Need assignment done

IM7.VU, AMOoMMLtab 4blO/t7/lliO.7S

Situational Leadership Theory: An Examination of a Prescriptive Theory

Robert P. Vecchio University of Notre Dame

In a study of 303 teachers representing 14 high schools, measures were taken of supervisory style (consideration and initiating structure), follower maturity, performaiioe, satisfaction with supervi- sion, and quality of leader-member exchange. A variety of statistical tests were conducted to test the prescriptions for effective supervision contained in Situational Leadership Theory (HerMy A Blanchard, 1982). Results suggest that the theory may hold only for certain types of employees. Specifically, the results imply that more recently hired employees may need and appreciate greater task structuring from their superior. These results have implications for reinterpreting the theory and examining it within the "substitutes for leadership" perspective (Kerr A Jermier, 1978).

Hersey and Blanchard's (1982) Situational Leadership The- ory (SLT) embodies one ofthe more widely known and, at the same time, least researched views of managerial effectiveness. Throughout this article, the term managerial will be used inter- changeably with the term leadership as SLT focuses on the effectiveness of nominal heads rather than on emergent or incre- mental (anas of power and influence (i.e., leadership per se). As noted by Graeff( 1983), the theory is often cited in academically oriented management textbooks. However, we can offer'little advice to our students, or to practicing managers, on the utility of the theory. Before we can endorse or critique the theory to our constituencies, a rigorous test ofthe theory's propositions is, of course, required. In this article, the origins and central elements of the theory, the available evidence of the theory's validity, and the requirements for a rigorous test ofthe theory's propositions are considered.

Origins and Elements o f Situational Leadership Theory

Situational Leadership Theory developed from the writings of Reddin (1967). Reddin's 3-Dimensional Management Style Theory posits the importance of a manager's relationship orien- tation and task orientation in conjunction with effectiveness.

. From the interplay of these dimensions, Reddin proposed a ty- p(4ogy of management styles (e.g., the autocrat, the missionary, the deserter). Although Reddin suggested that his framework explained effectiveness as a function of matching style to situa- tion, his approach did not identify specific situational attributes that could be explicitly incorporated into a predictive scheme.

Building on Reddin's (1967) suggestion that leader or man- ager effectiveness varies according to style, Hersey and Blanch-

Tbe author expresses his appreciation to Cindy Harrison, who helped to ensure the coqmation of all participants.

Correspondence conoeniing this article should be addressed to Rob- ert P. ^todlio, Department of Management, University oTNotie Dame, Notre Dame, Indiana 46SS6.

ard (1969) proposed a life-cycle theory of leadership. According to life-cycle theory, degrees of task orientation and relationship orientation must be examined in conjunction with the dimen- sion of follower maturity to account for leader effectiveness. The central precept of life-cycle theory (1969, p. 29) is that as the level of follower maturity increases, effective leader behavior will involve less structuring (task orientation) and less socio- emotional support (relationship orientation). Howevei; the de- cline in need for both of these leader behaviors is not straight- forward. During the early stages of an employee's tenure, a low level of relationship orientation coupled with high task menta- tion is considered to be ideal. As an emjdoyee (or group of em- ployees of roughly equal nuturity) gains in maturity, the need for supervisory social-emotional suiqxMi increases, while the need for structuring declines. Beyond a certain level of matu- rity, the need far both social-emotional support and structur- ing declines. At the highest levels of employee maturity, super- visory task and social behaviors became superfluous to effective employee perfonnance.

In a popular text (evidenced by its being in its 4th edition), Hersey and Blanchard (1982) attempted to provide still greater precision to these precepts. They suggested that fc^ower matu- rity can be broken into benchmark categories of high, moder- ate, and low, and that appropriate leader style can be summa- rized in terms of a leader primarily telling, selling, participat- ing, or delegating in relations with subordinates. This most recent statement ofthe Heney-Blanchard SLT model (1982, p. 152) is summarized in Figure 1.

Evidence ofthe Model's

At a purely theoretical level, SLT has been suggested as hav- ing a good deal of avedap with other popular views of leader and group behavior. In a comparison (rf'their views with those offered by other perspectives, Herxy and Blanchard (1982, ch^>. 13) achieved a synthesis of their concepts with those oon- tained in McGregof's (1960) Theory X and Y, Aisyris's (1957) maturity-immaturity continuum, Ukert's (1967) maniigBiiieat

SITUATIONAL LEADFRSHIP THEORY 445

STYLE OF LEADER High Task

and Kigh

Reiatxyiship

HiQh Reiattonship

and Low Task

Low Retatonshtp

and Low Task

I LOW)

HIGH

TASK BEHAVIOR

MOOERATE

• (HIGH)

LOW

M3 M2 M1 MATURrTY OF FOLLOWER(S)

S i t u a t i o n a l l e a d e r s h i p

Figure I. The situational leadership model. (From Managtng Organiza- tional Behavior [p. 152] by P. Hersey and K.. Blanchard. 1982. Engle- wood Qiffs, NJ: Prentice Hall. Copyright 1982 by Prentice-Hall. Inc. Reprinted by permission.)

systems, Maslow's (1954) need hierarchy, Herzberg's (1966) two-factor theory, McClelland's (1961) achievement theory, Schein's (1970) assumptions of human nature, transactional analysis (Berne, 1964; Harris 1969), French and Raven s (1959) power bases, parent effectiveness training concepts (Gordon, 1970), Greiner's (1972) phases of organizational growth, Lew- in's (1947) views of achieving behavioral change, behavior mod- ification (Skinner, 1953), and force field analysis (Lewin, 1947). Although one can cynically argue that this high degree of over- lap suggests that SLT is not saying much that is new or original, it can also be contended that many of the above theories can also be shown to contain a high degree of overlap. More posi- tively, one can argue that SLT is focusing on critical features of behavior that have been previously identified.

The fact that SLT can be shown to overlap to varying degrees with other theories is not in itself sufficient evidence of SLT's validity. This point is of clear importance when one considers that many ofthe previously cited theories have not achieved a high degree of empirical support (e.g., Herzberg's two-factor theory), despite a fair amount of attention in the academic liter- ature. In terms of internal theoretical coherence, Graeff (1983)

has provided the most comprehensive critique ofthe theory. In his review of SLT, Graeff argued that the theory may actually have been derived from a passage in an article hy Korman (1966), in which the suggestion of curvilinear relationships be- tween dimensions of leader behavior and other variables was taken to mean that a curvilinear relationship may exist between dimensions of leader behavior In addition, Graeff (1983) sug- gested that the manner in which components of dimensions in SLT are combined and the manner of graphic presentation ofa four-dimensional model (task orientation, relationship orienta- tion, follower maturity, and effectiveness) in only two dimen- sions are critical problems for the theory. Also, he has suggested that the popularly advocated measurement device for studying leader behaviors (the LEAD instrument) possesses unknown psychometric qualities. In the theory's favor, Graeff argued, however, that Sl T correctly focuses on issues of leader flexibility and the importance of subordinate attributes as the key situa- tional determinant of appropriate leader behavior.

At an empirical level, the theory has received little attention. One ofthe earliest published studies devoted to SLT concepts focused on the development ofa measure of follower maturity, yet It did not use the measure m a test of the model (Moore, 1976). In a study that approximated a test ofthe theory, Ham- bleton and Gumpert (1982) asked managers to select at random 4 of their subordinates to complete a survey instrument. For 65 participating managers (of 159 who were contacted), manager ratings of subordinate maturity were coded in conjunction with manager self-assessments of leadership style (high vs. low task and relationship orientation). From this coding, matches and mismatches were identified. Matches occurred in only 29% of the cases. A comparison of mean performance ratings—given for each employee by the managers—for the matches and mis- matches revealed that the matches received somewhat higher mean evaluations (/ = 6.47,;? < .01).

Although the findings of Hambleton and Gumpert (1982) are the only available supportive evidence for the model, they raise several concerns. First, the sample suffered severe attrition (i.e., less than half of the managers provided data). Second, the man- agers provided self-assessments of their own style. Such self-as- sessments of leader behavior are not regarded as being highly accurate (Schriesheim & Kerr. 1974). In addition, these assess- ments were taken on a version ofthe LEAD instrument, rather than on a more widely studied and accepted measure of leader behavior Last, the respondents were highly cognizant of SLT precepts (as evidenced by their having been asked to rate their knowledge of SLT and to assess the extent to which they used SLT in their work). This awareness of SLT principles on the part of the participants may have induced some respondents to attempt to complete their surveys in conformity with the theory (i.e., in order to appear to be applying their knowledge ofthe theory). Also, all ofthe respondents reported at least fair knowl- edge of and some use of SLT.

The most recently reported study of SLT (Blank, Weitzel, & Green, 1986) involved 27 hall directors and 353 resident advi- sors (subordinates) at two large universities. Respondents com- pleted the Leader Behavior Description Questionnaire (LBDQ- XIl; Stogdill & Coons, 1957) and a measure of maturity. Direc- tors provided performance ratings of resident advisors, and

446 ROBERT P VECCHIO

each resident advisor completed subscales of the Job Descrip- ticm Index (JDI) satisfaction measure (Smith, Kendall, & Hulin, 1%9). In their analysis of these daU, Blank, Weitzel, and Green did not report the results of noatching subordinate maturity and leader behaviors to predict subordinate performance and satis- faction, but instead examined interactions between maturity and each of the two leader behavior dimensions (consideration and structuring) in an attempt to predict subordinate perfor- mance and satisfaction. In general, their search for two-way in- teractions did not reveal support for the theory. However, it should be noted that the theory predicts a three-way interac- tion, and not separate two-way interactions, among the key variables (i.e., the interaction of maturity with structuring should not be examined independently of consideration, but jointly).

In summary, investigations of the theoretical and empirical robustness of SLT have been rare. Although the theory contains strong intuitive appeal, the veracity of the theory has not been assessed via a rigorous empirical test.

Issues Surrounding a Test o f the Situational Leadership Theory

In order to test SLT, several issues must be addressed that relate to the clarity of the theory's prediction. One major issue surrounds the unit of analysis for SLT: the individual versus the group. Although the theory is often stated in terms of group maturity, there are also many references to individual maturity as well. For example, in their definition of maturity, Hersey and Blanchard (1982, p. 151) stated that an "individual or a group" is their focus. They also recognized that when one relates to an entire group (e.g., a teacher speaking to a class of students), it is the maturity level of the group that is important. However, when one deals with an employee in a one-to-one setting (e.g., a teacher speaking with a single student), the maturity level of the individtial is most important. This recognition of the need for leaders to behave differently with individual group members than when they relate to an entire group is an important state- ment. Much of the recent research in the area of leadership has, in fact, focused on the issue of universal versus differential lead- ership st^e (cf. research on the Vertical Dyad Linkage Model; Liden & Graen, 1980). That the dynamics of SLT are presumed to operate at both levels is an important feature of the SLT

• framework. In the context of a test of SLT, it is necessary to specify and be consistent in studying leadership phenomena at a given level (and not across I^^Is) of analysis. It seems likely— in light of the preponderance of research at the individual level and the suspicion that group processes may mask individtial process—that SLT will be most robust at the individual level of social dynamics (i.e., leadership behavior that is in accord with the prescriptions of SLT will be more effective when it is tar- geted to a given individual's level of maturity).

An extended issue that is beyond the present investigation is whether individual maturity interacts with group maturity in determining leader effectiveness. It is easy to envision a situa- tion in which a subordinate is significantly more mature than his or her peers in a given position, yet the leader's behaviors (if they are often displayed for the benefit of the group) may be

grossly inappropriate. In such settings, the inoongruent subor- dinate will likely be dissatisfied with the leader's directions and may be resentful of the limits that his peers indirectly set via the leader's actions.

Pfcrhaps the least clear feature of SLT surrounds the defini- tion of effectiveness. In their book, Hersey and Blanchard (1982, pp. 96-99) define effectiveness and ineffectiveness as oc- curring when a leader's style is appropriate and inappropriate, respectively. Although they recognize that effectiveness can be viewed as a continuum, they do not acknowledge the multifac- eted nature of the concept. Also, the definition of effectiveness in terms of appropriateness of leader style is somewhat circtilar in its use of logic. In order to test SLT, it is useful to define effectiveness in broader, more traditional terms. For example, effectiveness can be defined in terms of output, cost reduction, enhancement of employee motivation, morale, and so forth. This restatement of effectiveness implies that SLT should be restated so that effectiveness is a possible outcome of appropri- ateness of leader behavior. The use of the word possible is im- portant in the foregoing sentence because an appropriate com- bination of leader style (in terms of SLTs prescription) may still not enhance subordinate behavior and attitudes. As noted by Kerr and Jermier (1978), situational attributes can o ^ e t and substitute for leader behaviors. In a sense, subordinate maturity in SLT is a substitute for leadership, in that subordinates of higher maturity need less attention or direction from leaders. Increases in subordinate self-sufficiency (maturity), which likely result from relevant work experience and training, can make leader behaviors increasingly irrelevant to subordinate performance and morale. Therefore, SLTs predictiim that highly mature employees require a low-structure-low-consider- ation style of supervision may be partially misstated. It may be more correct to say that supervisory style is comparatively more irrelevant, in terms of its impact on highly mature subordi- nates. In short, the conduct of highly mattire subordinates may simply be less predictable than that of other employees, from supervisory attributes.

A further issue centers on how to test in a statistical sense, the predictive accuracy of SLT. At its heart, the theory forecasts a three-w^ interaction of leader consideration, leader structtir- ing, and subordinate maturity. If one imagines the form of this hypothesized interaction by trying to graph the hyperplane that is predicted, it becomes apparent that the predicted interaction does not satisfy the statistical assumption of hcmioscedasdty. That is to say, the regression-based assumption of equal vari- ance around the regression plane does not hdd, by definition, for SLT as the predictions of superior performance only hold for specific points in the multidimensional space. In aU other locations in the space, the data are free to vary. Theref<xe, the theory only makes predictions for specific combinations of vari- ables. For all other combinations, the theory is silent. In essence, a test of the theory that uses the statistical technique of multiple r ^ e s s i o n may lead, erroneously, to the conclusion that the the- ory is incorrect for a given data set (i.e., a IVpe II error). To more fairly test the theory, it would be wrathwhile to examine the predictions in the manner proposed by the theory's devdap- ers (i.e., to compare the effectiveness oileadtcn whose styles are "appropriate" for given settings to leaders whose styles are pn-

SITUATIONAL LEADERSHIP THEORY 447

dieted to be "inappropriate" for the same settings). Although one may test whether a given data set violates the requirement of homoscedascity, the results of such a test would not, of course, shed light on the validity of Situational Leadership The- ory, perse.

The purpose of the present investigation was (a) to test SLT in a study that was designed to capture the critical variables proposed by the theory and (b) to explore SLT with analytic techniques that reflect traditional practice in organizational re- search (i.e., regression) as well as accommodate the need to study leadership as a situation-specific phenomena (i.e., sub- grouping analysis).

Method

Subjects and Procedure

Subjects were 303 full-time high school teachers, who represented 14 high schools in a large midwestem city and provided data in response to a coofidential survey. Because of the support of the head of the school district, cooperatioa was readily obtained from the principals of all 14 schools in the district. During monthly meetings with faculty, the prin- cipals distributed surveys to their teachers. At the meetings, time was devoted to the completion of the surveys. Both teachers and principals completed similar surveys, which focused on the behavior of the school principal (leader) and the individual teacher (subordinate). To ensure anonymity for the teachers, the surveys were coded with ID numbers. In addition, the completed surveys of both teachers and principals were collected at the meeting, placed in an envelope, and mailed directly to tbe author. With the exception of 34 teachers who did not attend their school's monthly meeting in January 1986, all of the principals and teachers responded to the survey.

Measures

Each teacher provided responses to the following scales: (a) JDI, satis- faction with supervision (Smith et al,, 1969); (b) Leader-Member Ex- change, quality of leader-member relationship (Liden &. Graen, 1980); (c) LBDQ-XII, leader consideration (StogdiU & Coons, 19S7); and (d) LBDQ-XIL leader initiating structure (StogdiU A Cooos, 19S7), Modi- 6ed versions of the LBDQ-XII measures of leader behavior were used in place of the LEAD instrument because of the relative psychometric advantages of the LBDQ-XII (i,e,, its reliability and construct validity have received more attention than the LEAD instrument, and it is a more widely accepted index of leader behavior than the LEAD instru- ment). In addition, the stems of the items in the LBDQ-XII used in this study wen modified to incorporate an individualized format (cf, >fecchio & Gobdel, 1984); sample item, "My principal acts without con- sulting me," Furthermore, each teacher was aiked to complete a fol- lower maturity index that contained items related to both task-relevant (e.g., understanding of job requirements) and psychological (e.g., com- mitment) forms of maturity (Hambleton, Blanchard, & Hersey, 1977),

Principals provided ratings for each teacher on dimensions of follower maturity and performance. Maturity was assessed on items related to task-relevant and psychological maturity, whereas performance was as- sested by summing ratings across dimensions of dependability, plan- ning, know-how, present performance, and expected performance (Li- den & Graen, 1980),

Analytic Techniques

The aocuncy of the principles of SLT was examined with several •litiiticti techniques. As will be shown, each technique represents a

somewhat different phrasing of the central research question. The first technique used was hierarchical regression analysis, in which a three-way multiplicative interaction term was created (Maturity X Con- sideration X Structuring), This interaction term was entered into a re- gression equation following the inclusion of main effects and two-way interaction terms to determine whether the inclusion of the three-way interaction term appreciably increased the variance accounted for in the performance criterion. As was noted earlier, the use of multiple re- gression for testing SLT can be critiqued on the grounds that certain assumptions of regression cannot be met when the principles of SLT are, in fact, correct. Nonetheless, the robustness of regression tech- niques and our present uncertainty as to just what would result with the use of the technique warrant the exploratory use of multiple regression to test the theory,

A second approach to testing SLT involves the creation of subgroups of employees for whom the theory is expeaed to hold or not to hold. This requires the creation of subgroups based on the combination of consideration, structuring, and maturity. For employees whose situa- tions are designated "matches," their mean performance should be su- perior to that of subordinates for whom the situations are "mis- matches," This comparison of matches and mismatches represents an omnibus test of SLT in that it ignores differences within specific catego- ries of maturity in favor of an overall test.

It can be argued, however, that an omnibus test is not the best possible device for assessing SLT. If the distribution of cases is not uniform across categories, then the results of the omnibus test may be biased. For example, no (or very few) cases may exist for some combinations of maturity and leader behavior. The peculiarities of these distributions and the possible associated mean differences for the categories could, thereby, produce spurious results, A third, more direct, assessment of SLT involves the creation of several categories on the dimension of fol- lower maturity. After creating these categories, comparisons could then be made within categories to determine whether subordinates who match on the leader dimensions are superior performers, relative to those who do not match on these dimensions. The need to first create cat^ories of maturity is perhaps critical in that different levels of fol- lower maturity are likely to be related to different levels of overall perfor- mance (although SLT does not directly address this critical issue). Therefore, the likely correlation of maturity with performance needs to be controlled by corulucting comparisons within levels of maturity.

To be sure, only the third, partitioned, technique is the most defensi- ble approach to assessing SLT, However, we presently know so little about SLT-related phenomena that all three techniques are reported here in the interest of completeness and in order to gain further under- standing.

Results

2Lero-order correlations among the predictors and criteria (see Table 1) reveal, first, that the variables are all, at least mod- erately, related to one another. Although this is to be expected from a survey-based study, it does not pose a serious obstacle to studying SLT principles in that SLT does not posit main effects, but interactions. Second, the existence of main effects makes it more difficult to identify complex forms of relation- ships because the main effects "steal" criterion variance (i.e., the present test of SLT can be viewed as being conservative in that it will be relatively difficult to uncover SLT effects). Table 1 also displays the internal consistency coefficients for the mea- sures of interest All of these coefficients are of reasonable mag- nitude (ranging from .82 to .94). Evidence that the niaturity measure at least partially taps work-relevant expoience was ob-

448 ROBERT P. VECCHIO

Table 1 Correlations Among Predictors and Criteria

1. 2. 3. 4. 5, 6,

Viable

Structuring* Consideration* Maturity" Performance" Leader-member quality* Satisfaction with supervisor*

M

16,89 18,89 40,26 29,62 21,46 47,03

SD

4,20 3.99 6.34 4.87 4.58

10.97

Range

5-25 5-25

15-48 10-35 7-28 2-57

1

82 52 26 27 60 79

2

83 35 31 79 74

3

— 93 85 34 31

4

—. 94 35 36

5

— 91 79

6

— 90

Note. Coefficient alphas are given on the priniary diagonal. All correlations are significant beyond the ,01 level. Decimal points are omitted, * Dau provided by subordinates," DaU provided by supervisors.

tained by correlating individual maturity with self-reported years of teaching experience. The obtained correlation sug- gested that years of experience was related to the index (r = . 15, /7 < .01). It should also be noted that the range of professional teachitig experience for the sample was substantial (range = 1- 31 years; A / = 2 0 , 5 years).

The results of the hierarchical regression analyses are pre- sented in Table 2 for the criteria of supervisor rating, leader- member exchange, and satisfaction with supervision. For each analysis, the three predictors of consideration, structuring, and maturity were entered simultaneously at the first step. Next, the two-way interaction terms were entered into the equation. Last, the three-way interaction term was included. The increment in R^ at each step (i.e., in variance accounted for) was calctllated and tested for significance (Cohen & Cohen, 197S). As the re- sults in Table 2 indicate, none of the three criteria tested yielded support for a three-way interaction.

Omnibus tests of mean differences on the criteria were also conducted. For these analyses, the distribution of follower ma- turity was trichotomized into high, moderate, and low maturity by cutting at the values of 44 and 40. Three categories, rather than four, of maturity were created to ensure a sufficient num- ber of cases for each subgroup. The predictions for the middle- range groups on maturity are identical (i.e., high consideration coupled with moderate structuring is prescribed). Next, the di- mensions of consideration and structuring were trichotomized and dichotomized, respectively. Cuts on the structuring dimen- sion were made at the values of 19 and 15. On the consideration dimension, the split was made at the value of 19. This resulted in a 3 X 2 cross tabulation, in accord with the SLT model. Em-

ployees whose values on maturity coincided with the prescribed levels of consideration and structuring were designated matches; all remaining employees were designated mismatches. It is, of course, predicted that mean values for the outcome vari- ables will be higher for the matched group, relative to the mis- matched group.

As Table 3 reveals, a large percentage of the employees were in the mismatched group (i.e., their situations were those for which the theory predicts lower effectiveness). This finding, in itself, is of some importance in that it suggests that the positive SLT prescriptions may have little relevance to a majority of em- ployees (i.e., the natural occurrence of the preferred combina- tions may be fairly low). Alternatively, it can be argued that there is a great untapped potential or significant need for creat- ing circumstances that the theory prescribes. To test this later point, mean differences were tested for significance for the matched versus mismatched groups. If the theory is correct, we can expect the matched group (albeit a smaller group) to have higher values on the outcome measures.

As the results reported in Table 3 indicate, the means were in the predicted direction for all three comparisons. In two of these comparisons, the means were significantly different (al- though the estimates of the effects' sizes were not substantial). The results suggest that employees who describe their superiors' behavior on the dimensions of consideration and structuring in accord with SLT prescriptions, given their specific level of maturity, tend to have somewhat higher performance ratings, and to report higher quality relationships with their supervisoi; as well as greater satisfaction with their supervisor.

Although the omnibus tests provided evidence of the accu-

Table 2 Summary of Regression Analyses

Source

Consideration (Q, structuring (S), maturity (M)

C X S . C X M , S X M C X S X M

Performance

R'

.743»*

.744**

.744**

^R'

.001

.000

Leader-member quality

R' A/?̂

.686»*

.697"* .011*

.697»* .000

Satisfaction

/e

.639*»

!641*»

Aje

.012*

.000

*p<.05.**p<.0l.

SITUATIONAL LEADERSHIP THEORY 449

Table 3 Results of Omnibus Tests

Group

Performance Match Mismatch

Leader-member quality

Match Mismatch

Satisfaction with supervision

Match Mismatch

M

30,4 29.5

22,8 21,2

51,0 46,3

SD

3.91 5.03

22,80 21,18

6,25 11,52

n

50 252

50 245

46 238

F

1,61

5,25

7,35

P

,21

,02

,01

Estimated effect size

,002

,014

,022

racy of SLT, they do not tell us precisely where these differences are occurring in the framework. Also, the supportive results of the omnibus tests may have capitalized on unique, sample-spe- cific differences that are correlated with uneven distributions of attributes. Therefore, sets of partitioned tests of the theory were conducted. These tests involved making mean difference com- parisons within maturity groups. That is to say, matches and mismatches were designated within each of the three levels of maturity. The results of these comparisons (see Table 4) indi- cate that the mean differences were in the correct direction in six of nine comparisons. Of these six, four were significantly different. The effect sizes for these differences ranged from ,032 to , 160. In one instance, the mean difference was significant in the reverse direction of that which was hypothesized. Also, the results were more generally supportive in the low-maturity cate- gory, somewhat mixed in the moderate-maturity category, and generally nonsignificant in the high-maturity category. This suggests a very different picture of the results, in which the the- ory is only correct for low-maturity employees and less correct for more mature employees. The implication is that low consid- eration coupled with high structuring is a superior combination for low-maturity employees. For moderate- and especially high- maturity employees, it is not clear that the combinations pre- scribed by SLT are associated with superior outcomes. Further- more, it is worth noting that the mismatched groups (Tables 3 and 4) were somewhat more variable than the matched groups. This is evidenced by the often larger standard deviations for the mismatched groups. This difference in variability confirms the eaiiier suggestion that the more stringent assumptions of para- metric statistical tests (e,g., regression analysis) may not be satisfied by the data generated by—as well as the logic attendant to—tests of Situational Leadership Theory.

Discussion

The present study represents one of the first comprehensive tests of Situational Leadership ThecM .̂ As such, it is not possi- ble to contrast the current findings with those obtained in other investigations. Therefore, the present evidence must be taken, for the moment, as providing the best available test of the theo- ry's principles. In general, the present study provides partial

support for the theory in that the omnibus tests and several of the partitioned tests point to the theory being partially accurate in its prescriptions. The results of the regression analyses are perhaps suspect in that various assumptions of regression anal- ysis are contrary to the essential features of SLT. Furthermore, the predictor variables that were used in the regression analysis were intercorrelated. This may be more than a common source effect problem because the predictor constructs may not be log- ically or empirically independent of one another.

The finding that SLT was most strongly supported in the low- maturity condition appears reasonable in that employees who are relatively lacking in task-relevant knowledge or commit- ment should require more structuring on the part of their super- visor. Displays of considerateness by superiors for low-maturity subordinates would be tantamount to sending improper signals to such subordinates. For subordinates of moderate maturity, it is not clear what style of supervision works best. The present

Table 4 Summary of Partitioned Tests

Group

Performance Match Mismatch

Leader-member quality

Match Mismatch

Satisfaction with supervision

Match Mismatch

Performance Match Mismatch

Leader-member quality

Match Mismatch

Satisfaction with supervision

Match Mismatch

Performance Match Mismatch

Leader-member quality

Match Mismatch

Satisfaction with supervision

Match Mismatch

M

27,3 25,0

24,2 19,1

53,4 41,3

32,1 30,1

19,6 22,1

47,0 48,6

33,4 33,9

24,1 22,7

51,4 49,6

SD n

Low maturity

3,50 4.54

3,06 4.79

3.86 12.68

Moderate

2,37 3,11

3,20 4,54

7,51 9,84

21 92

21 92

19 88

F

4,68

22,18

16.80

maturity

15 78

15 75

13 72

High maturity

2,10 1,70

2,81 3,89

6,15 9,66

14 82

14 80

14 78

5,65

4,18

<1

1,20

1,65

<1

P

,03

,00

,00

,02

,04

,28

,20

Estimated effect size

,032

,160

.129

,048

,034

,002

,007

450 ROBERT P. VECCHIO

data suggest that performance is greater for these same employ- ees if moderate structuring is combined with high consider- ation. Howevei; the same sample provides evidence that the quality of leader-member relationships may be significantly lower when this particular combination of styles is reported.

For high-maturity employees, the theory appears to be un- able to predict. Perhaps the theory needs to be rephrased to accommodate such high-maturity types. As it stands, the theory seems to suggest that highly mature employees can be relatively free from direction and do not need to receive "strokes" from their supervisors. Such a scenario is highly doubtful, as most employees probably appreciate supervisor considerateness and occasional signs of supervisor interest (as manifested by super- visor direction or structuring). In addition, the measurement of maturity poses a unique problem for testing Situational Leader- ship Theory. Self-reports of maturity are highly suspect. Sim- ilarly, peer ratings may largely reflect popularity rather than task orientation. Yet, the construct is so broad that a rating seems the most appropriate technique for addressing it. It is interest- ing that in the present sample, years of teaching was somewhat associated with supervisor ratings of maturity. However, experi- ence can still be regarded as independent of supervisor ratings. This makes sense as long-tenured employees may be, on the av- erage, more competent than are more recently hired individu- als. Yet, the job-relevant maturity of long-tenured employees can still be quite variable (i,e,, years of experience and job-rele- vant maturity are not likely to be highly correlated across situa- tions). Of further interest, the correlation of teachers' self-rat- ings of maturity were significantly correlated with their superi- ors' ratings on the same instrument (r = ,28, p < ,01),

It is possible that SLT is not well suited to making predictions in any given job category, in that a full range of maturity and leader behaviors may not be manifested in one job classifica- tion. Perhaps SLT is better viewed as being prescriptive across job categories. In this view, high to low maturity represents var- ious classes of jobs (e,g., professional to unskilled). For profes- sional (high-maturity) jobs, supervisors should display rela- tively less consideration and less structuring. Professionals, as the reasoning goes, should be capable and desirous of greater self-direction. For unskilled (low-maturity) jobs, supervisors should provide significantly greater structuring and less consid- eration. Unskilled workers perhaps expect, and may prefer, greater direction and less social-emotional attention on the part of supervisors (cf Vroom & Mann, 1960). This restatement of SLT suggests that the underlying principles of SLT may be valid but that the theory may be improperly conceptualized, such that the current focus is on maturity differences within jobs rather than across jobs.

An across-jobs perspective offers possibly greater ranges of maturity and, therefore, a greater likelihood of identifying sys- tematic relationships. However, this across-jobs perspective re- quires a modification of the term maturity. As it is used in a within-job perspective, its meaning is fairiy clear (i.e., employee level of task-relevant knowledge and commitment). In an across-job perspective, maturity may have to be replaced with a more level-appropriate concept such as normative expecta- tions. This across-jobs view suggests that job prestige (or job quality) dictates ^jecific norms for preferred styles of supervi-

sion. To be effective, a supervisor should be conscious of and responsive to these norms. The above viewpoint also can be seen as incorporating notions from Kerr and Jermier's (1979) suggestion of substitutes for leadership. In profesnonal and highly skilled positions, experience and knowledge can make a supervisor's influence less important. In less skilled positions in which employees lack experience and acquired knowledge, supervisory influences can be of far greater importance. How- ever, a testing of an across-jobs perspective should perhaps not be undertaken without further testing of the within-job view of Situational Leadership Theory. In addition, future tests of the theory should, to the degree possible, use independent measures of predictors.

In summary, the present study provided partial support f<v principles contained in Situational Leadership Theory. As is not uncommon in organizational research, an initial, seemingly simple research question yielded a complex set of results that only substantiated a portion of a set of propositions. Further- more, the results underscored the somewhat disheartening ob- servation that the approach taken in analyzing a theory deter- mines, to an extent, the form and degree of support that is ob- tained for the propositions. The present results most strongly suggest that more recently hired employees may require greater structuring from their superior. Nonetheless, the present results are sufficiently intriguing so as to suggest that SLT be stvidied further, with an across-jobs perspective and with a recognition that high-maturity conditions may obviate the need for supervi- sion, rather than specify a particular style of supervision.

References

Argyris, C, (1957), Personality ard organization. New Yorlc Harper A Row,

Berne, E, (1964), Games people play New York; Grove Press, Blank, W,, Wietzel, J,, A Green, S, G, (1986). Situational leadership

theory; A test of underlying assumptions. Proceedings (^the Academy ofMaruigement. 384,

Cohen, J,, & Cohen, P, (1975), Applied multiple regression/correlation aruilysis for the behavioral sciences. Hillsdale, NJ; Eribaum.

French, J. R. P., A Raven, B, (1959), The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150-167). Ann Aiboi; MI; Institute for Social Researeh.

Gordon, T (1970). Parent effectiveness training. New Ybrk; P. H. Mfy- den,

Graeff, C, L, (1983), The situational leadership theory; A critical view. Academy ofManagement Review 8.285-291.

Greiner, L. E. (1972, July-August). Evolution and revolution as oisani- zations grow. Harvard Business Review. 37-46.

Hambleton, R. K., Blanchard, K. H., A Hersey. R (1977). MaturUy Scale—Sdf-rating form. San Diego, CA; Learniog Resources Corpo- ration.

Hambleton, R. K., A Gumpol, R. (1982). The validity of Heney and Blanchard's theory of leader effectiveness. Groi^ and Organization Studies. 7,225-242.

Harris, T (1969). I'm OK—You 're OK: A practical guide to transac- tional analysis. New York; Harper A Row.

Herzbeis, F. (1966). Work and the nature of man. New \bik; Worid Publishing.

Henty; R, A Blanchard, K. (1969). Life<yde theory of Indcnhip. Tnuning and Development Journal, 23,26-34.

SITUATIONAL LEADERSHIP THEORY 451

Heney, P., A Blanchard, K. (1982). Management of organizational be- havior {4ih cd.). Englewood Qiffs, NJ; Prentice-Hall,

Ken; S,, A Jennici; J. M. (1978). Substitutes for leadership; Their mean- ing and measurement Organizational Behavior and Human Perfor- mance. 22. 375-403,

Korman, A, K. (1966), Consideration, initiating structure, and organi- zation criteria—A review. Personnel Psychology. 19. 349-361,

Lewin, K. (1947). Frontiers in group dynamics; Concept, method, and reality in social science; social equilibria and social change. Human Relations. 1,5-41.

Liden, R. C , A Graen, G, (1980), Generalizability of the vertical dyad linkage model of leadership. Academy ofManagement Journal. 23. 451-465,

Likert, R, (1967), The human organization. New York; McGraw-Hill, Maslow, A, (1954), Motivation and personality New York; Harper A

Row, McQelland, D, C, (1961), The achieving society Princeton, NJ; \za

Nostrand, McGregOA D, (I960), The human side of enterprise. New York: Mc-

Graw-Hill, Moore, L, I, (1976), The FMI: Dimensions of follower maturity. Group

arui Organization Studies. 1. 203-222,

Reddin, W, J, (1967). The 3-D Management Style Theory. Trainingand Development Journal. 21. 8-17.

Schein, E. H. (1970), Organizational psychology (2tui ed,), Englewood Qiffs, NJ; Prentice Hall,

Schriesheim, C, A,, A Kerr, S, (1974), Psychometric properties of the Ohio State leadership scales. Psychological Bulletin. 81. 756-765,

Skinner, B, F, (1953), Science and human behavior. New York; MacMil- lan.

Smith, P,, Kendall, L,, A Hulin, C, L, (1969), The measurement of satis- faction in work and retirement. Chicago; Rand-McNally,

StogdiU, R,, A Coons, A, (1957), Leader behavior: Its description and measurement (Research Monograph No, 88), Columbus; Ohio State University, Bureau of Business Researeh,

Vecchio, R. P, & Gobdel, B, C, (1984), The vertical dyad linkage model of leadership; Problems and prospects, Organizatioruil Behavior and Human Perfomuxnce. 34, 5-20,

Vroom, V, H,, A Mann, F C. (1960), Leader authoritarianism and em' p\oyee axtituda. Personnel Psychology. 13. 125-140,

Received July 7, 1986 Revision received January 8, 1987

Accepted January 12,1987 •