations.

DISCUSSION AND CONCLUSION

Summarizing results, we have de-scribed the development of a new four-item Theory X/Y attitude mea-sure and presented construct validity evidence. The measure is content valid, has adequate reliability, and be-haves as postulated with respect to a theorized nomological network.

However, there are a number of limitations and areas for future re-search that need to be addressed. First, the present validation evidence was provided primarily by employees who also happened to be students. Al-though we found very similar results upon splitting our sample based on years of work experience, it would be


t- C)1

Table 5 Hierarchical Regression of Four-item and 13-itemTheory X/Y Attitude Scales

on Closely and Distally-related Constructs

Two Models with Dependent Variables - Constructs Four-item and 13-item Same Closely Distally Four Unrelated Constructs X/Y Attitude Scales as the X/Y X/Y Faith Fast

rri Independent Variables Attitude Behavior in Food Leisure- Leisure- Leisure- in the Second Steps 13-item 19-item People Opinion TV Movies Reading

(A- u) Modell Theory X/Y Attitude 4-item

c.)

/3 (second step) .65*** .47*** .22*** .07 -.03 .01 .06

o AR 2 (second step) AF (second step)

.40 271.47***

.21 106.37**

.05 19.61***

.01 1.93

.00

.16 .00 .01

.00

.88 Total R2 (first step) .04 .04 .05 .00 .03 .05 .03 Total R2 (second step) .44 .25 .10 .01 .03 .05 .03

f'D

Total F (second step) 60.25*** 25.04*** 8.19*** .94 1.37 2.65* 1.25

IN7 Model 2 C/) Theory X/Y Attitude 13-item

/3 (second step) .37*** .27*** .04 -.09 .07 .05 1 AR 2 (second step) .13 .07 .00 .01 .01 .00

AF (second step) 59.18*** 26.62*** .46 2.17 1.35 .71 Total R2 (first step) .04 .05 .01 .02 .04 .03 Total R2 (second step) .17 .12 .01 .03 .05 .03 Total F (second step) 15.24*** 10.29* .65 1.78 2.93* 1.84 Note. In each case, age, gender, salary, and tenure were added in Step 1. /is of age, gender, salary, and tenure are not

shown for space reasons. The four-item Theory X/Y attitudes and the 13-item Theory X/Y attitude scales were added in the second step in separate analyses. *p < .05, two-tailed; **p < .01, two-tailed; ***p < .001, two-tailed.

KOPELMAN, PRCYITAS AND DAVIS 267

desirable to examine data drawn di-rectly from a field setting. Second, part of the construct validation pro-cess should include the examination of substantive results. We would have increased confidence in the validity of our measure if we had collected data showing that work groups led by Theory Y managers had higher levels of employee creativity, and perhaps even superior levels of work-unit per-formance. Third, we view the new four-item Theory X/Y attitude mea-sure as a start, and not the "final word" in terms of instrument devel-opment. Our pool of 17 items was comprised of far more Theory X (13) than Theory Y statements (four) . This may have contributed to the four-item Theory X/Y scale being com-prised solely of Theory X statements. Accordingly, future research, draw-ing on an expanded and more evenly balanced set of Theory X/Y state-ments, might yield a different, possi-bly multidimensional measure. This would empirically address one re-viewer's suggestion that the four-item scale might be alternatively labeled a Theory X scale. Future research should also attempt to tease out ge-neric versus personal attitudes (i.e., towards "employees in general" ver-sus "me as an employee"). Perhaps the present research may spur the un-dertaking of additional construct val-idation research.

Yet, as we noted earlier, the paucity of substantive research on the effects of Theory Y managerial assumptions/ attitudes may be attributed to the ab-sence of a construct valid measure that is freely available to researchers. How can McGregor's theory be tested if the focal construct has essentially gone unmeasured? Furthermore, in-terventions consistent with Theory Y attitudes, such as participative lead-

ership, should not be viewed as prox-ies for measuring managerial atti-tudes. Yet the key issue that seemingly has eluded most management schol-ars, even to this day, is that Theory Y pertains to an individual difference variable reflecting assumptions about people at work—it is not a specific set of recommended management prac-tices. For example, in his book review of Douglas McGregor, Revisited, Jacobs called the authors—Heil, Bennis, and Stephens—to task for balking "at in-volving workers to the degree con-templated in the Scanlon plans," rather instead endorsing "the diluted tonic of open book management as an acceptable substitute" (2004: 295).

There are many fascinating sub-stantive questions that can be re-searched now that the more funda-mental task of construct validation has been initially addressed. We list a few below.

Coaching and Development. Mc-Gregor (1966) asserted that manage-rial attitudes reflect deep-seated (and possibly unconscious) beliefs; simi-larly, Locke (2003) observed that the Pygmalion effect does not operate consciously and that leaders deny that they treat different people differ-ently. Perhaps this partially accounts for the difficulty Eden et al. (2000) en-countered in using one- to three-day workshops to "train" managers to adopt successfully the Pygmalion Leadership Style. Along these lines, Heil et al. wrote: "Douglas Mc-Gregor's most important legacy was neither Theory X nor Theory Y. It was his insistence that managers question their core assumptions about human nature. . . ." (2000: 20). Thus, our measure of Theory X/Y attitudes might serve as a self-administered di-agnostic tool that enables managers



to achieve greater self-awareness of their attitudes and assumptions about managing people at work.

There has also been general agree-ment among both academics and practitioners that a new social/psy-chological contract has been emerg-ing—one that emphasizes new em-ployer and employee responsibilities. Employers are now expected to pro-vide training, education, and skill de-velopment opportunities, involve em-ployees in decision making, and foster challenging and stimulating work op-portunities; and employees are now re-sponsible for developing their own ca-reers, taking initiative, and participating in organizational deci-sion making (Boswell et aL, 2001). From this perspective, the new em-ployment relationship assumes a The-ory Y view with respect to what em-ployees are willing and able to contribute to the organization, with corresponding employer responsibili-ties.

Boundary Conditions. Are there boundary conditions that moderate the efficacy of Theory Y managerial attitudes? Does organizational cli-mate serve as one such boundary con-dition; for example, will a manager with Theory Y inclinations be less suc-cessful in a command-and-control type of environment? Are Sutton and Woodman (1989) correct in their conjecture that a Theory Y manage-rial style will be more effective where the work entails challenge and uncer-tainty? Will employee expectations about how they should be managed moderate the effectiveness of Theory Y assumptions/attitudes?

Relatedly, there appears to be an increasing tendency for modern firms to adopt organic structures with participative involvement-oriented cultures, empowered self-managed

teams, and managers serving as "coaches" or "facilitators." This is in contrast to traditional mechanistic structures with control-oriented cul-tures emphasizing managerial com-mand and control (Stevens and Ash, 2001). Implicitly, the decision to em-power workers and to assign corre-sponding managerial behaviors as-sumes a TheoryYmindset, whereas the traditional manner of organizing work could be seen as more Theory X.

It is possible that it may be an over-simplification to categorize a man-ager as having either a Theory X or Theory Y mindset. According to leader-member exchange theory (Dansereau et aL, 1975), managers have different types of relationships with subordinates who are in-group members versus those who are out-group members. Furthermore, Campbell and Swift (2006) found ev-idence that managers differ in whether they make internal or exter-nal attributions for good and bad per-formance, depending on whether the subordinate is an in-group member or not. It is possible that managers have a Theory Y mindset with respect to in-group members and a Theory X mindset with respect to out-group members and engage in correspond-ingly different managerial behaviors.

Consequences. Most fundamentally, McGregor's theorizing about the ef-fects of managerial assumptions has not been rigorously examined. Using field data, the hypothesis that work units led by managers with a Theory Y orientation will be generally more effective could be tested—of course, with performance data at the work-unit level. Other theoretical issues might also be researched—e.g., how enduring are the effects of a Theory Y orientation? Along these lines, Liden,



Wayne and Stilwell (1993) found that initial managerial expectations influ-enced leader-member exchanges, but the effects on performance dissipated after six months.

There is no shortage of books on leadership. Taylor (2004) reported that Amazon listed 59,366 book tides under the heading "leadership." Yet a survey of 40,000 workers from 350 organizations found thousands of ex-amples of poor leadership (Taylor, 2004). Sample comments included: "They [supervisors] treat us like

criminals or as if we're on parole—treating us like we mean nothing and they have no problem getting rid of us." "This is a very negative, control place. . . ." "I have a manager that doesn't listen and simply wants to be the boss." Perhaps if more managers operated according to Theory Y as-sumptions, Dilbert's day-to-day expe-riences with the "pointy-haired boss" would be of less interest. But before we can test McGregor's Theory Y we must be able to measure the focal construct.

APPENDIX Items Measuring Theory X/Y Attitudes and Assumptions

I. Most people will try to do as little work as possible. 2. For most people, work is as natural as play or recreation. 3. Most employees must be closely supervised to get them to perform up to

expectations. 4. Most employees actually prefer to be told exactly what to do rather than

having to figure it out for themselves. 5. Most employees do not care much about the organization's goals. 6. Most employees would prefer increased responsibility to increased job se-

curity. 7. Most people will not use their own initiative or do things that they have not

been specifically assigned to do. 8. Employees generally do not have much to contribute when asked to partic-

ipate in making decisions or solving problems. 9. It is just basic human nature — people just naturally dislike work.

10. Most employees will not exercise self-control and self-motivation — manag-ers must do this for them.

11. Most employees have little ambition. 12. Most people do want responsibility. 13. Most employees prefer to have someone else set their goals and objectives. 14. Most people work to eat and pay their bills rather than because they need

to solve problems and be creative. 15. Most employees prefer supervising themselves rather than close supervision. 16. Most people are lazy and don't want to work. 17. Most employees can't be trusted.

References

Bandura, A. and E. A. Locke. 2003. "Negative Self-efficacy and Goal Effects Re-visited." Journal of Applied Psychology 88: 87-99.



Baron, R. A. and P. B. Paulus. 1991. Understanding Human Relations: A Pratical Guide to People at Work. New York, NY: Allyn and Bacon.

Boswell, W. R., L. M. Moynihan, M. V. Roehling and M. A. Cavanaugh. 2001. "Responsibilities in the 'New Employment Relationship': An Empirical Test of an Assumed Phenomenon." Journal of Managerial Issues 13: 307-327.

Campbell, C. R. and C. 0. Swift. 2006. "Attributional Comparisons Across Biases and Leader-Member Exchange Status." Journal of Managerial Issues 18: 393-408.

Chapman, A. 2005. "X-Y Theory Questionnaire." In Organizational Behavior in Health Care. Ed. N. Borkowski. Boston, MA: Jones and Bartlett.

Costley, D. L. and R. Todd. 1987. Human Relations in Organizations (3rd ed.). St. Paul, MN: West Publishing Company.

Dansereau, F., G. Graen and W. Haga. 1975. "A Vertical Dyad Linkage Approach to Leadership within Formal Organizations." Organizational Behavior and Hu-man Performance 13: 46-78.

Eden, D. 1990. Pygmalion in Management. Lexington, MA: Lexington Books. D. Geller, A. Gewirtz, I. I. Gordon-Terner, M. Liberman, Y. Pass, I.

Salomon-Segev and M. Shalit. 2000. "Implanting Pygmalion Leadership Style through Workshop Training: Seven Field Experiments." Leadership Quarterly 11: 171-210.

Fiman, B. G. 1973. "An Investigation of the Relationships among Supervisory Attitudes, Behaviors, and Outputs: An Examination of McGregor's Theory Y." Personnel Psychology 26: 95-105.

Gordon, J. R. 1999. Organizational Behavior: A Diagnostic Approach. Upper Saddle River, NJ: Prentice Hall.

Greenberg, J. 1999. Managing Behavior in Organizations (2nd ed.). Upper Saddle River, NJ: Prentice Hall.

Hair, J. F., R. E. Anderson, R. L. Tatham and W. C. Black. 1998. Multivariate Data Analysis. Upper Saddle River, NJ: Prentice Hall.

Heil, G., W. Bennis and D. C. Stephens. 2000. Douglas McGregor, Revisited: Man- aging the Human Side of the Enterprise. New York, NY: John Wiley and Sons.

Jacobs, D. 2004. "Book Review Essay: Douglas McGregor—The Human Side of Enterprise in Peril." Academy of Management Review 29: 293-311.

Joreskog, K. G. and D. Sorbom. 2002. LISREL 8.53. Chicago, IL: Scientific Soft-ware International, Inc.

Kelly, K. R. and W. Lee. 2002. "Mapping the Domain of Career Decision Prob-lems." Journal of Vocational Behavior 61: 302-326.

Liden, R. C., S. J. Wayne and D. Stilwell. 1993. "A Longitudinal Study on the Early Development of Leader-Member Exchanges." Journal of Applied Psychol-ogy 78: 662-674.

Locke, E. A. 2003. "Good Definitions: The Epistemological Foundation of Sci-entific Progress." In Organizational Behavior: The State of the Science (2nd ed.). Ed. J. Greenberg. Mahwah, NJ: Lawrence Erlbaum Associates.

McGregor, D. M. 1967. The Professional Manager. Eds. W. G. Bennis and C. McGregor. New York, NY: McGraw-Hill. 1966. Leadership and Motivation. Cambridge, MA: MIT Press. 1960/1985. The Human Side of Enterprise. New York, NY: McGraw-Hill.


View publication statsView publication stats

https://www.researchgate.net/publication/280114317

douglas mcgregor's theory x and y: toward a construct ... · ries found that theory x/y was...

Documents