Article Component Papers (Early Contingency Theories Of Effective Leadership)

y of Management Review 1979, Vol. 4. No. 2,249-257.

A Critique of the Vroom-Yetton Contingency Model

Of Leadership Behavior̂ R. H. GEORGE FIELD University of Toroiito

The Vroom-Yetton contingency model of leadership behavior is described and critiqued. While there is some evidence for the validity of the model, the use of concurrent validation with self-report data seriously questions this validity. The modei is of limited utility because It deals with only one aspect of leader behavior and is not as parsimonious as alternative mod- els.

The Vroom-Yetton contingency model of leader- ship behavior (21) is of importance to managers because it was developed as a model of how lead- ers should make decisions if they are to be effec- tive. If valid, the model could be used as a tool to aid in the selection of appropriate decision-making processes for different situations, resulting in in- creased decision and organizational effectiveness. Ttie full Vroom-Yetton contingency model deals with individual problems and group problems (19, 21). This paper will analyze the group model since it has been most frequently presented and tested (3, 5,7,14,16,17,18,20,21).

The model is deductive in nature, in that seven rules have been derived from the decision-making literature that are designed to protect decision qual- ity and acceptance. The rules are applied to the characteristics of the problem in the form of a deci- sion tree and 23 unique problem situations are iden- tified. Each problem situation has associated with it a feasible set of decision processes that are appro- priate for use in solving the problem (Figure 1).

Decision quality depends on objective aspects of 'he decision whereas decision acceptance de-

* wish to thank Hugh Arnold, Martin Evans, Bob House and Harvey Kolodny for their heipful comments on earlier versions of !fiis paper.

'•- '^^^t>y the Academy of Managemeni 0363-7425

pends on subjective aspects of the decision (the feelings of people) (9).

Decision processes that may be used range from autocratic (Al, All), to consultive (Cl, CM), to group process (Gil). Subordinate participation in the de- cision-making process increases in the order of Al, All, Cl, CM, Gil. When using process Al the mana- ger solves the problem alone, using information available at that time, while with All the manager obtains any necessary information from subordi- nates and then makes the decision. Cl is a consul- tive process in which the manager shares the prob-- lem with relevant subordinates individually and then makes the decision. With process CM the manager shares the problem with subordinates in a group meeting, and then makes the decision. Gil is a group decision process where the manager and subordinates aim to reach consensus on the deci- sion as a group.

When more than one decision-making process may be used for a problem (the feasible set), the manager is free to choose among them since deci- sion quality and acceptance are protected for all decision methods in the feasible set.

Vroom and Yetton (21) identify two criteria for choosing within the feasible set. When the manager is guided by the desire to make decisions with the least expenditure of personnel hours, the most au-

249

A. Does the problem possess a quality requirement?

B. Do I have sufficient information to make a high quality decision?

C. Is the problem structured?

D. Is

acceptance of the decision by subordinates important for effective implementation?

1

If I were to make the decision by myself is it reasonably certain that it would be accepted by my subordinates?

. Do subordinates share the organizationai goals to be attained in solving this problem?

I

conflict among subordinates likely in preferred solutions?

PROBLEM

SITUATION

Gil

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

A[,AII,CI,Olj

Al,AII.CI.CII

AI,AI!.CI.Oi

AI.AII.CI.e

Gil

CM

CI.CII

All.CI.CIIfl

AM.CI.CII

Gil

cn

CI.CII

AII.CI.CII.Gl

AII.CI.CII

CM

CII.GII

CM

CII.GII

Gil

CH

Figure 1 The Vroom-Yetton Contingency Model

of Leadership Behavior ^

Adapted with permission from V. H. Vroom, "Can Leaders Learn to Lead?" Organizational Dynamics, Voi. 4, No. 3 (1976), p. 19, © 1976 by AMACOM, a division of American Management Associations and from T. E. Hill and N. Schmitt, "Individual Differences in Leadership Decision Making," Organiza- tional Behavior and Human Performance, Vol. 19 (1977), p. 356.

250

•icratic of the decision processes in the feasible set -nould be chosen. This is identified as the short- term model since there is no emphasis on the de- elopment of subordinates. When the manager .ishes to emphasize development of individuals nd groups, the most participative of the decision

processes in the feasible set should be chosen. This is identified as the long-term or developmental "odei.

Research Results

Vroom and Yetton (21) found that when a mana- ger was asked to recall one problem recently solved .n the role of manager, 85% of these problems had a quality requirement and 90% had an acceptance equirement. They also found that managers' deci- ;on-making behavior fell in the feasible set 68% of 18 time. When managers were asked to indicate which

lecision processes they would use to solve stan- lardized case problems, it was found that their aecision-making behavior fell in the feasible set 69% of the time. Also, the short-term model that chooses the most autocratic decision method from ne feasible set was found to be more varied in :ecision processes prescribed and slightly more .utocratio than the managers tested. Vroom and Yetton attempted to test the overall

validity of the model. They reasoned that: The validity of the rules underlying the model would be indicated by the degree to which employment of methods used that were outside the feasible set result in decisions of lower quality or lower accep- tance than those within (21, p. 183).

They had 136 managers who were unfamiliar with the model describe one problem they recently had to solve. These managers also specified the decision process used, the status of the problem attributes (in order to determine the feasible set), 5nd decision quality and acceptance (on seven oint scales). Vroom and Yetton found that in 97 of 136 prob-

lems (71%) the manager's decision process fell in the feasible set, but almost every decision (132 of 36) was rated as having high quality and accep- n̂ce. Therefore, a comparison of quality and ac- 'Ptance of decisions inside and outside of the '=asible set showed little difference. Decisions

made with processes outside the feasible set were found to have slightly lower quality and acceptance than decisions made with processes inside the fea- sible set, but these differences were not statistically significant.

Vroom and Yetton also found that decision qual- ity or acceptance was lower when a quality or ac- ceptance rule was violated than when it was not, but again the differences were not statistically signifi- cant.

Evidence of Model Validity Vroom and Jago (20) have recently tested the

Vroom-Yetton model, extending the results of Vroom (17), and have avoided the lack of variance of decision quality and acceptance encountered by Vroom and Yetton (21). They had 96 managers who were unfamiliar with the model select and report one successful and one unsuccessful decision made in their managerial role. The managers then reported the decision method used to solve each problem and the overall effectiveness, quality, and acceptance of the solution. Managers were later trained in coding problem attributes and determined the feasible set for each problem.

Results show some evidence of model validity. Decisions made with processes inside the feasible set were found to be significantly more likely to be rated as successful than were decisions made with processes outside the feasible set. The model did not predict perfectly, however, in that decision proc- esses within the feasible set did not guarantee sue-. cess and decision processes falling outside the fea- sible set did not always result in failure (51 of 181 decisions fall in these categories).

Vroom and Jago (20) also found that decision quality or acceptance were significantly lower when the respective quality or acceptance rules were vio- lated. Overall, these results show some, but limited, evidence of the model's validity.

Methodological Issues Self-Report Effect

The major threat to the internal validity of the model is the fact that it has been tested with self- report data (20,21). In a major test of the model (20) managers were asked to list the details of one suc- cessful and one unsuccessful decision-making sit-

251

uation that they personally experienced in their role as a manager. This case writing occurs after the managers have been exposed to the five decision processes, a chapter concerning problem solving (9), and practice in choosing a decision process for each of 30 standardized cases. After writing the two cases, the managers report the decision process actually used for each case along with a rating of the decision's overall effectiveness, quality, and accep- tance. Managers are then trained in understanding problem attributes and subsequently code the prob- lem attributes of each of the two previously reported problems in order that the feasible set for each problem can be ascertained.

It is possible that managers report successful decisions as using a rational decision process that was appropriate to the situation, regardless of their actual behavior. Thus, self-reported rational beha- vior would tend to match the rational model and the model would be validated. Similarly, ineffective de- cisions could have leader behavior reported as fall- ing outside the feasible set. The leader would attrib- ute the ineffectiveness of the decision to others and would report the decision process used was inap- propriate to the actual situation, even though the decision process used was appropriate to the situa- tion as it then existed.

A further error could occur because the depen- dent variables of decision effectiveness, quality, and acceptance are self-reported on seven point scales. When asked to report an effective decision, the manager will likely report high decision effec- tiveness, quality, and acceptance and may have cognitively distorted the actual levels of these varia- bles. Similarly, when reporting an ineffective deci- sion the actual levels of effectiveness, quality, and acceptance may be cognitively distorted and re- ported at low levels. The effect of this error would be to artificially increase the difference between effec- tive and ineffective decisions and thereby increase the probability of supporting hypotheses concern- ing effective and ineffective decisions.

Jago and Vroom (6) report that subordinate's perceptions of managerial behavior do not correlate significantly with the superior's own descriptions of the same behavior. This raises the question of the validity of manager behavior self-reports. To test this, Jago and Vroom (7) matched what managers

said they would do in standard cases to what thev said they did in similar self-reported oases, and found a significant positive correlation. This resui suggests that managers act as they say they woulij act on a hypothetical decision, although thefindim is a correlation between two self-reports.

To evaluate the extent of these potential self-re- port effects, reported decision process, situatioo attributes, decision effectiveness, quality, and ac- ceptance must be validated against objective mea- surements of these variables. This validation couW be performed as part of a laboratory test of the Vroom-Yetton model, where variables are con- trolled and objective measurements are available. Experimenter Effect

An experimenter effect would be present if sub- jects tried to guess the hypotheses being tested and supplied supporting data. Vroom and Jago (20) have attempted to minimize this problem by giving subjects specific instructions and decision trees with feasible sets not identified. Further, they state that this is not a problem because of the complexity of the model and the difficulty in guessing the hy- potheses.

However, it is possible that an experimenter ef- fect is present in the Vroom and Jago (20) test ofthe model because subjects are exposed to Maier's chapter (9, ch. 1) on problem solving which sets forth three basic types of problems. These three types are: (a) decision quality most important, (b) decision acceptance most important, and (c) deci- sion quality and acceptance both important. Since the Vroom-Yetton model is similar to Maier's devel- opment in the emphasis on decision quality and acceptance it is possible that subjects have a basic understanding of the model before they are asked to supply data to test it.

Social Desirability Effect A social desirability bias toward fitting the model

could occur if for effective decisions managers re- port they used more participative decision process- es than they actually did. Managers are likely to do so because of the current social emphasis on parti- cipation.

This bias would occur because participative de- cision processes are included in the feasible set more often than are autocratic decision processes

252

Mote that the autocratic processes Al and All are found in the feasible set for 6 and 10 of the 23 problem situations respectively, whereas the parti- cipative processes Cl, CM, and Gil are found in the feasible set for 12, 19, and 12 of the 23 problem situations (Figure 1).

Standardized Cases

A potential internal validity problem exists in that '/room and Yetton (21) developed the model using JO standardized cases to represent different com- binations of problem attributes. A full manipulation of the seven problem attributes (coded Yes/No) would require 128 standardized cases. This num- ber was reduded to 30 by Vroom and Yetton be- cause of time limitations when gathering data. They ay this reduction can be made because of logical r empirically demonstrated relationships between ne seven variables, but reducing the number of elationships tested has eliminated the possibility of ;iscovering relationships that run counter to logic or previously demonstrated empirical relationships.

This criticism has been dealt with by Hill and Schmitt (3) who tested a shortened Vroom-Yetton model with a completely crossed design. They found that 37% of leader decision method variance was due to situational effects and 8% was due to individual effects. This result is very similar to that of Vroom and Yetton (21) who found variances of 30% and 10%, respectively.

Steinman (14) found that the 30 standardized -ases confound the hierarchical level of subordi- nates. If managers report they would use more participative decision processes with subordinates of higher hierarchical levels, then the Vroom and Yetton result that the seven situational variables predict participativeness may be spurious.

Jago (5) performed a test of the above result and ound that confounded differences in hierarchical ivel did significantly inflate, the relationship be- tween three situational attributes (B, D, and G of Figure 1) and the level of participation chosen by managers on the cases. However, Jago notes that ê size of this relationship has been reduced for nly three of the seven situational attributes, and at all situational attributes are still significantly

related to manager's choices of participative versus autocratic decision processes.

Jago (5) has developed a new set of 30 standar- dized cases that does not confound hierarchical level, thus increasing the internal validity of the instrument.

Reliability A final threat to internal validity when using the

model prescriptively is the reliability of problem at- tribute coding by model users. Vroom and Yetton (21) found that on standardized cases the percen- tage of subject problem attribute coding matching expert coding went from 30% without training to 63% with training. Allowing for minor differences in coding which make no difference in the feasible set selected, the percentage of subject coding match- ing expert coding is 45% without training and 78% with training. These results imply that coding relia- bility without training is not adequate. Even with training, errors in problem attribute coding will cause the selection of inappropriate decision proc- esses in some cases.

External Validity The major threat to external validity is that the

model may not be generalizable to naturally occur- ring problem situations, since the model was devel- oped and tested using standardized cases and self- reported decisions. Standardized cases were de- veloped to vary systematically on selected problem attributes and were rewritten until most persons (75%) reading the cases could agree on the pres- ence versus absence of the seven problem attri- butes. These cases are therefore fairly unambigu-" ous and do not reflect the ambiguity of real-life decisions. Self-reported decisions could be biased toward important, unambiguous, or short-term problems, because reporting such a decision would be more meaningful or easier for the manager. Generalizing the model to less important decisions, long-term decisions, or decisions where problem attributes are not as clear could prove difficult.

The Vroom-Yetton model presents a simplified view of reality, which lessens its usefulness as a prescriptive tool. While any model is an abstraction of reality, currently 55% to 60% of the variance in managers' selection of decision processes remains unaccounted for by the model (3, 21). Individual difference variables, structural variables (2, 14), and other situational variables could be added to

253

the Vroom-Yetton model to provide a more complex model that would likely explain more of the variance in managers' selection of decision processes. This additional complexity will avoid a simplified view of reality, but poses the threat of the model being too complex to use in actual practice (1, 14). Phelps and Shanteau (13) have, however, found that ex- perts judging livestock could use information from a large number of dimensions when explicitly pre- sented with the information in a hypothetical situa- tion, but used few dimensions when dealing with a more realistic situation. It seems then that the Vroom-Yetton model could be made more complex and still be used in actual practice if the problem situation dimensions were dealt with explicitly.

A basic assumption of the Vroom-Yetton model is that it should be prescriptive (21), allowing mana- gers and leaders to determine which decision meth- od to use in different situations. Later in their book, Vroom and Yetton declare that "we have viewed leadership development as the principal application of our work " (21, p. 208); a switch of emphasis from prescription to leadership training. In a later work, Vroom (18) maintains that the model is useful for training managers in leadership and decision mak- ing, but is not intended to train managers to pro- gram their choices of decision style. However,

• Vroom and Jago (20) find evidence for the overall validity of the Vroom-Yetton model and state that the model can be used prescriptively in its present form to reduce many errors found in current man- agerial practice.

It is not clear if the model is intended to be a training tool, a prescriptive device, or both. The limited evidence available supporting the validity of the model and the threats to model validity raised in this paper imply that the model should not as yet be used for prescriptive purposes or leader training.

Discussion Assumptions ofthe Vroom-Yetton Model

Vroom (15) reported that three classes of out- comes that are relevant to effectiveness and af- fected by participation in decision making are: (a) quality, (b) acceptance, and (c) time taken to reach the decision. Decision quality and acceptance are protected in the Vroom-Yetton model by the seven normative rules, but decision time is used only to

describe the short-term and developmental mod- els. Vroom and Yetton (21) argue that the time available for a specific decision is not an importait factor (usually managers have more than one week to make a decision), and so does not need to be explicitly included as a rule defining the normative model. However, the amount of time taken to solve a problem is the expenditure of a resource, andtfiis should be balanced against the benefits gained. Therefore, the time taken to make a decision and the expected benefits should be considered when selecting from the feasible set.

Decision processes that comprise the feasible set do not guarantee quality and acceptance, but offer a better chance than do the decision proc- esses not in the feasible set. Vroom and Yetton (21| state that decision processes left in the feasible set have equal likelihoods of meeting quality and ac- ceptance requirements. However, different deci- sion processes require different skills ofthe leaderif they are to be effective. Autocratic decisions require leader task knowledge and communication skills, while group participation decisions require leader discussion and conference skills. Since managers can be expected to vary on their levels of these skills, then they can be expected to be more effec- tive when using decision processes of the feasible set that most closely match their skills.

The goal-congruence rule of the Vroom-Yetton model implies that managers may act in the organi- zation's interests but that subordinates may not. Itis inconsistent to maintain that managers will not al- low their self-interests to interfere with their decision making while subordinates have no such restric- tion. This assumption limits the model's correspon- dence with reality as the model will not account for actions a manager may take that are motivated by self-interest.

Problem attributes of a situation should be simi- larly coded by all managers, but Hoffman (4) has noted that one problem of the model is that mana- gers are apt to answer problem attribute questions in a way that reflects their leadership style regard- less of the actual situation. For example, an auto- cratic manager could be expected to answer "^e^ to attribute E, "If I were to make the decision by myself, is it reasonably certain that it would be ac- cepted by my subordinates?" This will allow the

254

manager to continue using an autocratic decision process and avoid participation, even when the sit- ation calls for a participative process. This effect is

possible, because Vroom and Yetton (21) have found that individual differences accounted for 8 5% of the variance in managerial choices among alternative decision processes.

The Vroom-Yetton model has the form that man- agers attend to certain situational variables and select a decision process based on the states of these variables. The decision process chosen causes decision effectiveness. This conception of causality may be too limited, since Lowin and Craig (8) have found that subordinate performance can affect preferred leadership style. A manager may wish to closely supervise subordinates who are not performing well, while greater freedom and partici- ation may be allowed those performing well.

Boundary Conditions of the Vroom-Yetton Model

Vroom and his colleagues present a managerial raining program (16,17,18,21) that exposes man- .gers to the concepts underlying the model. The program includes peer discussion concerning lead- ership styles, and how to apply the model to a roblem situation and identify the feasible set.

Since this training is required in order to use the iiodel reliably (21), the population that may use the model prescriptively is limited.

Another boundary condition of the Vroom-Yetton model as a theory of leadership is that it deals with only one facet of leader behavior, that of selecting a decision process for a particular problem situation. he model ignores the discussion and conference kills required of the manager to actually solve the

problem (4). The Vroom managerial training pro- gram attempts to develop these skills, but no evi- ienoe of the effectiveness of the training program as been presented.

Theoretical Alternatives to the Vroom-Yetton Model

The Vroom-Yetton model is founded on Maier's definition of effective decisions being a function of iuality and acceptance (9), and focuses on choos- •̂g the appropriate decision process for each prob- 9m situation. A manager wishing a prescriptive

analysis of a decision situation could use a much simpler model developed by Maier (10).

In a readings book that summarizes fifteen years of research he and his colleagues conducted on problem solving and creativity, Maier presents a four-way classification of problems and how these decisions should be made (10, pp. 325-327). A Type I problem has a quality requirement, accep- tance is likely to be obtained easily, and the deci- sion should be made by the leader. Type II prob- lems do not have a quality requirement but acceptance is critical. These problems should be resolved by group decision. Type III problems do not have quality or acceptance requirements and should be decided by the flip of a coin. Maier cau- tions against using participative approaches with this type of problem as unnecessary conflicts may be generated. Type IV problems require both qual- ity and acceptance and are solved by using man- agerial skills. These leadership skills can be direct- ed toward persuasion or group discussion, but Maier emphasizes the latter.

It seems that Vroom and Yetton (21) have added unnecessary complexity to Maier's simple model of four basic problem types and decision processes that should be used to solve each type.

Further, Maier has gone beyond merely specify- ing problem types and suggesting leadership deci- sion methods, but has done considerable research and has specified the skills managers will need in solving problems and how these skills may be ac- quired. He and his colleagues have reported a num- ber of studies concerning solution quality, accep- tance, the role of the leader in group problem solv- ing, and conditions favorable to individual and group problem solving (10, 11). Maier has written a summary book of leadership methods and skills as well (9) and with colleagues has created a number of exercises that have been used for training and research (12).

Another theoretical alternative to the Vroom-Yet- ton model is a simplified model based on the one proposed by Vroom and Yetton (21). Note that in Figure 1, the decision process CM is in the feasible set for 19 of the 23 problem situations. Of the four situations where Cll is not in the feasible set, Gil is in the feasible set. A simple rule to guarantee a decision process in the Vroom-Yetton feasible set

255

and thereby protect decision quality and accep- tance is:

If acceptance of the decision by subordinates is critical to effective implementation and it is not rea- sonably certain that subordinates would accept an autocratic decision, but they share organizational goals (or decision quality is not important), use Gil; otherwise, use Cll.

This simple model uses four situation attributes in- stead of the seven used by the Vroom-Yetton model and uses only two decision processes rather than the five of the Vroom-Yetton model. Also, this sim- ple model offers a balance between the short-term time efficient model and the long-term group devel- opment model.

Since the Vroom-Yetton model can be signifi- cantly reduced in complexity and yet still utilize the core concepts of the model, those of protecting decision quality and acceptance, it seems that the current model is not parsimonious. The current complexity of the model has not been justified in the latest test of the model (20) because two of the seven rules underlying the model were found not to have a significant effect on decision effectiveness.

Conciusion While some evidence for the validity of

Vroom-Yetton model has been presented, thereare a number of criticisms of the model that question its validity and utility.

The major threat to the validity of the model is the use of concurrent validation with the problem ai butes, decision process, decision effectiveness, quality, and acceptance all being self-reported t the subject manager. A predictive validation study of the model should be undertaken, using objective coding of these variables. Rigorous research ol this kind is necessary before the model can be accepted as valid for use as a prescriptive decision-making or training tool.

The utility of the Vroom-Yetton model is ques- tionable for two reasons. First, it is not as parsimo- nious as other models of leader decision process choice. Second, it deals with only one aspect of leader behavior, that of selecting different decision processes for different problem situations. Maiers work on problem solving deals with this aspect of leader behavior plus requisite leader skills and theif development.

REFERENCES

1. Filley, A. C , R. J. House, and S. Kerr. Managerial Process and Organizational Behavior, 2nd ed. (Glenview, III.: Scott, Foresman, 1976).

2. Heller, F. A., and G. Yukl. "Participation, Managerial Deci- sion Making, and Situational Variables," Organizational Behavior and Human Performance, Vol. 4 {1969), 227-241.

3. Hill, T. E., and N. Schmitt. "Individual Differences in Leader- ship Decision Making," Organizational Behavior and Hu- man Performance, Vol. 19 (1977), 353-367.

4. Hoffman, L. R. "Review of Leadership and Decision-Mak- ing',' Journai of Business, Vol. 47 (1974), 593-598.

5. Jago, A. G. "A Test of Spuriousness in Descriptive Models of Participative Leader Behavior," Journai of Appiied Psy- chology, Vol. 63 (1978), 383-387.

6. Jago, A. G., and V. H. Vroom. "Perceptions of Leadership Style: Superior and Subordinate Descriptions of Decision- Making Behavior," in J. G. Hunt and L. L. Larson (Eds.), Leadership Frontiers, (Carbondale, III.: Southern Illinois University Press, 1975), 103-120.

7. Jago, A. G., and V. H. Vroom, "Predicting Leader Behavw From a Measure of Behavioral Intent," Proceedings, Am- ican Institute for Decision Sciences, Vol. 9 (1977), 503-505.

8. Lowin, A., and J. R. Craig, "The Influence of Level of Per- formance on Managerial Style: An Experimental OD) Lesson in the Ambiguity of Correlational Data," Organm- tionai Behavior and Human Performance, Vol. 3 (1968|. 440-458.

9. Maier, N. R. F. Problem Solving Discussions and Conb- ences: Leadership Methods and Skills (New York: Mc- Graw-Hill, 1963).

10. Maier, N. R. F. Probiem Soiving and Creativity in individmis and Groups (Belmont, Calif.: Brooks/Cole, 1970).

11. Maier, N. R. F. Psychoiogy in industriai Organizations.' ed. (Boston: Houghton Mifflin, 1973).

12. Maier, N. R. F., A. R. Solem, and A. A. Maier, Supervisory and Executive Development: A Manual for Role Play«i9 (NewYork: Wiley, 1957).

13. Phelps, R. H., and J. Shanteau, "Livestock Judges: Hw

256

Much Information Can an Expert Use?" Organizational Be- 18. havior and Human Performance, Vol. 21 (1978), 209-219. Steinman, J. I. Some Antecedents of Participative Decision •Making (Ph.D. dissertation. University of California, Berke- 19. ley, 1974). Vroom, V. H. "Industrial Social Psychology," in G. Lindzey and E. Aronson (Eds.), Handbook of Sociai Psychology, 20. Vol. 5 (Reading, Mass.: Addison-Wesley, 1969), 196-268. Vroom, V. H. "Decision Making and the Leadership Proc- ess," Journal of Contemporary Business, Vol. 3, No. 4 21. (1974), 47-64. Vroom, V. H. "Can Leaders Learn to Lead?" Organizational Dynamics, Vol. 4, No. 3 (1976), 17-28.

Vroom, V. H. "Leadership." in M. D. Dunnette (Ed.). Hand- book of industrial and Organizational Psychology (Chica- go: Rand McNally, 1976), 1527-1551. Vroom, V. H.. and A. G. Jago. "Decision Making as a Social Process: Normative and Descnptive Models of Leader Be- havior," Decision Sciences. Vol. 5 (1974), 743-769. Vroom, V. H., and A. G. Jago. "On the Validity of the Vroom- Yetton Model," Joumai of Applied Psychoiogy, Vol. 63 (1978), 151-162.

Vroom, V. H., and P. W. Yetton, Leadership and Decision- Making (Pittsburgh, Penn.: University of Pittsburgh Press, 1973).

R. H. George Field is a doctoral candidate at the Faculty of Management Studies, University of Toron- to, Canada.

Received 2/16/78

257