Importance of Supportive Relationships in Goal Setting - American ...

13 downloads 40 Views 494KB Size Report
between the actual difficulty level of the goal and the number of ideas generated in the as- signed condition, r(2&) = .59, p < .05, the participative condition, r(28)Ā ...
Journal of Applied Psychology 1979, Vol. 64, No. 2, 151-156

Importance of Supportive Relationships in Goal Setting Gary P. Latham and Lise M. Saari University of Washington The importance of supportive behavior by an authority figure when setting goals was tested using a brainstorming task. Ninety college students were randomly assigned in a 2 X 3 design to a supportive or nonsupportive condition and to one of three goal-setting conditions (assigned, participative, and "do your best"). Goal difficulty was held constant between the assigned and the participative conditions. Supportive behavior resulted in higher goals being set than nonsupportive behavior. Participatively set goals led to better performance than assigned goals. Participation appears to be important in that it increases understanding of task requirements. Three key aspects of a System 4 (Likert, 1967) management theory are supportive relationships, participative decision making, and goal setting. The principle of supportive relationships states that a supervisor should be perceived by a subordinate as a person who is primarily interested in building and maintaining the subordinate's sense of personal worth. Likert argues that the more often a supervisor is perceived as supportive by the subordinate, the better will be the effect of the supervisor's behavior on the subordinate's performance. The empirical evidence cited in support of this principle, however, consists primarily of correlational studies (e.g., Bowers & Seashore, 1966; Carroll & Tosi, 1970). The principles of participation and goal setting have been subjected to rigorous experiments in both laboratory and field settings. This is particularly true for goal setting. It has been shown consistently that speThe authors are grateful to R. Likert, E. A. Locke, and T. R. Mitchell for their helpful comments in reviewing this article. We also thank A. Collins and C. Fay respectively for their assistance in conducting the experiment and performing the data analysis. Requests for reprints should be sent to Gary Latham, Department of Management and Organization, DJ-10, University of Washington, Seattle, Washington 98195.

cific goals lead to better performance than general goals and that difficult goals, if accepted, lead to better performance than easy goals (Latham & Yukl, 1975b; Locke, 1968.) Conclusions about the importance of participation, especially with respect to goal setting, have remained elusive. For example, in an experiment with educationally disadvantaged loggers, participation was found to lead to the setting of higher goals and greater goal attainment than was the case when the goal was assigned (Latham & Yukl, 1975a). But in a second experiment, no significant difference was found between the performance of typists with assigned goals versus participatively set goals (Latham & Yukl, 1976). Nor was there any difference in measures of goal acceptance or attainment. The goals were equally difficult in the two experimental conditions. In a third experiment, participation led to significantly higher goals being set by engineers and scientists than was the case when the goals were set unilaterally by supervisors (Latham, Mitchell, & Dossett, 1978). However, there was no significant difference between the responses of the engineers and scientists in these two conditions on measures of goal acceptance, per1 ceived goal difficulty, or probability of goal attainment. Latham et al. concluded that participation is important insofar as it influences goal difficulty and, hence, performance,

Copyright 1979 by the American Psychological Association, Inc. 0021-9010/79/6402-0151$00.75

151

152

GARY P. LATHAM AND LISE M. SAARI

but that goal specificity and goal acceptance saved by deleting the participation process can be attained as easily through assigned could be substantial. goals as through participatively set goals. In reviewing these studies, Likert (Note In a laboratory experiment on brainstorm- 1) stated that participation appears to be ing, Latham and Saari (in press) tested the important in that it results in employees setclaim of modern organizational theorists (e.g., ting high goals without perceiving them as difLikert, 1967) that even with goal difficulty ficult. This comment is based primarily on the held constant, participatively set goals lead study by Latham et al. (1978) in which to greater productivity (ideas generated) engineers and scientists set higher goals and than do assigned goals. The hypothesis was attained them as often as individuals with not supported. Performance, goal acceptance, assigned goals; there was, however, no sigand goal attainment were not significantly nificant difference in the employees' percepdifferent in the two goal-setting conditions. tions of goal difficulty between the two goal Latham and Saari concluded that participa- conditions. Moreover, Likert (Note 1) noted tion of subordinate employees is important that when the assigned goal had been effeconly to the extent that it leads to the setting tive, the supervisor had always behaved in a of higher goals than a supervisor would set highly supportive manner. The authors' obunilaterally. servations in each of these studies and the Dossett, Latham, and Mitchell (in press) subjects' comments support this interpretareplicated the above study in the context of tion. (a) a concurrent validity study, and (b) a The purpose of the present study was to performance appraisal study. Goal difficulty test experimentally the importance of Likert's was held constant between the two goal-set- (1967) principle of supportive relationships ting conditions. In the first experiment, Dos- to goal setting with goal difficulty held consett et al. found that employees who were stant. The experiment was conducted in the assigned a specific test score as a goal at- laboratory because of practical and ethical tained their goal significantly more often considerations in manipulating supportiveness than individuals who had participated in in a field setting. The task used was brainsetting their goal. storming, since this is a procedure used by In the second field experiment, Dossett many organizations and since the differences, et al. found a marginally significant differ- if any, that exist between an industrial and ence between the performance appraisal mea- a university setting are minimal. The hysures for the two goal-setting groups. Those potheses tested were as follows: (a) A supwith assigned goals performed better than portive management style leads to higher employees who participated in setting the performance than a nonsupportive style; (b) goals. Moreover, goal acceptance was sig- setting a specific goal leads to better pernificantly greater in the assigned than in formance than adopting a general attitude the participative condition. The subordinates of "doing one's best"; and (c) allowing an in the assigned group were more likely than individual to participate in goal setting leads those in the participative group to say that to better performance than assigning a goal. goal attainment would lead to favorable or These hypotheses are consistent with Likert's rewarding consequences such as job security, theory of effective managerial behavior. future pay increases, promotion, and coworker respect. Method Dossett et al. concluded that if an assigned Ninety college students (44 males and 46 females) goal is difficult and challenging, the result is were randomly assigned to one of six conditions likely to be performance that is equal, if in a 2 X 3 factorial design: (a) supportive/parnot superior, to that which occurs as a re- ticipatively set goals, (b) supportive/assigned goals, sult of participatively set goals. Moreover, (c) supportive/do best, (d) nonsupportive/participatively-set goals, (e) nonsupportive/assigned goals, if a supervisor has a large number of sub- and (f) nonsupportive/do best goals. The age of ordinates reporting directly to her, the time the subjects ranged from 18 to 24.

GOAL SETTING AND SUPPORTIVE RELATIONSHIPS In the supportive conditions, the experimenter followed a memorized script that emphasized (a) giving the subject a friendly welcome, (to) reassuring the subject that he or she would do well, (c) using words of encouragement and support (e.g., "Do you feel comfortable with that goal?"), (d) encouraging the subject to ask questions, and (e) asking rather than telling the subject to do things. In the supportive/participative condition, the experimenter, who was blind to the hypotheses of the study, explained that this research was sponsored by a wood products company and that the task consisted of generating ideas for uses of wood. The subject was asked to set a specific, difficult, but attainable goal in terms of the number of ideas that he or she could list within 20 minutes. If the goal the person set was extremely high or extremely low, the experimenter reemphasized that the goal should be difficult but attainable, and asked the subject if he or she was satisfied that the set goal fit that definition. His decision to make this statement was based on his knowledge that the average number of ideas generated within 20 minutes in a previous study (Latham & Saari, in press) was 55. However, this number was not conveyed to the students because we did not want the subjects to feel compelled to set a goal equal to or higher than this number. Once a goal was agreed upon, the experimenter gave the subject a clock and a sheet of paper on which he or she was to list and number the ideas. This was done to provide each individual with knowledge of results regarding his or her performance in relation to the goal. After answering any questions, the experimenter left the room and immediately stopped a concealed stopwatch that had been running since he had first begun to talk with the subject. This was done to determine the time it took to explain the brainstorming procedure and to set the goal. The supportive/assigned condition was identical to the supportive/participative condition, except that the experimenter assigned the goal to each individual. This goal was identical to the one set by a person who had been randomly assigned to work in an adjoining room in the supportive/participative condition. The procedure for the supportive/do best condition was the same as for the two supportive goalsetting conditions except that rather than providing a specific goal for the subject to try to attain, the experimenter simply asked each individual to do his or her best to list as many ideas as possible within 20 minutes. Instructions given in the nonsupportive conditions did not differ from .those given in the supportive conditions, but for each condition, the experimenter acted differently toward the subject. For example, in the nonsupportive conditions, the experimenter (a) told the subject that he was in a hurry, (b) told the subject to listen because he did not have a lot of tune, (c) tossed the sheet of paper to the

153

subject, and (d) continually glanced at his watch. In sum, the experimenter behaved in a rude and abrupt manner. At the end of the 20-minute brainstorming task, each subject completed a two-part questionnaire. Part 1 contained questions about who had influenced the goal set ("How much influence did you have over the goal?"; "Compared to the experimenter, how much influence did you have over the goal that was set?"), goal difficulty ("How difficult was it for you to attain the goal?"), and goal acceptance ("How committed were you to attaining the goal?"; "How much satisfaction did you experience from working toward the goal?"; and "How important was goal attainment to your feeling of achievement and accomplishment on this task?"). The subjects responded to each of these questions on a 5-point Likert-type scale. Part 2 required the subjects to rate the experimenter on the following 11 bipolar adjectives, each with an 8-point scale: pleasant-unpleasant, cold-warm, supportive-hostile, considerate-4nconsiderate, accepting-rejeoting, nasty-nice, distantclose, friendly-unfriendly, gloomy-cheerful, quarrelsome-harmonious, and kind-unkind.

Results Manipulation Checks The goal-setting manipulation was effective. A 2 X 2 unweighted means analysis of variance (omitting the do best groups) on the sum of the two questions (Cronbach's alpha = .92) revealed a main effect for goal setting only, F(l, 56) - 14.53, p < .05. The subjects in the participative condition felt that they had more influence (M = 7.53, SD ~ 2.24) than those in the assigned condition (M = 5.07, SD = 2.72) in setting the goals. No interaction effect was obtained. A 3 X 2 unweighted means analysis of variance of the ratings on the 11-item supportiveness scale (Cronbach's alpha = .96) revealed a main effect for supportiveness only, F(l, 84) = 24.68, p < .05. No interaction effect was found. The subjects in the supportive conditions (M = 77.96, SD = 11.95) rated the experimenter significantly higher than did those in the nonsupportive conditions (M - 64.05, SD= 14.20). Nevertheless, it should be noted that the mean response of those in the nonsupportive condition was on the favorable side (5.34 on an 8-point scale). Thus it cannot be said that the subjects in this condition perceived the

154

GARY P. LATHAM AND LISE M. SAARI

experimenter as nonsupportive, but only that he was perceived as less supportive than he was in the supportive condition. Goal Difficulty No significant difference was found between the participative and assigned goalsetting conditions regarding the subjects' perceptions of goal difficulty. The mean goal set in the two conditions was 83.75 (SD = 27.17). However, there was a significant difference between the supportive (M = 3.11, SD=1,23) and nonsupportive (M = 2.48, SD = 1.19) conditions in terms of perception of goal difficulty, F(l, 56) = 5.95, p < .05. This perception was supported by the fact that the goals set by the individuals in the supportive condition were higher than those set in the nonsupportive condition (M = 98.67, SD = 24.01, and M = 70.93, SD = 23.59, respectively); F(l, 56) =25.16, p < .05. No interaction effects were obtained. Goal Acceptance No significant main or interaction effects were found with respect to the three questions on goal acceptance (Cronbach's alpha = .53). The grand mean was 3.68 (SD ā€” .95) on a 5-point scale. Time Length Significant main effects for goal setting, F(2, 84) =41.03, p < . 0 5 , and for supportiveness, F(\, 84) = 25.30, p < .05, were found for the time (seconds) required to explain the study to each subject. No interaction effects were found. Mean time for the supportive condition was 121.24 (SD = 44.53), compared with 90.52 (SD = 35.40) in the nonsupportive condition. Planned t tests revealed a significant difference between the participative and assigned goal setting conditions (M = 144.87, SD = 41.01, and M = 90.59, SD = 30.02, respectively; #(58) = 5.78, p < .05). There was no significant difference between the assigned goal condition and the do best condition (M ā€” 82.20, SD= 25.54).

Performance A 3 X 2 unweighted means analysis of variance revealed a main effect for goal setting only, F ( 2 , 84) = 3.23, p < .05. No interaction effect was obtained. Planned comparisons indicated that participatively set goals led to significantly higher performance (M - 84.37, SD= 18.65) than either assigned goals (M = 71.48, SD = 19.31), # ( 8 7 ) = 2.03, p < .05, or do best goals, (M = 69.67, SD = 32.57), # ( 8 7 ) = 2.33, p < .05. The performance of those with assigned goals did not differ significantly from those individuals with do best goals. There was a significant linear relationship between the actual difficulty level of the goal and the number of ideas generated in the assigned condition, r(2&) = .59, p < .05, the participative condition, r(28) = .67, p < .05, and the combination of assigned and participative goal conditions, r(58) = .61, p < .05. A supportive management style resulted in a mean of 79.16 (SD = 23.38) ideas being generated compared with a mean of 71.18 ideas (SD=23.28) with a less supportive management style. This difference was not significant. Discussion This study provides partial experimental support under well-controlled conditions for Likert's (1967) System 4 theory of management and for Locke's (1968) theory of goal setting. The major findings of the present study follow. A supportive management style leads to higher goals being set than is the case when an authority figure behaves in a less supportive manner. High goals resulted in higher performance than easy goals. The absence of a significant effect for supportiveness on performance indicates that it may not have an independent effect on performance. This finding is consistent with goal-setting theory (Locke, 1968). Supportiveness appears to be important primarily because it gives subordinates and supervisors the confidence to set high goals, which in turn lead to high levels of performance.

GOAL SETTING AND SUPPORTIVE

The finding that there was not a direct relationship between supportiveness and performance may raise questions regarding the external validity of our manipulation of managerial style to organizational settings, particularly in light of the brevity of the study and the relatively high ratings of the experimenter in the nonsupportive condition. Likert (1967) stated that time is a key variable in evaluating organizational outcomes of a System 2 and a System 4 managerial style. In large organizations, he argued, 3 to S years may elapse before outcomes are discernible. He agreed (Likert, Note 1) with the behavior demonstrated in the present study by the experimenter in the role of the System 4 manager (e.g., showing confidence and trust in subordinates, setting goals participatively, and indicating that subordinates should feel free to talk to him), but he questioned the authenticity of the experimenter in the role of System 2, and suggested that the role was too weak. It may be that over time, performance of individuals with a nonsupportive supervisor compared with performance of workers with a supportive manager would be readily observable, but we suspect that the mediating variable would still be goal difficulty. Contrary to previous laboratory experiments (e.g. Latham & Saari, in press) and field experiments (e.g., Dossett et al., in press), specific goals did not lead to better performance than a generalized goal of do your best. Goal specificity only increased performance when the goal was set mutually by the experimenter and the subject. And contrary to previous experiments in which goal difficulty was held constant, participatively set goals led to higher performance than did assigned goals. However, participation did not affect goal acceptance. Apparently, the presence of an authority figure (see Ronan, Latham, & Kinne, 1973) is the key to ensuring goal acceptance; whether the goal is assigned or is participatively set has little or no effect on this variable. The major drawback of adopting a supportive managerial style and allowing individuals to participate in the goal-setting process is time. As Dossett et al. (in press) implied, this variable could prove costly

RELATIONSHIPS

155

when the supervisor has a large number of employees reporting to her. Correlational analyses of field studies, however, indicate that the increase in performance may offset this cost (Likert, 1967). Attention needs to be given to this variable in subsequent field studies. The inconsistency in the findings on participatively set goals between the present study and the previous laboratory study by Latham and Saari (1979) is difficult to reconcile, since the subject population, task, setting, and procedure were the same. The findings of the earlier study were in agreement with previous field experiments (but not with correlational studies) which had shown that participation only affects performance to the extent that it leads to the setting of high goals. When goal difficulty was held constant, assigned goals led to performance that was equal, if not superior, to participatively set goals. The findings of the previous studies suggest that if the present experiment had been conducted in the field, or for a longer duration, the assigned goals might have proven to be as effective as the participatively set goals, given that the supervisor behaved supportively. If this is true, the two key principles to System 4 management are goal setting and the principle of supportive relationships. Nevertheless, the present experiment and the correlational analyses conducted by Likert and his colleagues (Likert, 1967) suggest that participation by itself may affect performance. Why participation in goal setting is important remains elusive. In the present experiment, participation could not affect goal difficulty, and it did not affect goal acceptanceThe only difference between this study and the previous study (Latham & Saari, 1979) was the difficulty levels of the goals set. The goals in this study were set much higher (M = 83.75, SD= 27.17) than in the previous study, (M = 55.0, SD = 25.24), t (48) =2.67, pā€”