TT.‘ k I!" 9' 91"," {1" . gfifl ATTT” L~"‘_.'9 Time T; TTT-‘f f1} :13, m I. . . .giéh L #11:; igfi.‘ . ‘ ,1: . .1}: 3% 1T}: if c I g‘.;’ l: ' ;h:'¥: A‘ 5; F527} 77]" fa} t ' 91.1.17: .:1,::T2:E;.I .9: ‘ a ‘ ‘ aif‘; “15553141119? TT‘T‘TT'T'T ”.H‘ 1., .. 31,331“ ~ A ‘ffigififg‘ZTFHfiaég 'Cfii}. Jig; :{hLL‘ 3f: “‘1" To“; ”'7!!!“ T . :1' 7. :53 TTH . 'T“ “i :TTAQ-i "“"flfg 113.1431”. “11:11.11- 111:1 1- 1 ..... J .43- i: s. .3 '1 #3:.” 1' ' 1.21111TTTT11: 11$ '13?“ 51W «:1: .. ,1 1111.111! TTi‘T TV" ‘TTTT"TTT1.TT’~'52:}JT;FT to ‘. Q‘ I .i 7 IT! 2‘ l- H '- .VLTTW‘T‘DP-wu 5:411:51;th . "g 5,. T . 1 ET“ - .. Ha.tTT-TTITMTMFT TT' £9344 ‘éf. I . ‘ .o H; '1. 0‘9 ._. .o ‘. ‘1 r - . . a. u: .’ 7 . ‘1‘: V1 - :. 4. 1: 3‘11 .. . . ‘1-‘n I a .154 I ,4: quxT-KJ I ‘ :11, Q: 1! ‘33... .N. :.~. 0.. ('3 ‘z N .. ‘23:! ‘.-- o .u..« (Th 'L 'L t» .nv 7". av .- “flu-C. . ~. -A'T‘.'f."‘.r.. . “v o o... C32.- E:: ’3 T :‘3 :fi‘1i‘33? . 27’ 1 2:32‘113TTT‘TTTT‘TTT’Q ‘1. ' . . . I. .‘o .7 ~. 5; . .T .T ‘ ' '0' ’1'? I {._ {.qu 1.1.1. "WATT . x.. mm?” .4313sz .1 513;} 11:;th £11111... g: 43%»,- ‘TVT‘TT‘ "3;:f12‘21216hih‘11é! 3° : T TM: """" lTT‘Tgi‘pT i'fiZ‘ J‘E" .é-‘T‘ : '1 n L 3". m. T 33551.“; ‘T‘T‘TT lava; if??? T713 §J§;;‘QTE§I:T:TT .1 {Tail}: iii: .~.-_, :‘l'fS-‘éé'éfl. 7“ File: A II t It (Ed, him 3 L? 1%! ‘3 Is" :1: £15111}? 2'"? ‘¢€e}; ‘r: 11" PT 1 73¢! \: 7 5‘; 0,1, “F ’2, ’TT’TTTJTTHT'TT' {777‘ F1! $.3'T‘l'331fi‘19 I” ,. . -—-—-— J! - re . u ‘ -_-.n. 1. .~ . o '- 0. up .. a... . 13’.“ '- t o . l . n {no .1. a" a a v - Iv . w... .l o . . I ¢ - v I . - yo. 0 .. u I -‘ l—Jl‘. ‘iw as .....%,r .. o... o 4 M v 4 w>94-,.. u n ”.31.... -0 . . A .. I I. v a?!“ «1’2“ .LJKJ‘L. 3‘: at: _ m... .. _ "I‘ - o~ . . . - ".,.':J . -. as... .0. v I5 I b. C'- q..- ' ‘0‘. Lizzj‘; - - ...- :31: .. 3“: qéfi'O; .n. 32:? #54" to” 232.211.1373: " ‘ 2‘ m. T1. Tm . 111:1 '11‘T*1T~T“1" ' TL NE“! 5'. u- . . m... A “'0‘. . r . 0‘ u .. ‘5‘."1-1 .. {n ,1 K11" ........ " 'i' {’31 ‘ - o'- > 7 r- m7 : ATTT‘TL 1'3“! T T TTTTTT '. o ' l 11 T5“; TT TT 111111 ETTTT O '13 u‘ {TVTT ‘HTTTHTTTTTMTJ TT’ :T- 11411.15 T T ‘T‘TTT‘TTT 1"“ ".TTTTESTT 'TTTTT ”‘1 1‘va ‘TT‘T T“ T 1‘ T" 2.1 111 T 3T T1s"f‘*T WT T TTTT‘T‘TT‘T ‘It TTTTTT TTT’TT TT w T! TLTTTT.WT"1TTT”:T MT. TTT T"’TT12T .' E“; ”a. av"- ”q .nl-o¢ p ' .... u - ‘5.“ - .. _ “I J. «Av IT“; . ,.,, .- l’ “1254 c IIIIIIIIIIIII II II II II II IIIIIIII 98991 I LIBRARY Michigan State University This is to certify that the dissertation entitled Reward Contingency as an Explanatory Variable in the Incentive - Goal Setting Relationship presented by Patrick Michael Wright has been accepted towards fulfillment of the requirements for Ph. D degreein Busuness Administration //////7 W/QM Major professor Date /2,/22:/(97 MSU is an Affirmative Action/Sq ual Opportunity Institution 0-12771 NIJV‘ ”1"”9 I996 I I ‘ '0 9: MSU LIBRARIES m RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. “FM. I ”Mir,- 4;}; , ' ?' ex C5 7 ._ a" ‘7'. ”Mn {:15 1133351 .' WW, 2 5 I993: t JUN 0 9 2011 r h T. 4‘ L: 17 I .l I." «"_ / ‘0.’ Ir5 L4 .2 *I‘? P expectancy. Second, they judge the contingency between performance and outcomes, referred to as the P->O expectancy. Thus, traditional expectancy theory can be viewed as a process involving judgements about E->P->O (Lawler, 1971). In contrast, NPI theory views the process as involving three contingencies diagrammed as a->p->e->o, where a = commitment of resources to an act, p = products produced from an act, e = the evaluation from some other about the products, and o = the outcomes received. The first contingency, a->p involves an individual's perception of the functional relationship between committing additional resources (e.g., time, effort, etc.) and the additional products resulting from that commitment. The second and third contingencies are a more detailed breakdown of the p->o contingency of expectancy theory. The p->e contingency is the individual's belief about the relationship between different levels of products and the level of evaluation from some external source on a positive to negative continuum. Finally, the e->o contingency refers to the individual's perception of the relationship between the level of evaluation and the utility of outcomes ‘ obtained. 10 NPI theory proposes that the motivational process consists of three sequential processing stages. At each stage a prior utility function is converted to a new utility function as one of the contingency relationships is integrated. The first utility function is the "Outcome Valence Function" which represents the individual's belief about the utility of different levels of reward. This is diagrammed in Figure 1a. Like Naylor et al. (1980), a traditional utility function having decreasing marginal utility for both increasing and decreasing amounts of the outcome is used. The second function in the motivational sequence is the e->o contingency function, and is depicted in Figure 1b. For the sake of simplicity, a positive linear function is used indicating that the individual believes that the more favorable the evaluation, the higher the anticipated level of outcome received. Combining these two functions results in the "Evaluation Utility Function" depicted in Figure 1c. V Next, the p->e contingency function (Figure 1d) is combined with the Evaluation Utility Function resulting in the "Product Utility Function" (Figure 1e). This describes the individual's perception about the various levels of outcome utility associated with various performance levels. Finally, combining the a->p function (Figure 1f) with the Product Utility Function results in the "Commitment to Act Utility Function" (Figure 1g). This describes the 11 Magnitude Utility of Outcome (+) (+) Magnitude I Magnitude .of (-) (+) 0f (-?i (+) Outcome Evaluation (-) (~) (3) Outcome Valence Function (b) Evaluation to Outcome Contingency . . Function Magnitude of Utilitv Evaluation (+) (+) Magnitude Amount of (-) (+) 0 (+) of Evaluation Product (-) (-) (c) Evaluation Utility Function (d) Product to Evaluation Contingency Function ." Figure 1. Components of the Motivational Sequences in NPI Theory. (Adapted from Naylor and Ilgen, E19843) 12 Figure 1 (cont'd) Amount of Product Utility (+) (+) Amount 0 (+) of Product (e) Product Utility Function (f) Act to Product Contingency Function Utility (+) Commitment 0 + ( ) to Act M (3) Commitment to Act Utility Function 13 anticipated utility that the individual associates with different degrees of commitment to the act. It is the Product Utility Function which serves as the basis for discussing differences in reward contingency conditions. This utility curve can vary according to the way in which rewards are tied to performance and can thus, describe differences between incentive systems (i.e., piece-rate, goal attainment bonus, and hourly rate). These respective utility curves will be described first when considering pay as the only extrinsic reward. They will then be described when considering the addition of a goal. Finally, they will be depicted with the consideration of costs (i.e. effort and the perception of punishment) to individuals. First, figure 2a depicts the relationship between monetary payoff and performance under a piece-rate reward system. In this situation the individual perceives a strong positive relationship between performance and reward as every additional unit of product results in additional money earned. (The marginal utility curve is not used in this study because the duration of the experimental task was such that it was not likely that individuals would earn enough money for the upper end of the curve to begin to display diminishing returns.) Figure 2b shows the Product Utility Function under a situation where an individual receives a bonus for some level of output. In this condition, the individual receives some set salary level so 14 meowuwncoo zocowcwucoo uum3ou m50flum> qu mcowuoczm muwafiu: uoscoum I N unawflm meosocxm - DBHAHE—«z» ramsom mszom HzmzzHe functions being displaced as now individuals know more clearly what is expected of them in order to receive a favorable evaluation. These authors stated "...the goal setting intervention has its initial and primary effect upon both the motivation and subsequent behavior of the individual by causing a modification, or distortion, in the individual's already existing Cp->e contingency function" (107). This displacement may be due to additional non-monetary rewards gained as a result of 16 the goal, whether they be intrinsic (satisfaction with performance) or extrinsic (the favorable evaluation by an external agent). Because the utility functions are combined multiplicatively, this results in a corresponding change in the Product Utility function. The three conditions are now depicted in Figure 3. Note that there is no reason to expect the additional non-monetary utility would be different across the reward contingency conditions, thus, they are shown as being equal. An additional concept should be considered at this point. Previously we have only considered the positive utility associated with performance. According to Atkinson and Birch (1970), each motivational tendency is the result of two opposing forces--a positive "action" tendency and a negative "negaction" tendency. This negaction works to decrease the action tendency. Thus any costs associated with performance (such as time, effort, or punishment) works to decrease the action tendency. In order to illustrate the importance of this concept, consider the following example. Individuals are working on a task under one of the three following reward conditions: piece-rate, goal attainment bonus, and hourly (non-contingent) pay. The task is of ten minute duration and entails subjects loading 50 pound boxes from a dock 50 feet away into the back of a truck. A goal of 20 boxes is assigned across all three conditions. Assume a measure of utility to be called a "util." To load one box results in a 17 Hmow m wo cowufipvm osu :uw: m:0wuficcoo xucowcwucoo nausea msofium> Low mcofluucnm auHHHu: noncoum I m ouswwm HUDDOmm >4m302 mazom HzmzzHwumwmc mo cowuflcwouou cue: mcofluavcoo aocomcwucoo uum3ou mzowun> HOW mcofiuoczm xuflaflu: Doscoum I q ouswwm meusnomm wqmaom mazom BzmzzHOZMGZEZOO Dm<>>wm >._..50_un=o 1.6.00 Ow20_mm< 24 oocmEcotmd co 3.355 .80 8:936. new xocooczcoo pcmsmm coming case—825 9: Lo BooE oomooocn. - o 050E 32328“.me .200 om20_wm< Z< OE. Pzws—tEEOO .300 4OZmGZ_._.ZOo Dm<>>wm Eumomua $00 8205?. 25 clarify the effect of incentives on choice of goal difficulty level" (p. 83). Locke et al. (1968) found that goal attainment bonus subjects had higher goals than piece-rate subjects, but this result is misleading since the former had goals assigned to them. Furthermore, while Terborg (1976) noted that incentive subjects may have set higher goals than no-incentive subjects, he lacked the data to test this hypothesis. Given the lack of research on the effect of reward contingency on goal level, NPI theory can be used to make some predictions. The negative utility associated with failure to attain a goal in the goal attainment bonus condition would likely influence subjects to set low goals (i.e., goals that they are sure to attain). Locke and Latham (1984) pointed out that one risk inherent in paying people for goal success is that "To the extent that they have influence over what goals are set, they may be motivated to set easy goals..." (p. 114). Piece-rate subjects, on the other hand, would be influenced to set high goals, as higher performance is associated with higher rewards. Subjects under the hourly pay condition would have no extrinsic reasons for setting either high or low goals. Thus, the following hypothesis is proposed: Hypothesis 1 Reward Contingency will exert an effect on personal goal level choice. Subjects in the Goal Attainment Bonus condition will set the lowest goals and subjects in the piece—rate condition will set the highest goals with hourly 26 pay subjects setting goals between the other two groups. Determinants of Assigned Goal Commitment Mowen et al (1981) found that reward contingency conditions moderated the relationship between assigned goal level and performance, as subjects assigned difficult goals in the goal attainment bonus condition displayed lower performance than those assigned difficult goals in the piece-rate condition. In essence a linear relationship was observed between goal difficulty and performance in the piece-rate condition, and a curvilinear (inverted U) relationship was observed in the goal attainment bonus condition. Locke et al. (1987) stated that this lower performance could have been due to their lack of acceptance of the assigned goal. While this explanation seems reasonable, their rationale is problematic. Locke et al. (1987) gave no explanation as to why these subjects would have lower acceptance than those individuals assigned the same goals in the piece-rate condition. In fact, expectancy theory would predict the same level of goal commitment across the two conditions. In addition, this runs counter to the argument made by Locke et al (1981) that money increases goal commitment. According to Naylor and Ilgen (1984) two conditions must be met in order for goals to be accepted. First, the individual must believe that the goal is a reality. The reality issue is determined by whether or not the goal truly displaces the p->e contingency. Second, the individual must believe that he or she can attain the goal. 27 The attainability issue is concerned with the individual's a->p contingency. In the goal attainment bonus condition, the goal (as mentioned before) would have a more profound displacement effect on the Product Utility Function than in the piece-rate condition. This stems from the fact that the individuals e->o as well as p->e contingency function would be distorted in this condition. Thus, in Naylor and Ilgen's (1984) terms, it may heighten the reality of the goal. With this heightened salience of the goal, the attainability issue may become more important. In situations where no rewards were given for performance, even unattainable goals have been accepted and resulted in increased performance (Garland, 1983; Locke et al. 1984). In the situation where a piece-rate incentive system exists, goals with a low probability of attainment (and possibly even unattainable goals) might not be rejected since performance below the goal still results in a monetary payoff for increased performance. Under a goal attainment bonus condition, however, increases in performance at all levels below the goal results in no additional monetary payoff. This would make individuals in this situation less amenable to accepting a goal which they perceive as not being attainable. Thus, as the goal becomes more difficult, subjects in the goal attainment bonus condition would be less likely to accept the goal than those in other conditions. 28 While research on the issue of goal commitment has most often concerned the initial acceptance of a goal, Locke et al. (1987) and Hollenbeck and Klein (1987) stated that commitment also refers to the tenacity with which one holds a goal. Thus, an individual may accept a goal before performing the task, but reject that goal at some point during the task. This is another possible explanation for the findings of Mowen et al. (1981). Subjects assigned difficult goals in the goal attainment bonus condition may have realized that success was impossible at some point during the task and quit. This would reconcile the conflicting findings of Mowen et al. (1981) and Campbell (1984) who found that goal contingent payment resulted in higher performance than non-contingent or no payment. The task in the Campbell (1984) study did not provide feedback on performance until after the task had been performed and subjects would therefore be less likely to quit. In the Mowen et al. (1981) study, subjects had some knowledge as to their progress toward the goal. Finally, Locke et a1. (1987) hypothesized that having individuals set a personal goal before being assigned a hard goal might lower their commitment to the assigned goal. They stated "the deleterious effect of self-set goals preceding assigned goals will be most marked when the assigned goals are more difficult than the self-set goals." (I). 16). 29 Locke et al. (1987) pointed out that this notion of the discrepancy between personal goal and assigned goal as a determinant of commitment is problematic. Goal "rejection" by subjects in easy goal conditions may not be the same as goal "rejection" in difficult goal conditions. Rejection in the first case may entail setting a harder additional goal. Thus, while personal goal might influence commitment to an assigned goal, it is also likely that assigned goal level will moderate the relationshipbetween personal goal and commitment to an assigned goal. Thus, the following hypotheses are proposed with regard to the determinants of goal commitment: Hypothesis 2a Personal goal level will moderate the relationship between assigned goal and commitment to an assigned goal. Hypothesis 2b The effect of reward contingency on commitment to an assigned goal will be through the moderated mediating effect of assigned goal level and personal goals. Determinants of Performance As was previously mentioned, Locke (1968) stated that incentives may affect performance through the mediating effect on goals. Integral to Locke's (1968) theory of goal setting was the notion of acceptance of or commitment to goals. He stated that only difficult goals that were accepted would motivate higher performance. Although this notion was central to his theory, Hollenbeck and Klein (1987) pointed out that recognition of this is lacking in the empirical studies to date. They stated that of the 109 published studies they reviewed, only three have measured goal commitment and 30 tested its role as a moderator of the goal difficulty-performance relationship. Thus, the following hypotheses are proposed: Hypothesis 3a Goal commitment will moderate the relationship between assigned goal level and performance. Hypothesis 3b Reward contingency will moderate the relationship between goal difficulty and performance. This moderating relationship will be such that a linear relationship will be observed between goal level and performance for the piece-rate and hourly rate groups but the relationship will not be linear for the goal attainment bonus condition. METHOD Overview This study examined the effects of reward contingency and assigned goal level on personal goal level, commitment to an assigned goal, and performance. The design consisted of manipulating reward contingency (three levels) and assigned goal level (two levels). Subjects were assigned an easy, moderate or difficult goal, and performed the experimental task under one of the following reward conditions: Piece-rate, hourly rate, or goal attainment bonus. Dependent variables were personal goals, commitment to an assigned goal, and performance. Tit-fl The task consisted of computer card sorting. Each card contained information on three demographic variables: income (three levels), sex (two levels) and marital status (two levels), allowing for twelve possible configurations. Subjects were required to read each card and then place it into the proper pile on a sorting board. The cards were punched with three holes corresponding to the information on the card. Pegs were present on the sorting board corresponding to each of the twelve possible configurations of holes. Thus, the actual sorting task consisted of fitting the cards onto the proper set of pegs. This task was chosen for two reasons. First, due to the motivational nature of goal setting, the task allowed for performance to reflect the effects of differential 31 32 motivation, as opposed to a task where required ability accounts for so much variance as to mask motivational differences. Numerous studies on goal setting have used this task and have shown the goal setting effect (e.g., London and Oldham, 1976; Pritchard and Curts, 1973; Rakestraw and Weiss, 1981). On most goal setting tasks, performance can vary in terms of either quantity or quality. Bavelas and Lee (1978) showed that subjects sacrificed quality while trying to meet a quantity goal. The fact that this task eliminates variance on the quality dimension was the second reason for choosing it. Since subjects were required to sort cards by placing them on the wooden pegs, performance quality differences are impossible because an incorrect placement of a card immediately becomes evident by its failure to fit. Subjects performed the task for two 20 minute periods. The first trial was a practice trial during which subjects were asked to sort as many cards possible. The purpose of this trial was twofold. First, it gave them some idea of how many cards they could reasonably sort in a 20 minute period. Second, it provided them enough practice to begin to approach their ability peaks. (London and Oldham [1976] found practice effects across three seven minute trials.) It was hoped that this, combined with the monotony of the task would make any performance differences across 33 experimental groups due mainly to motivational differences caused by the interventions. Subjects Subjects were 243 students from Michigan State University recruited from two large management classes. A majority were business majors and all were sophomore level and above. They received course credit for participation, and were told that participation would enable them to make between $5 and $15 for a one and one-half hour time commitment. Manipulated Variables Assigned Goal The assigned goal consisted of two levels operationalized as a goal at either the 67th or 99th percentile of the performance of a pilot study group on their second trial. The goal at the 67th percentile will be referred to as a "moderate" goal and can be thought of as difficult, but attainable. The goal at the 99th percentile will be referred to as a "difficult" goal and can be considered unattainable for almost all subjects. Reward Contingency, The method of payment for subjects depended upon the condition. Subjects in the piece-rate condition were paid $.03 for every card they sorted. Each subject in the non-contingent condition was yoked to a subject in the piece-rate system, and paid an hourly rate equal to that earned by their yoked partner. The goal attainment bonus condition subjects received a $5.00 salary plus a bonus of $5.05 (moderate goal) or $7.00 (difficult 34 goal) if they attained their goal. In addition, they received $.03 for every card they sorted above the goal. This made their pay equal to that of subjects in the piece-rate condition who were performing at or above their goal level. It should be noted that this operationalization of the piece-rate and goal attainment bonus conditions is similar to that used by Mowen et al. (1981). The Product Utility Curves for each of the three reward contingency conditions with three goal levels are depicted in Figure 7. Note that these curves only recognize monetary payoffs, and are meant only to graphically illustrate how subjects earned money in the study. Operationalizing the pay systems in this way accomplished two objectives. First, it made the expected monetary payoffs equal for all subjects across the contingent pay conditions (peice-rate and goal attainment bonus) performing at or above their assigned goal level. Equality between piece-rate and non-contingent subjects was achieved by yoking. Second, it made the potential pay equal across all goal levels for the contingent reward conditions. Subjects assigned moderate goals could work beyond their goal level and those that achieved performance at the difficult goal level received the same reward as those assigned difficult goals who achieved them. This eliminated the potential . confounding of amount of reward with assigned goal level. 35 A>Hu>wuumdmou meow ufizuwwMHv won ounuovoe ou wcficcoammuuou ox can Exv Amwwozmd aunuocoa aaco wcwuflcwououv mcofluficcou Housmaflumnxm m50Hum> mom mooaumaaqwcms cofluocsm auflku: noncoum I n unawam u n m ex Ex ox Ex ex ex I _ I b b h a a a a q d _ _ w . \ H. p \ H \. A x H H D yamso: mszom Hzm22H vo>aomno momnuupo> Houeoaguonxu u:oe< mcoquoaouuoououcu can .moucnm .mcouuc«>oa unaccoum .cooz H mqmp contingency in determining goal commitment (Naylor and Ilgen, 1984). The fact that assigned goal level and personal goal level both were related to goal commitment but in opposite 64 directions leads to the possibility that it was in fact the discrepancy between these two variables that predicts commitment. In other words, the more congruent the assigned goal is with the personal goal, the higher the commitment. This alternative formulation was tested by computing a discrepancy measure between assigned goal difficulty and personal goal and correlating this measure with goal commitment. This discrepancy accounted for the same amount of variance (25.7%) as the combination of personal goal and assigned goal tested in hypothesis 2a, lending some support for this idea. The analySis regarding Hypothesis 2b and the mediating role of personal goal in the incentive - goal commitment relationship provided some interesting findings regarding the incentive - goal commitment process. First, as was already discussed, incentives affected personal goals, and personal goals were directly related to goal commitment. Thus, reward contingency has an indirect effect on goal commitment through its effect on personal goals. Remember that this effect was such that piece-rate subjects had the highest goals, followed by goal attainment bonus subjects, and finally hourly subjects. This effect on goal commitment would present similar rank orderings. Second, the results also supported the proposition that incentives directly affect goal commitment (Hollenbeck and Klein, 1987; Locke et al. 1981; Locke et al. 1987). Goal attainment bonus subjects expressed the highest 65 commitment, followed by piece-rate and hourly condition subjects respectively. This finding is important in that it provides empirical support for this much discussed, but previously unresearched idea. This finding was also important in that the direction of the effect was not parallel to that of the effect on personal goals and, in fact, was exactly opposite to that which Locke et al. (1987) used to explain the results of Mowen et al. (1981). Mowen et al. (1981) found that subjects assigned difficult goals in the goal attainment condition had the lowest performance, and Locke et al. (1987) reasoned that this was due to their lower goal commitment. In the present study, however, subjects in the goal attainment bonus condition expressed significantly higher levels of commitment. Naylor and Ilgen (1984) suggested that goal acceptability is determined by an individual's beliefs about whether or not s/he could attain the goal and by his/her beliefs about the reality of the goal. The former determinant was supported by the positive relationship between personal goal and goal commitment and the negative relationship between assigned goal difficulty and goal commitment. Little attention, however was paid to the effect of the reality issue in formulating the hypothesis. It should be remembered that the reality of the goal is determined by the extent to which the goal distorts the Cp->e function. It was hypothesized that the different 66 reward contingency conditions would distort the entire Product Utility Function heightening the reality of the goal. (Remember that this function describes the individuals' beliefs about what levels of Utility are associated with various performance levels.) Accompanying this heightening of the reality of the goal, it was proposed that the attainability of the goal would become more important. This reasoning ignored the potential direct impact of the Product Utility Function on goal commitment in addition to the indirect impact through the heightened salience of attainability. The higher commitment of goal attainment bonus subjects shows the potentially important role of the Product Utility function in determining goal acceptance. Naylor et al. (1981) proposed that the glgpg of the Utility Curve acts as the major motivational mechanism. Although the present study presents no concrete data, these results lend support to the potential role of that slope in determining goal commitment as well as performance. As can be seen in figure 7, the slope of this curve is steepest for the goal attainment bonus group, followed by the piece-rate and hourly groups, respectively. The ranking of the groups in terms of goal commitment parallels that of the ranking of the groups in terms of steepness of slope of the Product Utility Curve around the assigned goal. One issue deserving attention is that of the potential rival hypothesis that the amount of the bonus was a large 67 determinant of commitment. It should be remembered that this was presented as a potential confound but was deemed to be unimportant because it would have worked in a direction opposite that of the hypothesis. Due to the fact that the hypothesis was not supported, the question arises as to whether or not the confound was at work. If the confound had an effect that effect would have been that commitment and performance in the goal attainment bonus groups would have increased (rather than decreased) relative to the other incentive groups when comparing moderate and difficult goals. Empirically, this would have appeared as a significant reward contingency X assigned goal interaction. To test this, commitment was regressed hierarchically on reward contingency, assigned goal and the interactions. No significant interactions were observed. As was already mentioned, however, range restriction existed in the goal difficulty variables. Thus, while no evidence indicates that the confound existed, this conclusion should be considered tentative. Another potential criticism relevant to the amount of the bonus is that the goal attainment bonus subjects were getting larger bonuses than piece rate subjects for goal attainment (to sort the goal attaining card paid piece-rate subjects $.03 while goal attainment bonus subjects earned an additional $5.05 or $7.00). However, it is important to note that the total amount of reward was equal across the ~two groups at that level of performance. The slope of the 68 Product Utility Function would have been different across the conditions. Remember that the main contention of the study was that it is the slope rather than just the amount that can affect goals, commitment and performance. This once again supports the proposition that the reward contingency must be considered when studying the role on incentives in determining goal commitment. Determinants of Performance Support was not shown for Locke's (1968) original proposition that goal commitment moderates the relationship between goal difficulty and performance. Rather, goal commitment exerted a direct positive effect on performance in this study. These results resemble those found by other researchers (Earley, 1985; Barley and Hulin, 1985; Erez and Zidon, 1984; Hollenbeck, Williams and Klein, 1987; Locke, Frederick, Buckner and Bobko, 1984). These findings also support Locke et al.'s (1987) formulation of commitment having a positive relationship with performance. This deviation from the hypothesis stemmed from utilizing only difficult goals (as was previously mentioned). Hollenbeck et al. (1987) noted that the moderating effect of goal commitment takes place only when the entire range of goals is present (easy, moderate, difficult) such that performance is high only when both goal level and goal commitment are high. When, on the other hand, only difficult goals are being utilized a positive relationship between goal commitment will be 69 observed. Thus, in their study which explored the determinants of commitment to difficult goals, goal commitment accounted for 13 percent of the variance in future academic performance. In addition, Earley, (1985b) reported that goal commitment was related to performance within each of a number of goal difficulty levels, but this relationship was not observed when collapsed across all difficulty levels. This implies that the goal levels utilized in a goal setting study have implications for the type of relationship that will be observed between goal commitment and performance. Since the present study utilized only difficult (67th and 99th percentile performance) goals, it is not surprising that the direct positive relationship between goal commitment and performance was observed rather than the original moderating effect proposed by Locke (1968). A troubling finding with regard to the determinants of performance was the failure to replicate the relatively strong research findings of a positive relationship between goal difficulty and performance (Mento et al, 1987; Tubbs, 1986). It should be noted that both the Mento et al. (1987) and Tubbs (1986) studies consisted of meta-analyses of the goal difficulty-performance relationship. This procedure entails cumulating effect sizes across a number of studies to determine the "true" population relationship 70 between those variables (Hunter, Schmidt and Jackson, 1983). With the effect size as the variable of interest in any meta-analysis, one must be cautious in interpreting the results. This caution stems from the fact that one important determinant of the effect size in experimental research is the levels at which the experimental variable (in this case the goals) is set (Kerlinger, 1973).. In fact, Kerlinger stated in support of maximizing effect sizes when designing research to "Design, plan, and conduct research so that the experimental conditions are as different as possible" (p. 308). Thus, one must recognize that the observed relationship between goal difficulty and performance in any experimental study is partly a function of the levels at which the goal difficulty construct is operationalized. Because the present study was designed to extend the Mowen et al. (1981) study, it is necessary to look at how these researchers operationalized goal difficulty. They set goals at performance levels that either 100% (easy). 50% (moderate), or 0% (difficult) of the pilot subjects had attained (i.e., the lst, 50th and 99th percentiles). The present study, on the other hand, operationalized goal difficulty at only two levels, those being at the 67th and 99th percentiles of pilot study subjects. This restriction of range may be the cause of a failure to detect a 71 significant positive relationship between goal difficulty and performance. This problem also was evident in Campbell's (1984) study. He hypothesized that his failure to find the goal difficulty - performance relationship was due to the fact that both the moderate and difficult goals were in fact difficult goals. In support of this, he cited the fact that very few subjects attained either goal. This prompts the question as to what, exactly, is meant by goal difficulty. Relevant to this discussion is the need to distinguish between the various operationalizations of goal difficulty in past research. The construct of goal difficulty in past research has been operationalized in three ways. First, many studies have utilized absolute goal levels (e.g. the 335 and 400 cards in the present study). For the sake of the following discussion this operationalization will be referred to as "goal level." Second, some studies have operationalized goal difficulty as goal levels which certain percentages of the subject population could be expected to attain (e.g. the 67th and 99th percentiles in this study). This operationalization will be termed "normative goal difficulty". Finally, some studies have operationalized goal difficulty as a discrepancy score, usually as the difference between the goal and an individual's past performance. This operationalization will be called "subjective goal difficulty" because it describes the 72 difficulty of the goal to the individual. Because goal setting theory is an individual level theory, it seems that this operationalization tends to be the most accurate as an operationalization of goal difficulty. To illustrate the differences between these operationalizations, Figure 8 displays a normal distribution with a mean of 500 and a standard deviation of 100. In addition, the standard error distributions for low, medium and high ability individuals are also shown. This figure displays the fact that goal level differences may not be the same as normative goal difficulty differences. Although the difference between 500 and 600 units equals the difference between 700 and 800 units with regard to goal level, this is not true for normative goal difficulty. According to the normative operationalization, approximately 34% few subjects would attain the higher goal in the first case, but only 2% fewer would attain the goal in the later case. Similarly, goal difficulty according to goal level and normative operationalizations may differ from a subjective operationalization. A study using goals of 500 and 600 units would likely find a goal difficulty effect using either of the former operationalizations, but these goals would have different subjective difficulty meanings to different subjects. Both goals would be difficult for the low ability subject (although the latter would be more SOIr . only the latter goal would be difficult for the medium 73 com muoofibsm 54:22 fine ago .938: .33 non ocoflofltog uowum vuwucmum saws mocmshowuom mo cowusnwuumfia Hushoz I m ouswfim I x Z N A x ooh coo com ooq com CON I>< 74 ability subject, and neither would be difficult for the high ability subject. The importance of this discussion relative to the present study lies in explaining why no relationship was observed between goal difficulty and performance. Although the goal levels may have differed substantially (335 vs. 400), according to the normative difficulty operationalization the goals were not that different. In the present study, none of the subjects performed on the practice trial above the goal that they were subsequently assigned. It is reasonable to assume that the average subject (having sorted 216 cards on the first trial) would view even the moderate goal of 335 as quite difficult. In addition, only 17% of the subjects in the moderate goal condition and approximately 3% of the subjects in the difficult goal condition performed above their goal level on the experimental trial. In the sample as a whole, only 10% of the subjects actually exceeded their goal. Thus, a restriction in range in goal difficulty by normative standards may have existed. If, in fact, the failure to detect a relationship between goal difficulty and performance stems from this hypothesized restriction in range, than one way to eliminate this range restriction would be to utilize a subjective goal difficulty operationalization. This would consist of using the discrepancy between the assigned goal and the subjects' performance on the practice trial as a 7S measure of goal difficulty. Thus, rather than having only two levels of difficulty, many levels are present, and these levels reflect the individual's subjective goal difficulty. One additional problem is created by doing this, however. Due to the fact that a strong correlation was observed between the two performance trials (r=.81), then to correlate the discrepancy measure of goal difficulty with the pure performance measure would result in a strong negative relationship between the two variables (i.e. subjects who had high ability would have a lower discrepancy score but higher performance on the experimental trial). Thus, performance in this case was also operationalized as a gain score. Additional analyses were performed operationalizing goal difficulty and performance as discrepancies between the assigned goal and practice trial performance and performance and practice trial performance respectively. This resulted in an observed correlation of .28, consistent with the often observed positive relationship between goal difficulty and performance. Since the major problem with using discrepancy scores to operationalize variables stems from their unreliability (Rogosa et al. 1982) a correlation of .28 between two discrepancy scores may underestimate the true relationship between the two variables.' This result lends support to the idea that range restriction in assigned goals was a problem in the study. 76 In light of this, the hypotheses dealing with the effect on goal difficulty on goal commitment and performance were retested using these measures. These results are displayed in Tables 12 and 13. These analyses showed no further support for the hypotheses, and differed from the previous analyses only in that goal difficulty accounted for 22 percent of the variance in goal commitment and 8 percent of the variance in performance. A problem that arises when comparing these results to those of Mowen et al. (1981) is that the performance decrement due to the goal attainment bonus condition observed in the latter study was not observed here. This divergence might, however, support the Locke et al. (1987) explanation that the performance decrement in the Mowen study was due to lower goal commitment. In this study, goal attainment bonus subjects were more committed to the goal than piece-rate subjects, and this commitment was directly related to performance. Thus, if in fact subjects were less committed to the goals in the Mowen study, and commitment was related to performance, then the Locke et al. explanation would be supported. The main question to be answered concerns why subjects in that study would be less committed to goals than subjects in the present study. One explanation may be that the differences were due to differences in the valence or utility of the incentives. The Mowen study used poker chips which could be redeemed 77 TABLE 12 Results of Regressing Goal Commitment on Personal Goal, Goal Difficulty and the Interactiona’b Hierarchical 2 2 Step Variable R P AR P of 1 Personal Goal .03 .01 .03 .01 2 Goal Difficulty .27 .01 .24 .01 3 Pets. Goal x Goal Diff. .27 .01 .00 N.S. aN—219 bPersonal and Goal Difficulty Re-operationalized Commitment - .02 Personal Goal -.05 Goal Difficulty + .00 Interaction + 37.76 78 TABLE 13 Results of Regressing Goal Commitment on Personal Goal, Goal Difficulty and the Interaction Hierarchical Step Variable R2 P AR2 P of A 1 Goal Difficulty .08 .01 .08 .01 2 Goal Difficulty .12 .01 .04 .01 3 Pers. Goal x Goal Diff. .12 .01 .00 N.S. aN—219 bPerformance and Goal Difficulty Re-Operationalized Performance - .07 Goal Difficulty + .73 Goal Commitment + .00 Interaction + 13.25 79 for school supplies. This type of incentive might not be evaluated as having as great a utility as monetary incentives. The monetary incentive used in this study could be redeemed for a wider variety of valent objects (e.g., food, clothes, gas, etc.). It is also likely that the total value of redemption may have been higher (i.e., subjects could buy a larger amount of school supplies) in the present study. If this is true, then the Product Utility Function would be less steep in Mowen study than in the present study. If the steepness of the Product Utility Function does, in fact, act as one determinant of goal commitment, this would explain why the Mowen subjects would be less committed to the assigned goal. Subjects in the Mowen study may have been more likely to reject or abandon the assigned goal since the utility in goal attainment was not that high. This might be consistent with Pritchard and Curts' (1976) finding that the amount of reward plays a role in the effectiveness of goals for influencing performance. While recognizing that this explanation is purely speculative, the differing results between the two studies certainly displays the need for further research into the mediating role of goal commitment in the incentive - performance relationship. 80 Conclusions About the MediatinggRole of Goals Although the purpose of this study was to test hypotheses specifically with regard to reward contingency, data relevant to Locke (1968) and Locke et al.‘s (1981) hypotheses regarding the mediating role of goals in the incentive - performance relationship should be discussed. As stated before, these authors believed that goals mediated the incentive - performance relationship through spontaneous goal setting, affecting goal level, and/or affecting goal commitment. The present study tested the latter two propositions. In order to evaluate these hypotheses, one must first understand the concept of a mediating relationship. James and Brett (1984) pointed out that a complete mediation model has the form x->m—>y where y is the antecedent, m is the mediator and y is the dependent variable. The relationship can be defined as follows: " g is a mediator of the probabilistic relation y=f(x) if m is a probabilistic function of g and y is a probabilistic function of m" (p. 310). This relationship in the past has been tested empirically by establishing a relationship between the antecedent and consequence, and then showing that this relationship disappears when the mediator is controlled for. James and Brett (1984) stated that all mediation models have in common the attribute that the mediator transmits influence from an antecedent to a consequence. 81 They, however, also point out that this transmission need not involve all of the influence of the antecedent on the consequence. Given this definition of a mediator, support has been shown for the mediating role of goals in the incentive - performance relationship. Hypotheses 1 and 2b showed that reward contingency (incentives) affected personal goal level and goal commitment respectively. Hypothesis 3a showed the relationship between goal commitment and performance, and Table 1 shows the correlation between personal goals and performance. (Additional analyses also showed that personal goal was related to performance even after controlling for goal commitment). In summary, this study supported two of Locke et al.'s (1981) hypotheses for how goals mediate the relationship between incentives and performance. This should provide a basis for additional research on the these two as well as the third proposition. Potential Limitations of the Study Locke et al. (1987) discussed indirectly measuring goal commitment by asking subjects their personal goals. Earley (1985) used this indirect measure of goal commitment and found that it correlated .76 with a self-report questionnaire measure of goal commitment. In light of this, one might question the construct validity of using personal goal as a predictor of assigned goal commitment. 82 The present study did not use personal goal as an indirect measure of assigned goal commitment. The study was designed such that individuals were asked to report their own personal goal before knowing what goal they would be assigned. This measure of personal goal was influenced by subjects' past performance and the incentive condition. Although reward contingency affected both personal goals and assigned goal commitment, the effect was different for the two dependent variables as piece-rate subjects expressed the higher goals but lower goal commitment than goal attainment bonus subjects. Schwab (1980) notes that one step in the construct validation process is to show that the variable of interest relates differently to an independent variable than a similar but theoretically distinct construct. This provides support for the contention that personal goals in this study were more than an indirect measure of assigned goal commitment. One apparent problem with this study was the nature of the task. Although the task was chosen to maximize the observability of motivational differences, it seems likely that ability played a substantial role in performance. First, the task is similar to pegboard ability tests used in selection decisions. In addition, past research successfully utilizing this task have used a much shorter trial period (7 or 10 minutes). The two twenty minute trials likely caused fatigue, thus, adding to the ability (endurance) component in performance. 83 One may also question the external validity of this study due to its laboratory setting. This issue is especially relevant due to the recent debate in the organization behavior literature about the use of laboratory research (Gordon, Slade and Schmitt, 1986,1987; Greenberg, 1987). Two issues are relevant to this debate. First, one must ask whether the student sample utilized in the present study provides any generalizability power. Second, one must determine if generalizability is the goal of this study. With regard to the first issue, Dipboye and Flanagan (1979) reviewed the subject populations utilized in studies published in some of the major organizational behavior journals. These authors found that the assumption that research conducted in field settings is inherently more generalizable than research conducted in laboratory setting is oversimplified, and in fact, possibly erroneous. They stated "field research in industrial-organizational psychology has dealt with a rather narrow subset of settings, actors, and behaviors (p.146)." They concluded that studies should not be accepted or rejected because of the setting, but that a careful examination of the organizations, people, and responses sampled should determine the possible limits on external validity. Greenberg (1987) proposed a strategy of conducting numerous studies on different sets of homogeneous 84 subpopulations, and determining the "boundary conditions" (Fromkin and Streufert, 1976) of any theory based on how the theory predicts behavior in each of the different studies. According to him, differences between subject populations may not only be expected but provide a valuable source of information. He stated "...the assumption that nonstudent samples allow for broader generalizability than samples of students appears unwarranted (p. 158).". Thus, while recognizing the limited generalizability of the results to other populations, there is little reason to believe that the sample was less representative than a nonstudent sample. Thus, this study may provide one necessary link in the overall chain of establishing the generalizability of the Locke et al. (1981) propositions. With regard to the second issue, Berkowitz and Donnerstein (1982) questioned whether or not generalizability should be a goal of all research. They recognized that in order to establish population parameters a representative sample is a must. However, these authors pointed out that laboratory experiments are mainly oriented toward testing some causal hypothesis, and are not carried out to determine the probability that a certain event will occur in a particular population. Mook (1983) more specifically questioned that assumption that generalizability is the goal of all research, stating that it is the purpose of the research that dictates the importance of the representativeness of a 85 sample. He pointed out that when the purpose of research is to apply findings to the real world, then studies should be designed so as to identify the target population and then select a random sample from that population. He also noted, however, that some research is designed to test a theory rather than to generalize to some population. In these cases, representativeness of the sample is a trivial issue since the intended conclusion is about a theory and not a population. He stated that "external validity...is a concept that applies only to a limited subset of the research we do (p. 386)." The purpose of the present study was not to establish population parameters, nor specifically to apply the findings to the general population. The purpose of the present study was to test Locke et al.'s (1981) often discussed, but previously untested propositions about how goals might mediate the relationship between incentives and performance. To the extent that no theoretical reason exists to expect that these propositions might apply gply to real world subjects working in real world situations, and not to student subjects working in a laboratory setting, this study provides a vital step toward a better understanding of how goals mediate the incentive-performance relationship, and generalizability becomes less of an issue. 86 Summary and Conclusions This study explored the mediating role of goals and goal commitment in the incentive - performance relationship. While not achieving its aim to reconcile conflicting past research findings with regard to the interaction of goals and incentives in determining performance, this study provides two concrete additions to the literature on incentives, goals and performance. First, it presented a theoretical base for further exploration into the incentive - goal - performance relationship. NPI Theory (Naylor et al. 1980) was used to depict the various types of incentives. These incentives were characterized by contingency relationships and labeled reward contingency. This theoretical framework aids in better understanding the process through which goals and incentives interact to determine performance as opposed to the current reliance on traditional expectancy theory (e.g. Porter and Lawler, 1971). Second, this study shed some empirical light on the issue of how goals mediate the incentive - performance relationship. Although researchers have often hypothesized about how incentives might affect goal level choice (Campbell, 1982) or goal commitment (Hollenbeck and Klein, 1987; Locke, 1968; Locke et al. 1981) very little data exists, and that which does exist is of dubious value. In the present study, reward contingency was shown to affect both personal goals and goal commitment as hypothesized by Locke (1968) and Locke et al. (1981). As was discussed, 87 however, the effect of reward contingency on goal commitment was in the opposite direction of that discussed by Locke et al (1987). The results observed here elicit some interesting questions to be answered by future research in this area. The theoretical framework and empirical results presented here hardly comprise the answers to all of the questions that have been asked about the role of incentives and goals in determining performance. They do, however, fill a long existing void, and form an important starting point in the quest to better understand the relationship among goals, incentives and human performance. APPENDIX A 88 MANIPULATION CHECKS REWARD CONTINGENCY Which of the following statement best describes what your pay was based on? Put a check next to the one that best describes it. 1. I was paid based upon attaining the goal assigned to me. 2. I was paid based upon how many cards I sorted. 3. I was paid just for participating on the trial. ASSIGNED GOAL What was the goal that was assigned to you by the experimentor? cards on the twenty-minute trial. 89 PERSONAL GOAL Having been told what your pay will be based on. I'd like you to set a personal goal that you think would be a reasonable goal for you to work toward on the next trial. What is that goal? cards on the twenty-minute task 90 GOAL COMMIMBNT (PR8) INSTRUCTIONS Please rate how strongly you agree or disagree with the following statements regarding the performance goal that was assigned to you. Circle the appropriate number after each item based on the scale below. Please respond to gIIIstatements. 1 2 3 4 5 Strongly Disagree Uncertain Agree Strongly Disagree Agree 1. I am strongly committed to pursuing this performance goal 1 2 3 4 5 2. I am willing to put forth a great deal of effort in order to achieve this goal 1 2 3 4 5 3. Quite frankly, I don‘t care if I achieve this goal or not. ‘ 1 2 3 4 5 4. There is not much to be gained by trying to achive this goal. 1 2 3 4 5 5. It is quite likely that this goal may need to be revised, depending on how things go on the trial. 1 2 3 4 5 6. It wouldn't take much for me to abandon this goal. 1 2 3 4 5 7. It's unrealistic for me to expect to reach this goal. 1 2 3 4 5 8. It's hard for me to take this goal seriously. 1 2 3 4 5 9. I think this goal is a good goal to shoot for. 1 2 3 4 5 10. This goal will be quite difficult to attain. 1 2 3 4 5 91 GOAL COMMIMENT (POST) INSTRUCTIONS Please rate how strongly you agree or disagree with the following statements regarding the performance goal that was assigned to you. Circle the appropriate number after each item based on the scale below. Please respond to all statements. 1 2 3 4 5 Strongly Disagree Uncertain Agree Strongly Disagree Agree 1. I was strongly committed to pursuing this performance goal 1 2 3 4 5 2. I put forth a great deal of effort in order to achieve this goal 1 2 3 4 5 3. Quite frankly, I didn't care if I achieved this goal or not. 1 2 3 4 5 4. There was not much to be gained by trying to achieve this goal. 1 2 3 4 5 5. I revised the goal during the trial. 1 2 3 4 5 6. I abandoned this goal. 1 2 3 4 5 7. It was unrealistic for me to expect to reach this goal. 1 2 3 4 5 8. It was hard for me to take this goal seriously. 1 2 3 4 5 9. I think this goal was a good goal to shoot for. 1 2 3 4 5 '10. This goal was quite difficult to attain. 1 2 3 4 5 92 FOOTNOTES It should be noted that the relationship between goal difficulty and performance has been shown to be somewhat robust (Mento, Steel and Karren, 1987; Tubbs, 1986). Recently, some authors have questioned the generalizability of this relationship to more complex tasks (Earley, in press; Huber, in press; Wood, in press). The present study will utilize a simple card sorting task, while recognizing that the model proposed may not be generalizable to more complex tasks. 93 NOTES Saari, L. and Latham, G. P. (1980). Hypotheses on reinforcing properties of incentives contingent upon performance, Unpublished manuscript, University of Washington 94 REFERENCES Atkinson, J. W. and Birch, D. (1970) The Dynamics of Action. New York: John Wiley and Sons. Berkowitz, L. and Donnerstein, E. (1982). External validity is more than skin deep: Some answers to criticisms of laboratory experiments. Americal Psychologist, 3;, 245-257. Campbell, D. J. (1982) "Determinants of choice of goal difficulty level: A review of situational and personality influences." Journal of Occupational Psychology,§§, 79-95. Campbell, D. J. (1984) "The effects of goal contingent payment on the performance of a complex task," Personnel Psychology, 31, 23-40. Chung, K. H,. and Vickery, W. D. (1976). Relative effectiveness and joint effects of three selected reinforcements in a repetitive task situation. Organizational Behavior and Human Performance, lg, 114-142. Cohen, J. and Cohen, P. (1983). Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences, Hilldale, N.J.: Erlbaum Associates. Dipboye, R. and Flanagan, M. (1979). Research settings in industrial and organizational psychology: Are findings in the field more generalizable then in the laboratory? American Psychologist, 34, 141-150. 95 Dossett, D. L., Latham, G. P., and Mitchell, T. R. (1979). The effects of assigned versus participatively set goals, RR, and individual differences when goal difficulty is held constant. Journal of Applied Psychology, £5. 291-298. Earley, P. C. (1985). Influence of information, choice and task complexity upon goal acceptance, performance, and personal goals. Journal of Applied Psyghology, 70, 481-491. Earley, P. C. (1985b). The influence of goal settinf methods on performance, goal acceptance, self-efficacy, expectations, and expectancies across levels of goal difficulty. Paper presented at the APA 93rd Annual Meeting, Los Angeles. Erez, M., Earley, P. C. and Hulin, C. (1985). The impact of participation on goal acceptance and performance: A two-step model. Academy of Management Journal, gg, 50-66. Earley, P. C. and Kanfer, R. (1985). The influence of component participation and role models on goal acceptance, goal satisfaction and performance. Organizational Behavior and Human Decision Processes, g9, 378-390. Erez, M. (1977). Feedback: A necessary condition for the goal setting-performance relationship. Journal of Applied Psychology, g3, 624-627. 96 Erez, M. and Kanfer, R. (1983). The role of goal acceptance in goal setting and task performance. Academy of Management Review, g, 454-463. Erez, M. and Zidon, I. (1984). Effect of goal acceptance on the relationship of goal difficulty to performance. Journal of Applied Psyghology, pg, 69-78. Erez, M. Earley, P. and Hulin, C. (1985). The impact of participation on goal acceptance and performance: A two-step model. Academy of Management Journal, gg, 50-66. Fromkin, H. and Streufert, S. (1976). Laboratory Experimentation. In M Dunette, (ed.) Handbook of Industrial/Organizational Psychology, Garland, H. (1983). Influence of ability, assigned goals, and normative information on personal goals and performance: A challenge to the goal attainability assumption. Journal of Applied Psychology, QB, 20-30. Gordon, M., Slade, L. and Schmitt, N. (1986). The "science of the sophomore" revisited: From conjecture to empiricism. Academy of Management Review. ll, 191-207. Gordon, M., Slade, L. and Schmitt, N. (1987). Student guinea pigs: Porcine predictors and particularistic phenomena. Academy of Management Review, 1;, 160-163. Greenberg, J. (1987). The college sophomore as guinea pig: Setting the record straight. Academy of Management Review, ya, 157-159. 97 Hollenbeck, J. R. and Brief, A. P. (1986). The effects of individual differences and goal origin on goal setting and performance. Organizational Behavior and Human Decision Processes, in press. Hollenbeck J. R. and Klein, H. J. (in press) Goal Commitment and the goal setting process: Problems, prospects and proposals for future research. Journal of Applied Psychology. Hollenbeck, J. R., Williams, C. and Klein, H. (1987). An examination of the antecedents of commitment to difficult goals. Working paper, Michigan State University. Hunter, J. Schmidt, P., and Jackson, G. (1982). Advanced meta-analysis: Quantitative methods for cumulating research findingsiacross studies. Beverly Hills, CA: Sage. Ilgen, D. R., Nebecker, D. M., and Pritchard, R. R. (1981). Expectancy theory measures: An empirical comparison in an experimental simulation. Organizational Behavior and Human Performance, gg, 189-223. Ilgen, D. R. and Wiggins, A. (1985). El paso del tiempo. Un factor olvidado en la sequencia establecimiento do metas-desempno-retroinformacion. Revista Interamerican de Psicologia OcupationalL 4, 21-33. James, L. R. and Brett, J. M. (1984). Mediators, moderators, and tests for mediation. Journal of Applied Psyphology, g3, 307-321. 98 James, L. R., Mulaik, S. A. and Brett, J. M. (1982). Causal analysis: Assumptions, models, and data. Beverly Hills, CA: Sage. Kerlinger, F. (1973). Foundation of Behavioral Research, Holt, Rinehart, Winston, Publishers. Latham, G. P. and Baldes, J. (1975). the practical significance of Locke's theory of goal setting. Journal of Applied Psychology, §_, 122-124. Latham, G. P. and Locke, E. A. (1979). Goal setting: A motivational technique that works. Organizaional Dynamics, g, 68-80. Latham, G. P., Mitchell, T. R. and Dossett, D. L. (1978). The importance of participative goal setting and anticipated rewards on goal difficulty and job performance. Journal of Applied Psychology, pg, 163-171. Latham, G. P. and Wexley, K. N. (1981). Increasing Productivity Through Performance Appraisal, Reading, Mass.: Addison-Wesley. Locke, E. A. (1968). Toward a theory of task motivation and incentives. Organizational Behavior and Human Performance, 3, 157—189. Locke, E. A. (1982). Relation of goal level to performance with a short work period and multiple goal levels. Journal of Applied Psychology, Q1, 512-514. 99 Locke, E. A. and Latham, G. P. (1984). Goal setting: A Motivational Tool that Works, Englewood Cliffs, N. J.; Prentice-Hall, Inc. Locke, E. A., Bryan, J. P., and Kendall, L. M. (1968). goals and intentions as mediators of the effectss of monetary incentives on behavior. Journal of Applied Psychloqy, £2. 104-121. Locke, E. A., Cartledge, N. and Knerr, C. S. (1970). Studies on the relationship between satisfaction, goal setting, and performance. Organigational Behavior and Human Performance, Q, 135-158. Locke, E. A., Feren, D. B., McCaleb, V. M., Shaw, K. N. and Denny, A. T. (1980). The relative effectiveness of four methods of motivating employee performance. In K. Duncan, M. Gruneberg, and D. Wallis, (Eds.), Changes in working life, New York: Wiley. 363-388. Locke, E. A., Frederick, E., Buckner, E. and Bobko, P. (1984). Effect of previously assigned goals on self-set goals and performance. Journal of Applied Psychology, g2, 694-699. Locke, E. A., Frederick, E., Lee, C. and Bobko, P. (1984). Effects of self-efficacy, goals, and task strategies on task performance. Journal of Applied Psychology, 69, 241-251. Locke, E. A., Latham, G. P. and Erez, M. (1987) The determinants of Goal Commitment. Academy of Management Review, in press. 100 Locke, E. A., Mento, A. J. and Katcher, B. C. (1978). The interaction of ability and motivation in performance: An explanation of the meaning of moderators. Personnel Psychology, 33, 769-780. Locke, E. A., Shaw, K. N., Saari, L. M. and Latham, G. P. (1981). Goal setting and task performance: 1968-1980. Psychological Bulletin, 33, 125-152. London, M. and Oldham, G. R. (1976). Effects of varying goal types and incentive systems on performance and satisfaction. Academy of Management Journal, 33, 537-546. Lopes, L. L. (1976). Individual strategies in goal setting. Organizational Behavior and Human Performance, 33, 268-277. Masters, J. C., Furman, W. and Barden, R. C. (1977). Effects of achievement standards, tangible rewards, and self-dispensed achievement evaluations on children's task mastery. Child Development, 33, 217-224. Mento, A. J., Steel, R. P. and Karren, R. J. (1987) A meta-analytic study of the effects of goal setting on task performance: 1966-1984. Organizational Behavior and Human Decision Processes, 33, 52-83. Mento, A. J., Cartledge, N. D., and Locke, E. A. (1980). Maryland vs. Michigan, vs. Minnesota: Another look at the relationship of expectancy and goal difficulty to. 101 task performance. Organizational Behavior and Human Performance, 33, 419-440. Milkovich, G. T. and Newman, J. M. (1984). Compensation. Plano, Texas: Business Publications, Inc. Mook, D. (1983). In defense of external invalidity. American Psychologist, Mowen, J., Middlemist, R., and Luther, D. (1981). Joint effects of assigned goal level and incentive structure on task performance: A laboratory study. Journal of Applied Psycholqu, 33, 598-603. Naylor, J. C. and Ilgen, D. R. (1984). Goal setting: A theoretical analysis of a motivational technology. Research in Organizational Behavior, 3, 95-140. Naylor, J. C., Pritchard, R. D. and Ilgen, D. R. (1980). A Theory of Behavior in Organizations, Academic Press. Patten, T. H. (1982). A manager's guide to Performance Appraisal. New York, NY: The Free Press. Pearson, E. F. and Hartley, H. O. (1958). Biometrika Tables for Statisticians, New York: Cambridge Press. Pritchard, R. D. and Curts, M. I. (1973). The influence of goal setting and financial incentives on task performance. Organizational Behavior and Human Performance, 39, 175-183. Rogosa, D. Brandt, D. and Zimowski, M. (1982). A growth curve approach to the measurement of change. Psycholog3cal Bulletin, 33, 726-748. 102 Rosswork, S. G. (1977). Goal setting: The effects on an academic task with varying magnitudes of incentive. Journal of Educational Psychology, 33, 710-715. Schwab, D. P. (1980). Construct Validity in Organizational Behavior. Research in Organizational Behavior, 3, 3-43. Strauss, G. (1972). Management by Objectives: A critical view. Training and Development Journal, 10-15. Stedry, A. C. and Kay, E. (1966). The effects of goal difficulty on performance: A field experiment. Behavior Science, 33, 459-470. Terborg, J. R. (1976). The motivational components of goal setting. Journal of Applied Psychology, 33, 613-621. Terborg, J. R. and Miller, H. E. (1978). Motivation, behavior and peformance: A closer examination of goal setting and monetary incentives. Tolchinsky, P. D. and King, D. C. (1980) "Do goals mediate the effects of incentives on performance?" Academby of Management Review, 3,455-467. Tubbs, M. E. (1986). Goal setting: A meta-analytic examination of the empircal evidence. Journal of Applied Psychology, 13, 474-483. Winer, B. J. (1971). Statistical Principles in Experimental Design (2nd. Ed). New York: McGraw-Hill. "71111171114111