Goal setting and task performance...
Psychological Bulletin 1981, Vol. 90. No. I, 125-152 Copyright 1981 by the American Psychological Association, Inc. 0033-2909/81/9001-0125SOO.75 Goal Setting and Task Performance: 1969-1980 Edwin A. Locke College of Business and Managementand Department of Psychology University of Maryland Lise M. Saari Department of Psychology University of Washington Karyll N. Shaw College of Business and Management University of Maryland Gary P. Latham Graduate School of Business Administration University of Washington A review of both laboratory and field studies on the effects of setting goals when performing a task found that in 90% of the studies, specific and challenging goals lead to higher performance than easy goals, "do your best" goals, or no goals. Goals affect performance by directing attention, mobilizing effort, increasing persistence, and motivating strategy development. Goal setting is most likely to improve task performance when the goals are specific and sufficiently challenging, the subjects have sufficient ability (and ability differences are controlled), feed- back is provided to show progress in relation to the goal, rewards such as money are given for goal attainment, the experimenter or manager is supportive, and assigned goals are accepted by the individual. No reliable individual differences have emerged in goal-setting studies, probably because the goals were typically assigned rather than self-set. Need for achievement and self-esteem may be the most promising individual difference variables. In this article we summarize research re- lating to (a) the effects of setting various types of goals or objectives on task perfor- mance and (b) the factors (other than the goals themselves) that influence the effec- tiveness of goal setting. Ail-encompassing theories of motivation based on such concepts as instinct, drive, and conditioning have not succeeded in explain- ing human action. Such theories have been gradually replaced by more modest and lim- ited approaches to motivation. These ap- proaches do not presume to explain all mo- tivational phenomena their domains are more restricted. The study of goal setting is one such limited approach. The concept of goal setting falls within the broad domain of cognitive psychology and is consistent with recent trends such as cog- nitive behavior modification (Meichenbaum, 1977). The present interest of researchers in goal setting has two sources, one academic Preparation of this manuscript was supported by Of- fice of Naval Research Contract N00014-79-C-0680. Requests for reprints should be sent to Edwin A. Locke, College of Business and Management, University of Maryland, College Park, Maryland 20742. and the other organizational. The academic source extends back in time from Ryan (1970) and G. Miller, Galanter, and Pri- brani (1960), through Lewin, to the Wurz- burg School and the associated concepts of intention, task, set, and level of aspiration (see Ryan, 1970, for a summary). The or- ganizational source is traced from Manage- ment by Objectives programs, now widely used in industry (see Odiorne, 1978, for a summary), back to the Scientific Manage- ment movement founded by Frederick W, Taylor (1911/1967). These two strains of thought converge in the more recent work of Locke (1968), Latham (Latham & Yukl, 1975b), and others who have studied the effects of goal setting on task performance. Goal setting is also an important component of social learning theory (Bandura, 1977), which has become increasingly influentialin recent years. Even the literature on organi- zational behavior modification can be inter- preted largely within a goal-setting frame- work (Locke, 1977). Research on goal setting is proliferating so rapidly that recent reviews (Latham & Yukl, I975b Locke, 1968 Steers & Porter, 1974) are now outdated. To provide a longer 125
126 LOCKE,SHAW, SAARI,AND LATHAM term perspective than just the last 6 years, our review includes research published since 1968. Studies that are explicitly clinical and social-psychological in nature are not in- cluded (for a detailed review of the latter, see Fishbein & Ajzen, 1975). review here. We also examine the mecha- nisms by which goals affect action, the ef- fects of feedback, participation, and money on goal-setting effectiveness, the role of in- dividual differences, and the determinants of goal commitment. The Concept of Goal Setting A goal is what an individual is trying to accomplish it is the object or aim of an ac- tion. The concept is similar in meaning to the concepts of purpose and intent (Locke, 1969). Other frequently used concepts that are also similar in meaning to that of goal include performance standard (a measuring rod for evaluating performance), quota (a minimum amount of work or production), work norm (a standard of acceptable be- havior defined by a work group), task (a piece of work to be accomplished), objective (the ultimate aim of an action or series of actions), deadline (a time limit for complet- ing a task), and budget (a spending goal or limit). Earlier attempts by behaviorists to reduce concepts like goal and purpose to physical events have been strongly criticized (e.g., see Locke, 1969, 1972). Goal setting might be called "stimulus control" by a modern be- haviorist, but the key question then becomes, What is the stimulus? If it is only an as- signed goal (an environmental event), then the importance of goal acceptance isignored an assigned goal that is rejected can hardly regulate performance. If goal acceptance is considered relevant, then the regulating stimulus must be a mental event���ultimately the individual's goal. The environment, of course, can influence goal setting as well as goal acceptance, an issue that is dealt with in some of the recent research, The basic assumption of goal-setting re- search is that goals are immediate regulators of human action. However, no one-to-one correspondence between goals and action is assumed because people may make errors, lack the ability to attain their objectives (Locke, 1968), or have subconscious con- flictsor premises that subvert their conscious goals. The precise degree of association be- tween goals and action is an empirical ques- tion that is dealt with in the research we Goal-Setting Attributes' Mental processes have two major attri- butes, content and intensity (Rand, 1967). The content of a goal is the object or result being sought. The main dimensions of goal content that have been studied so far are specificity or clarity (the degree of quanti- tative precision with which the aim is spec- ified) and difficulty (the degree of profi- ciency or level of performance sought). The terms task difficulty and goal difficulty are often used interchangeably,but a distinction between them can be made. A task is a piece of work to be accom- plished. A difficult task is one that is hard to do. A task can be hard because it is com- plex, that is, requires a high level of skill and knowledge. For example, writing a book on physics is a harder task than writing a thank- you note. A task can also be hard because it requires a great deal of effort: digging the foundation for a pool takes more effort than digging a hole to plant a flower seed. Since a goal is the object or aim of an action, it is possible for the completion of a task to be a goal. However, in most goal- setting studies, the term goal refers to at- taining a specific standard of proficiency on a task, usually within a specified time limit. For example, two individuals are given the same task (e.g., simple addition), but one is asked to complete a large number of prob- lems within 30 minutes, and the other, a small number. The harder goal would be achieved by expending greater effort and attention than would be expended to achieve the easy goal. Harder goals, like harder tasks, also can require more knowledge and skill than easier goals (e.g., winning a chess ' Our view of what constitutes a goal attribute differs from that of Steers and Porter (1974) who, for example, called participationan attribute of goals. We treat par- ticipation as a mechanism that may affect goal content or goal acceptance.
GOAL SETTING AND TASK PERFORMANCE 127 tournament vs. coming in next to last). To summarize the distinction between the terms, goal difficulty specifies a certain level of task proficiency measured against a standard, whereas task difficulty refers simply to the nature of the work to be accomplished. Although greater task difficulty should lead to greater effort (Kahneman, 1973 Kaplan & Rothkopf, 1974 Shapira, Note 1), the relation of task difficulty to perfor- mance is problematic. If more work is trans- lated into a goal to get more done, task difficulty may be positively related to performance (Sales, 1970). On the other hand, if harder tasks require more ability or knowledge, most people will, at least ini- tially, perform less well on them, even if they try harder, than they would on easier tasks (e.g., Shapira, Note 1). An experiment by Campbell and Ilgen (1976) demonstrated that the distinction between task and goal difficulty has prac- tical utility. They manipulated both dimen- sions independently, On chess problems dif- ficult goals led to better performance than easy goals training subjects on hard prob- lems (tasks) led at first to poorer perfor- mance but later to better performance than training subjects on easier problems (tasks). Presumably the harder goals led to greater effort than the easier goals, and training on the harder chess problems led to the acqui- sition of more skill and knowledge than training on easier ones. Although there has been extensive re- search on the effects of goal specificity and difficulty on performance, little attention has been paid to two other dimensions of goal content: goal complexity (the number and interrelation of the results aimed for) and goal conflict (the degree to which attaining one goal negates or subverts attaining an- other). The second attribute of goals, intensity, pertains to the process of setting the goal or of determining how to reach it. Intensity would be measured by such factors as the scope of the cognitive process, the degree of effort required, the importance of the goal, and the context in which it is set. Goal in- tensity may be related to goal content for example, a more intense psychological pro- cess is needed to set complex goals and to figure out how to attain them than the pro- cess needed to set and attain simple goals. Goal intensity has not been studied as such, although a related concept, goal commit- ment, has been measured in a number of experiments. Relation of Goal Dimensions to Performance Goal Difficulty In an earlier review of the goal-setting lit- erature, Locke (1968) found evidence for a positive, linear relation between goal diffi- culty and task performance (assuming suf- ficient ability), and more recent studies have supported these findings. Four results in three experimental field studies demon- strated that harder goals led to better per- formance than easy goals: Latham and Locke (1975) with logging crews Yukl and Latham (1978) with typists and a simulated field study by Bassett (1979). In a separate manipulation, Bassett also found that shorter time limits led to a faster work pace than longer time limits. Twenty-five experimental laboratory stud- ies have obtained similar results with a wide variety of tasks: Bavelas (1978), with a fig- ure-selection task Bavelas and Lee (1978) in five of six experiments involving brain- storming, figure selection, and sum estima- tion tasks Campbell and Ilgen (1976) with chess Hannan (1975) with a coding (credit applications) task LaPorte and Nath (1976) with prose learning Latham and Saari (1979b) with brainstorming Locke and Bryan (1969b) with simple addition Locke, Cartledge, and Knerr (1970) in four studies, three with reaction time and one with simple addition Locke, Mento, and Katcher (1978) with perceptual speed London and Oldham (1976) with card sorting Masters, Furman, and Barden (1977) in two studies of 4- and 5-year-old children working on a color dis- crimination task Mento, Cartledge, and Locke (1980) in two experiments using a perceptual speed task Rothkopf and Bil- lington (1975) and Rothkopf and Kaplan (1972) in more complex prose-learningstud- ies than that of LaPorte and Nath (1976) and Sales (1970), using anagrams. In Sales's
128 LOCKE, SHAW, SAARI, AND LATHAM study, task rather than goal difficulty was manipulated by means of varying the work load given to the subjects. Presumably sub- jects developed implicit goals based on the amount of work assigned to them. Ness and Patton (1979) also found that a harder task led to better weight-lifting performance than an easier task when subjects were deceived as to the actual weights. Four studies found conditional2 support for a positive relation between goal difficulty and performance. Becker (1978) with an energy conservation task, Erez (1977) with a clerical task, and Strang, Lawrence, and Fowler (1978) with a computation task, all found that only subjects who had high goals and who received feedback regarding their performance in relation to those goals per- formed better than subjects with low goals. This pattern of results seems also to have been present in Frost and Mahoney's (1976) first study using a reading task (see their Table 1, p. 339). Subjects with high and moderately high goals who apparently re- ceived frequent feedback performed better than those with average goals, whereas the opposite pattern was obtained for subjects given no feedback during the 42-minute work period (interaction p = . 11 / tests were not performed). Six experimental laboratory studies found no relation between goal level and task per- formance. Bavelas and Lee (1978) allowed only 15 minutes for an addition task and gave subjects no information either before or during the task of how fast they needed to go to attain the goal. Frost and Mahoney (1976) found negative results with a jigsaw puzzle task, although their range of goal difficulty was limited: from medium to hard to very hard (actual probabilities of success were .50, .135, and .026, respectively). The same narrow range of difficulty (very diffi- cult to moderately difficult) may explain the negative results of Oldham (1975) using a time sheet computation task. Moreover, not all subjects accepted the assigned goals in that study, and it is not clear that ability was controlled when Oldham (1975, pp. 471- 472) did his post hoc analysis by personal goal level. Organ (1977) also compared moderate goals with hard goals using an ana- gram task. However, since no group average even reached the level of the moderate goal, the hard goal may have been totally un- realistic. The fifth negative study, by Motowidlo, Loehr, and Dunnette (1978), using a com- plex computation task, examined the goal theory-expectancy theory controversy. Goal theory predicts that harder goals lead to bet- ter performance than easy goals, despite their lower probability of being fully reached. In contrast, expectancy theory predicts (other things being equal) a positive relation be- tween expectancy and performance, the op- posite of the goal theory prediction. Motow- idlo et al. found a positive relation between expectancy and performance, which is in agreement with expectancy theory. One pos- sible confounding factor is that the subjects in the Motowidlo et al. study did not make their expectancy ratings conditional upon trying their hardest to reach the goal or to win (pointed out by Mento et al., 1980, based on Yates & Kulick, 1977, among oth- ers). Thus, low expectancy ratings could mean that a subject was not planning to ex- ert maximum effort, whereas high ratings would mean the opposite. This would yield a spurious positive correlation between ex- pectancy and performance. Furthermore, Motowidlo et al. did not provide their sub- jects with feedback regarding how close they were coming to their goals during task per- formance. (The importance of this factor is documented below.) The two studies by Mento et al. (1980), which avoided the er- rors of the Motowidlo et al. study and in- corporated other methodological improve- ments, found the usual positive relation between goal level and performance and no relation between expectancy and perfor- mance. Forward and Zander (1971) used goals set by groups of high school boys on a team- coding task as both independent and depen- dent variables. Success and failure as well as outside pressures were covertly manipu- 2 Partially or conditionally supportive studies were distinguished from nonsupportive studies as follows: A study was called partially supportive if the treatment was significant for one subsample of the full sample of subjects or for one of several experimental treatments or criteria. If an entire sample or study found no sig- nificant effects, it was callednonsupportive.
GOAL SETTING AND TASK PERFORMANCE 129 lated to influence goal setting, which oc- curred before each trial of the task. Under these somewhat complex conditions, goal discrepancy (goal minus previous perfor- mance level) was either unrelated or nega- tively related to subsequentperformance. The results of the experimental studies were, to varying degrees, supported by the results of 15 correlational studies. Andrews and Farris (1972) found that time pressure was associated with high performanceamong scientists and engineers. Hall and Lawler (1971), with a similar sample, found no re- lation between time pressure and perfor- mance but found a significant relation be- tween both quality and financial pressure (implied goals?) and work performance. Ashworth and Mobley (Note 2) found a sig- nificant relation between performance goal level and training performance for Marine recruits. Blurnenfeld and Leidy (1969), in what also could be called a natural field ex- periment, found that soft-drink servicemen who were assigned higher goals serviced more machines than those assigned lower goals. Hamner and Harnett (1974) found that subjects in an experimental study of bargaining who expected (tried?) to earn a high amount of money earned more than those who expected (tried?) to earn less money. Locke et al. (1970), in the last of their five studies, found a significant corre- lation between grade goals on an hourly exam and actual grade earned. The majority of the correlational studies found only a conditional positive relation between goal difficulty and performance and/or effort. Carroll and Tosi (1970) found a positive relation only for managers who were mature and high in self-assurance Dachler and Mobley (1973) found it only for production workers (in two plants) with long tenure (1 or 2 years or more) Dossett, Latham, and Mitchell (1979), found it in two studies of clerical personnel, but only for those who set goals participatively Hall and Hall (1976) found it for the class perfor- mance of second through fourth grade stu- dents in high-support schools and Ivancev- ich and McMahon (1977a, 1977b, 1977c) found it for skilled technicians who had higher order (growth) need strength, were white, and had higher levels of education. Negative results were obtained by For- ward and Zander (1971) with United Fund campaign workers, Hall and Foster (1977) with participants in a simulated manage- ment game, and Steers (1975) with first- level supervisors. All the correlational studies are, of course, open to multiple causal interpretations. For example, Dossett et al. (1979) implied that their results may be an artifact of ability, since ability was considered when setting goals in the participative groups but not in the assigned groups. In fact, none of the cor- relational studies had controls for ability. Also, many relied on self-ratings of goal dif- ficulty or performance. The Yukl and La- tham (1978) study found that only objective goal level, not subjective goal difficulty, was related to typing performance. None of the correlational studies measured the individ- ual's personal goal level, a measure that Mento et al. (1980) found to be the single best motivational predictor of performance. Their measures of subjective goal difficulty did not explain any variance inperformance over and above that explained by objective and personal goal levels. Goal Specificity Specific hard goals versus "do best" goals or no goals. Previous research found that specific, challenging (difficult) goals led to higher output than vague goals such as "do your best" (Locke, 1968). Subsequent re- search has strongly supported these results, although in a number of studies, no distinc- tion was made between groups told to do their best and those assigned no specific goals. The latter were typically labeled no goal groups. We have not found any differ- ences in the results obtained by studies in which no goals are assigned and those in which subjects are explicitly told to do their best. No goal subjects, it appears, typically try to do as well as they can on the assigned task. Twenty-four field experiments all found that individuals given specific, challenging goals either outperformed those trying to do their best or surpassed their own previous performance when they were not trying for specific goals: Bandura and Simon (1977)