‘COMMON METHOD BIASES IN BEHAVIORAL RESEARCH 881
‘There are several important conclusions that can be drawn from
Table I. For example, the entry in the fist column ofthe First row
indicates that even though two atitude constructs are perfectly
correlated, the observed correlation between their measures is only
2 because of measurement errr. Similarly, the entry in the last
Column of the firs row indicates that even though two atitude
constructs are completely uncorrelated, the observed correlation
between their measures is 23 because of random and systematic
‘measurement errr. Both of these numbers ate troubling but for
diferent reasons. The enties in the entre frst column are trou-
bling because they show that even though two tats are perfectly
correlated, typical levels of measurement error cut the observed
correlation berween their measures in half and the variance ex:
plained by 70%. The last column of ences is troubling because it
shows that even when two constrct are completely uncorrelated,
‘measurement error causes the observed correlation between their
‘measures to be greater than zero. Indeed, some of these numbers
are not very different from the effect sizes reported in the beh
ioral literature. In view ofthis, i is disturbing that most studies
‘ignore measurement ertor entirely and tht even many ofthe ones
that do try to take random measurement error into account ignore
systematic measurement err. Thus, measurement error 2m it
flate or deflate the observed correlation between the measures,
‘pending on the correlation between the methods. Indeed, 3s
noted by Cote and Buckley (1988), method effects inflate the
‘observed relationship when the coreltion between the methods is
higher than the observed correlation berween the measures with
method effects removed and deflate the relationship when the
correlation between the methods is lower than the observed cor-
relation between the measures with method effects removed.
Potential Sources of Common Method Biases
Because commen method biases can have potentially serious
effects on research findings, itis important to understand thie
sources and when they are especially likely to be « problem.
“Therefore, in the next sections of the article, we identify several of
the mos likely causes of method bias and the research setings in
which they are likely to pose particular problems. As shown in
Table 2, some sources of common method biases result fom the
fact thatthe preictor and criterion variables are obtained from the
same source or ater, whereas others are produced by the measure-
iment items themselves, the context of the items within the mes-
surement instrument, and/or the context in Which the measures are
obtained.
Method Effects Produced by a Common Source or Rater
Some methods effects result from the fect thatthe espondent|
providing the measure ofthe predictor and criterion variable isthe
Same person. This type of sefteport bias may be said to result
from any artifactual covariance between the predictor and criterion
variable produced by the fact that the respondent providing the
‘measure of these variables isthe same.
Consistency motif. There isa substantial amount of theory (ct
Heider, 1958; Osgood & Tannenbaum, 1953) and research (ef.
“MeGuite, 1966) suggesting that people try to maintain consistency
between their cognitions and atitudes. Thus, it should not be
surprising that people responding to questions posed by research-
15 would have a desire (0 appear consistent and rational inthe
responses and might search for similarities in the questions asked
of them—thereby producing relationships that would not other
wise exist at the same level in real-life settings. This tendency of
respondents to try to maintain consistency in their responses to
similar questions oF to organize information in consistent ways is
called the conssteney motif Qobns, 1994; Podsakoft & Organ,
1986; Schmit, 1994) or the conslotency effect Salancik & Prefer,
1977) and is likely to be particularly problematic in those situa.
tions in which respondents are asked to provide retrospective
accounts oftheir atituds, perceptions, andor behaviors
Implicit theories and illusory correlations. Related to the no-
‘ion of the consistency motif as a potential source of common
method variance are illusory correlations (cf. Berman & Kenny,
1976; Chapman & Chapman, 1967, 1968; Smither, Collins, &
Buda, 1989), and implicit theories (ef. Lord, Binning, Rush, &
‘Thomas, 1978; Phillips & Lord, 1986; Staw, 1975). Berman and
Kenny (1976) have indicated tha illusory correlations result fom
the fact that “raters often appear to possess assumptions concem-
ing the co-occurrence of rated items, and these assumptions may
introduce systematic distortions when correlations are derived
fom the ratings” (p. 268); Smither et al. (1989) have noted that
these “illusory correlations may serve asthe basis of job schema oF
implicit theories held by rates and theteby affect aitenton to and
‘encoding of ratee behaviors as well a later recall” (p, 599). This
suggests that correlations derived from ratees' responses are com-
posed of not only true relationships but also artifactual covariation
based on rates” implicie theories
Indeed, there is a substantial amount of evidence that implicit
theories do have an effect on respondents’ ratings in a variety of
different domains, including ratings of leader behavior (eg, Eden
& Leviatin, 1975; Lord et al, 1978; Phillips & Lotd, 1986),
tributions ofthe causes of group performance (ef. Guzzo, Wag-
ner, Maguire, Her, & Hawley, 1986; Staw, 1975), and perceptions
about the relationship between employee satisfaction and perfor
‘mance (Smither etal, 1989). Taken together, these findings ind
cate thatthe relationships researchers observe between predictor
and criterion variables on a questionnaire may not oly reflet the
actual covariation tat exists between these events but may alsa be
‘the result of the implicit theories that respondents have regarding
the relationship between these evens
Social desirability: According to Crowne and Marlowe
(1964), social desirability “refers to the need for social approval
1nd acceptance and the belief that it can be attsined by means of
culturally acceptable and appropriate behaviors” (p. 109). It is
generally viewed as the tendeney on the part of individuals to
present themselves in 2 favorable light, regardless of their tue
feelings about an issue or topic. This tendency is problematic, not
only because ofits potential to bias the answers of respondents
(je, to change the mean levels of the response) but also because
it may mask the true relationships between two or more variables
(Ganstr, Hennessey & Luthans, 1983). Ganster tal (1983) have
noted that social desirability can produce spurious relationships,
serve as a suppressor variable that hides the tre relationship
berween variables, o serve as a moderator variable that influences
the nature ofthe relationships between the variables
Leniency bases. Guilford (1954, p. 278) has defined leniency
biases as the tendency for raters “to rate those whom they know
well, or whom they are ego involved, higher than they should
Research on this form of bias (Schriesheim, Kinicki, &882 PODSAKOFF, McKENZIE, LEE, AND PODSAKOFF
Table 2
Summary of Potemial Sources of Common Method Biases
Potential ease Defiion
Refer to any artifactual covariance between th predictor and criterion variable produced by the Gt that the
Tespondent proving the measre ofthese variables ithe same.
Refers othe propensity fr respondent to ty fo maintain consistency in thee responss to questions
Refer to respondents eles about he covariation among pariclrtis, behaviors, andlor oucones
Common rater effets
Consistency motif
Inlet theories (and ilsory
‘stltons)
Social desiality Refers to the tendeney of sme people to respond to items more a8 result ofthe scil acceptability than
thei tne fins.
Refer othe propensity for respondents to abu socially desirable wats atiudes, sade behavior to
omcons thy Know and ie tha to someone they dle
Refer othe propensity for espondems 1 ate (r disagree) with quetonire items independent of thie
Refers wo the propensity of respondents to view themssives andthe world arond them n generally negative
terms (oegaliveafectiviy) orth propensity of espondent o view themslves andthe world sound
them in generally postive terms (postive aleetivip)
‘Refers tothe impact of elavelyfeent mood inducing even nfvence the manner in which respondents
‘iow themselves and the world sound them.
Refer to any aificual covariance thal is caused bythe influence o interpretation that a eespondent might
‘sserbe to an item solely Because of specie propre or characteristics the item possesses
Refers tthe Tact that items maybe writen in Such away aso rect more socially desirable aides,
‘behaviors, oF perceptions
Refer othe fat tat me may convey hidden cues a ohow to respond othe
Refers to the fact hat items that sre ambos alow respondents fo respond to thm systematically using
thei own hers or respond othe randomly.
Refer to atfsctial covariation produced by the ite ofthe same scale format eg. Likert sales, semantic
“ferential seals, Taes” sales) on questionaire.
Refer to the repeated us of the same anchor points (eg, exremely, always, never) ona questions
[Refes fo the fact that the use of postvely (nepalivel) worded tems may produce aisetal relationships
‘on The questionaire
Refer to any influence or iferpretation tat a respondent might serie oan itm solely because ofits
‘elation fo the other tems making up an instument (Wainer & Kisly, 1987)
Refer to the fact tat the positioning ofthe peste (or eriterin) variable onthe quesonnaze can make
tat variable mre salient othe espondent and imply a causal atonship with athe vara
Refers othe fct hat neatal items ended in the content of either positively or negatively worded items
‘vl take on the evaluative properties of has ers.
Refers to when the Bat question for st of gestions) encountered on the questionnaire indices mood for
responding othe rmsiner ofthe questionaite
‘Refers to theft that if scales have fewer tems, responses to previous items are mor likely o be
‘acessible in sorter eiemory and wo be recall’ when responding fo oer iems,
Refers to theft that stems fom difret constr alae grouped together my decrease inraconstrst
‘comelaons and increase lneconsructconeltions.
Lenieny bases
Acquescence biases (yen-aying
‘ad ay-saying)
Mood sats (postive oe negative
sMfeatviy, positive or nepalive
‘emotionli)
‘eansen eno state
em charcteritie est
ter sci desirabily
term demand characteris
hem ambiguity
Common scale formats
Commen sele anchors
Postve and negative item
wording
tem context effets
tem priming effects
te embedednest
Contextinduced mood
Seale length
Imecriing (or grouping) of
tome or contri the
uestionaite
Measurement context effects
Predictor an eiteron variables
messed tthe same point in
Predictor and enteronvaribles
‘easies inthe sme loation
Predictor and enteon variables
measured using the sme
red
Refer to any arifacual covariation produced fom th context in which the measures are obiied
Refers to the fact that measures of different constructs messed at the sane pot in ne may produce
‘onfacal covariance independent of the content ofthe cosets themsles.
Refers to the fct hat measures of ifferent constructs measured in the same location may pode
‘ifotual covariance independent ofthe content ofthe constucts hemslves
Refers to the fat that menor of efferent constrict measured withthe same medium may produce
‘ntfatual covariance independent ofthe content ofthe const themselves
Schriesheim, 1979) has shown that it produces spurious corelations
bonween leader-consdertion behavior and employee stisfction and
perceptions of group productivity, drive, and cohesiveness but not
bberween leader iiating stuctue behavior and these same eiterion
‘variables. This suggests that the consideration scale is not socaly
‘neutral and that leniency biases tend t0 influence the relationships
obtained between this sale and employee attitudes and perceptions
(One might also expect leniency biases to produce spurious corel
tions in other staics that examine the relationship between respon-
dents’ ratings of liked (r disliked) others andthe respondents ratings
Of the performance, atiudes, and perceptions of others.
Acquiescence (yea-saying or nay-saying). Winkler, Kanowse,
and Ware (1982, p_ $55) have defined acquiescence response set
1s the “tendency to agree with atiude statements regardless of
content” and have noted that this response set is problematic
“pecause it heightens the correlations among items that are worded
similarly, even when they are not conceptually related.” Although
Winkler ef al, (1982) focused specific attention on the effects of |
acquiescence on scale development processes, it is easy to see how
this form of bias might also cause spurious relationships between
‘so or more constructs. Thus, acquiescence may also be a potential
cause of artifactual variance inthe relationships between two orCOMMON METHOD BIASES IN BEHAVIORAL RESEARCH 883
more variables, other than the true varianee between these
variables
Positive and negative affectviy. Waston and Clark (1984)
defined negative afectivity 3s a mood-