CET suggests that rewards may be harmful to employee performance due to
detrimental eVects on intrinsic task motivation. Intrinsic motivation is experienced
when a person performs an activity that is of interest to them even when no apparent
reward is received. Extrinsic motivation refers to motivation to perform an activity
strictly for the rewards themselves (Daniel and Esser 1980 ). CET suggests that
intrinsic motivation is adversely aVected by rewards when reward recipients per-
ceive the reward as controlling or as a challenge to competence. This suggestion has
spurred a number of studies to test propositions of the theory, with mixed results.
A series of meta-analyses (Cameron and Pierce 1994 , 1997 ; Deci et al. 1999 ;
Eisenberger and Cameron 1996 ; Wiersma 1992 ) suggests that the extent to which
rewards undermine intrinsic motivation is limited and circumscribed. The most
supportive of these meta-analyses is Deci et al. ( 1999 ), but as pointed out by Gerhart
and Rynes ( 2003 ), Deci et al.’s study found no support for detrimental eVects on
attitudes; moreover, theirWndings suggesting negative eVects for the ‘free-time’
measure of intrinsic motivation were much more pronounced in school-aged
children as opposed to an adult (college-age) population. Finally, some studies
(e.g. Fang and Gerhart 1999 ) have reportedpositiverelationships between pay and
intrinsic motivation. Thus, although the CET literature has generated a provocative
and healthy dialogue on the motivational value of incentives (e.g. Gupta and Shaw
1998 ; Kohn 1998 ), research does not overwhelmingly support the assertion that
incentives and rewards undermine individuals’ motivation or task interest. More-
over, even if Deci and colleagues are more right than wrong—even if incentives and
rewards have the deleterious eVects they suggest—the cumulative evidence
(reviewed below) belies the claim that ‘incentive plans cannot work’ (Kohn 1993 ).
An early meta-analysis by Guzzo et al. ( 1985 ) examined the productivity eVects
of a number of organizational initiatives, including ‘Wnancial compensation’, in
studies published during the 1971 – 81 time period. WhileWnancial initiatives had no
eVect on employee withdrawal, the eVect on productivity was both statistically and
practically signiWcant. In fact, the eVect size associated withWnancial compensation
initiatives was more than twice that of any of the other classes of initiatives (e.g.
training, appraisal and feedback, goal-setting, etc.).
Jenkins et al. ( 1998 ) conducted a more recent meta-analysis of studies published
from 1960 to 1996. Jenkins et al. ( 1998 ) only analyzed individual-level research that
used experimental or quasi-experimental designs. Of the forty-seven separate
studies, forty-one focused on performance quantity and six on performance
quality. While the cumulative results across the six studies focusing on perform-
ance quality found no relationship between incentives and quality (the eVect was
positive, but small and insigniWcant), the eVect size for performance quantity was
substantial (. 34 ). While incentive eVects were evident in all settings, the eVect sizes
in bothWeld (. 48 ) and simulations (. 56 ) were much higher than those reported in
laboratory settings (. 24 ). Since the majority of studies are conducted in lab settings,
this may indicate that many studies underestimate the performance eVects of
remuneration: pay effects at work 347