Articol mediere

7/30/2019 Articol mediere

http://slidepdf.com/reader/full/articol-mediere 1/10

Jou rnal of Pe ~n al i ty and Social Psychology Copyright 1986 by the American Psychological Association, Inc .1986, Vol. 51, No . 6, 1173-1182 0022-3514 /86/$00.75

T he M o de r a t o r - M e di a t o r Va r i a b l e D i s t i nc t i o n i n So c i a l P sy c ho l o g i c a l

Research: C on cep tual , Strateg ic, and Stat is t ica l C ons idera t ions

R e u b e n M . B a r o n a n d D a v i d A . K e n n yU n i v e r s i ty o f C o n n e c t i c u t

In th i s a r t i c l e , we a t t em pt to d i s t ingu i sh be tween the p rope r t i e s o f mo dera to r and med ia to r va r iab les

a t a numb er o f leve ls . F i rs t , we seek to make theor i s t s and re sea rche rs aware o f the im por tance o f

no t us ing the t e rms moderator and mediator interchangeably by careful ly e laborat ing, both concep-tua l ly and s t ra teg ical ly , the m any ways in which mod era to r s and med ia to r s d i f fe r . We then go bey ond

th i s l a rge ly pedagogical func t ion and de l inea te the concep tua l an d s t ra teg ic im pl ica t ions o f m aking

use o f such d i s t inc t ions wi th rega rd to a w ide range o f phenomena , inc lud ing con t ro l and s t r es s ,

a t t i tudes , and pe r sona l i ty t r a i t s. We a l so p rovide a speci fi c compend ium of ana ly t i c p rocedures ap -p ropr ia te fo r making the mos t e f fec tive use o f the m odera to r and med ia to r d i s tinc t ion , bo th sepa -

ra te ly and in t e rms o f a b roa der causa l sys tem th a t inc ludes bo th m odera to r s and med ia to r s.

T h e p u r p o s e o f t h i s a n a ly s i s is t o d i s t i n g u i sh b e t w e e n t h e

p r o p e r t i e s o f m o d e r a t o r a n d m e d i a t o r v a r i a b le s i n s u c h a w a y

a s t o c l a r i f y t h e d i f f e r e n t w a y s i n w h i c h c o n c e p t u a l v a r i a b l e s

m a y a c c o u n t f o r d i ff e r e n c e s i n p e o p l e s ' b e h a v i o r . S p e c i fi c a ll y ,

w e d i f f e r e n t ia t e b e tw e e n t w o o f t e n - c o n f u s e d f u n c t i o n s o f t h i r d

v a r i a b le s : (a ) t h e m o d e r a t o r f u n c t i o n o f t h i r d v a r i a b le s , w h i c h

p a r t i t i o n s a f o c a l i n d e p e n d e n t v a r i a b l e i n t o s u b g r o u p s t h a t e s -

t a b l i s h i t s d o m a i n s o f m a x i m a l e f f e c ti v e n e s s i n r e g a r d t o a g i v e n

d e p e n d e n t v a r i a b le , a n d ( b ) t h e m e d i a t o r f u n c t i o n o f a t h i r d

v a r i a b le , w h i c h r e p r e s e n t s t h e g e n e r a ti v e m e c h a n i s m t h r o u g h

w h i c h t h e f o c a l i n d e p e n d e n t v a r i a b l e is a b l e t o i n f l u e n c e t h e

d e p e n d e n t v a r i a b l e o f i n te r e s t .

A l t h o u g h t h e s e tw o f u n c t i o n s o f t h i r d v a r i a b le s h a ve a r e l a -t i v e l y l o n g t r a d i t i o n i n t h e s o c i a l s c i e n c e s , i t i s n o t a t a l l u n c o m -

m o n f o r so c i a l p s y c h o l o g ic a l r e s e a rc h e r s t o u , t h e t e r m s mod-

erator and mediator i n t e r c h a n g e a b l y . F o r e x a m p l e , H a r k i n s ,

L a ta n 6 , a n d W i l li a m s 0 9 8 0 ) f ir st s u m m a r i z e d t h e i m p a c t o f

i d e n t i f i a b i l i t y o n s o c i a l l o a f i n g b y o b s e r v i n g t h a t i t " m o d e r a t e s

s o c i al lo a f in g " ( p . 3 0 3 ) a n d t h e n w i t h i n t h e s a m e p a r a g r a p h

p r o p o s e d " t h a t i d e n t i f ia b i l it y s a n i m p o r t a n t m e d i a t o r o f so c i a l

l o a f i n g : ' S i m i la r l y, F i n d l e y a n d C o o p e r ( 1 9 8 3 ), i n t e n d i n g a

m o d e r a t o r i n t e r p r e t a t i o n , l a b e l e d g e nd e r , a g e , r a c e , a n d s o c i o -

e c o n o m i c l e v el a s m e d i a t o r s o f t h e r e l a t i o n b e t w e e n l o c u s o f

c o n t r o l a n d a c a d e m i c a c h i e v e m e n t . T h u s , o n e l a r g el y p e d a g o g i -

Th is r e sea rch w as suppor ted in pa r t by Na t iona l Sc ience Founda t ion

Gran t BNS-8210137 and Na t iona l Ins t i tu te o f Menta l Hea l th Gran t

R01 MH -40295-01 to the second author. Su ppo rt was also given to himdur ing h i s sabba t i ca l yea r (1982-83) by the M acA r thur Founda t ion a t

the C enter for Advanced Studies in the Behavioral Sciences, Stanford,

California.Than ks a re due to Ju d i th Harack iewicz, Char le s Judd , S tephen Wes t ,

and Har r i s Cooper , who p rovided commen ts on an ea r l i e r ve r sion o fthis ar t ic le . Stephe n P. Needel was instrum ental in the be ginnin g s tages

of th i s work .Cor respondence conce rn ing th i s a r t i c l e shou ld be add ressed to Reu-

ben M. Baron, Dep ar tmen t o f Psycho logy U-20 , Unive rs i ty o f Connec t -icut , Storrs , Co nnec ticut 06268.

c a l f u n c t i o n o f t h i s a r t i c l e is to c l a r i f y f o r e x p e r i m e n t a l r e -

s e a r c he r s t h e i m p o r t a n c e o f r e s p e c t i n g t h es e d i s t i n c ti o n s .

T h i s i s n o t , h o w e v e r , th e c e n t r a l t h r u s t o f o u r a n a l y s i s . R a t h e r ,

o u r m a j o r e m p h a s i s i s o n c o n t r a s ti n g th e m o d e r a t o r - m e d i a t o r

f u n c t i o n s i n w a y s t h a t d e l i n e a t e t h e i m p l i c a t i o n s o f th i s d i s t i n c -

t i o n f o r t h e o r y a n d r e s e a rc h . W e f o c u s p a r t ic u l a r l y o n t h e

d i f f e r e n t i a l i m p l i c a t i o n s fo r c h o i c e o f e x p e r i m e n t a l d e s i g n , r e -

s e a r c h o p e r a t i o n s , a n d p l a n o f s t a t i s ti c a l a n a l y s i s .

W e a l s o c l a i m t h a t t h e r e a r e c o n c e p t u a l i m p l i c a t i o n s o f th e

f a i lu r e to a p p r e c i a t e th e m o d e r a t o r - m e d i a t o r d is t i n c ti o n .

A m o n g t h e i s s u e s w e w i ll d is c u s s in t h i s r e g a r d a r e m i s s e d o p -

p o r t u n i t i e s t o p r o b e m o r e d e e p l y i n to t h e n a t u r e o f c a u sa l

m e c h a n i s m s a n d i n t e g r a te s e e m i n g l y i r r e c o n c i l a b l e t h e o r e t ic a lp o s i t io n s . F o r e x a m p l e , i t is p o s s ib l e t h a t i n s o m e p r o b l e m a r e a s

d i s a g r e e m e n t s a b o u t m e d i a t o r s c a n b e r e s o l v e d b y t r e a t i n g c e r-

t a i n v a r i a b l e s a s m o d e r a t o r s .

T h e m o d e r a t o r a n d m e d i a t o r f u n c t i o n s w il l b e d i s c u s s ed a t

t h r e e l e v e ls : c o n c e p t u a l , s t r a t e g ic , a n d s t a t i st i c a l . T o a v o i d a n y

m i s u n d e r s t a n d i n g o f th e m o d e r a t o r - m e d i a t o r d i st i n c t i o n b y e r -

r o n e o u s l y e q u a t i n g i t w i t h t h e d i ff e re n c e b e tw e e n e x p e r i m e n t a l

m a n i p u l a t i o n s a n d m e a s u r e d v a r i a b l e s , b e t w e e n s i t u a t i o n a l a n d

p e r s o n v a r i a b l e s , o r b e t w e e n m a n i p u l a t i o n s a n d v e r b a l s e l f - r e -

p o r t s , w e w i l l d e s c r i b e b o t h a c t u a l a n d h y p o t h e t i c a l e x a m p l e s

i n v o l v i n g a w i d e r a n g e o f v a r i a b l e s a n d o p e r a t i o n s . T h a t i s ,

m o d e r a t o r s m a y i n v o l v e e i t h e r m a n i p u l a t i o n s o r a s s e s s m e n t s

a n d e i t h e r s i t u a t i o n a l o r p e r s o n v a r i a b l e s . M o r e o v e r , m e d i a t o r sa r e i n n o w a y r e s t r i c t e d t o v e r b a l r e p o r t s o r , f o r t h a t m a t t e r , t o

i n d i v i d u a l - l e v e l v a r i a b l e s .

F i n a l l y , f o r e x p o s i t o r y r e a s o n s , o u r a n a l y s i s w i l l i n i t i a l l y

s t re s s t h e n e e d t o m a k e c l e a r w h e t h e r o n e i s t e st i n g a m o d e r a t o r

o r a m e d i a t o r t y p e o f m o d e l . I n t h e s e c o n d h a l f o f t h e a r ti c l e,

w e p r o v i d e a d e s i g n t h a t a l l o w s o n e t o t e s t w i t h i n t h e s t r u c t u r e

o f th e s a m e s t u d y w h e t h e r a m e d i a t o r o r m o d e r a t o r i n t e r p r e t a -

t i o n i s m o r e a p p r o p r i a t e .

A l t h o u g h t h e s e i s s u e s a r e o b v i o u s l y i m p o r t a n t f o r a l a r g e

n u m b e r o f a r e a s w i t h i n p s y c h o l o g y , w e h a v e t a r g e t e d t h i s a r t i c l e

f o r a s o c i a l p s y c h o l o g i c a l a u d i e n c e b e c a u s e t h e r e l e v a n c e o f t h i s

d i s t i n c t i o n i s h i g h e s t i n s o c i a l p s y c h o lo g y , w h i c h u s e s e x p e r i -

1173



11 74 REUBEN M. BARON AND DAVID A. KENNY

mental operations and at the same time retains an interest in

organismic variables ranging from individual difference mea-

sures to cognitive constructs such as perceived control.

T h e N a t u r e o f M o d e r a t o r s

In general terms, a moderator is a qualitative (e.g., sex, race,

class) or qua ntitative (e.g., level of reward) variable tha t affects

the di rect ion and /or s t rength o f the re la t ion between an inde-

pendent or predictor variable and a dependent or cr i ter ion vari-

able.

Specifically within a correlational analysis framework, a

mod erato r is a third variable tha t affects the zero-ord er correla-

tion between two other variables. For example, Stem, M cCants,

and Pet t ine (1982) found that the pos i t ivi ty of the re la t ion be-

tween ch anging life events and severity of i l lness was considera-

bly stronger for uncontrollab le events (e.g., death of a spouse)

than for controllable events (e.g., divorce). A mode rator effect

within a corre la t ional framework may also be said to occur

where the direction of the correlation changes. Such an effectwould have occur red in the Stern et al. s tudy if controllable life

changes had red uced the likelihood o f illness, the reby chang ing

the direction of the relation b etween life-event change and ill-

ness from positive to negative.

In the mo re fam iliar analysis of variance (ANOVA) erms, a

basic moderator effect can be represented as an interaction be-

tween a focal independent variable and a factor that specifies

the appropria te condi t ions for it s operat ion. In the disson ance-

forced compliance area, for example , i t became apparent that

the ability o f investigators to establish the effects of insufficient

justification required the specification of such mod erators as

commitment, personal responsibility, and free choice (cf.

Brehm & Cohen, 1962).An example of a mod erator- type effect in this context i s the

demonstra t ion of a crossover interact ion of the form that the

insufficient justification effect holds under public commitment

(e.g., atti tude change is inversely related to incentive), whereas

attitude change is directly related to level of incentive when the

coun teratti tudin al action occu rs in private (cf. Collins & Hoyt,

1972). A mo derator-in teraction effect also would be said to oc-

cur if a relation is substantially reduc ed instead of being re-

versed, for example, if we find no difference under the private

condition.

Toward Establishing an Ana lytic Framew ork

for Testing Moderator Effects

A common framework for capturing both the corre la t ional

and the exp erimental views of a moder ator variable is possible

by us ing a path diagra m as both a descriptive and an analyt ic

proce dure. Glass and Singer's (1972) finding of an interaction

of the factors stressor intensity (noise level) and controllability

(periodic-aperiodic noise) , of the form that an adverse impact

on task performan ce occu rred only when the onset of the noise

was aperiodic or unsignaled, will serve as our substantive exam-

ple. Using such an appro ach, th e essential properties of a mod -

erator variable are summarized in Figure 1.

The mod el diagramm ed in Figure 1 has three causal paths

that feed into the outcom e variable o f task performance: the

Figure 1.Moderator model.

impact of the noise intens i ty as a predictor (Path a) , the impact

of controllability as a mo dera tor (Path b), and the inter action

or produ ct o f these two (Path c) . The mo derator hypothes is issuppo rted if the interaction (Path c) is significant. There may

also be significant mai n effects for the pred ictor and the m oder-

ator (Paths a and b), but these are no t directly relevant conc ep-

tually to testing the mode rator hypo thesis.

In addition to these basic considerations, i t is desirable that

the moderator variable be uncorrela ted wi th both the predictor

and the criterion (the dependent variable) to provide a clearly

interpretable interact ion term. Another pro perty of the moder-

ator variable apparen t fro m Figure 1 is that, unlike the me dia-

tor-predictor relation (where the predictor is causally anteced-

ent to the mediator), m oderators and predictors are a t the same

level in regard to their role as causal variables antecedent or

exogenou s to certain criterion effects. Th at is , mo dera tor vari-ables always function as independent variables, whereas medi-

ating events shift roles from effects to causes, depend ing o n the

focus oftb e analysis.

Choosing an Appropriate Ana lytic Procedure:Testing Moderation

In this section we consider in detail the specific analysis pro-

cedures for appropriate ly measuring and tes t ing mod erat ional

hypotheses . Within this framework, moderat ion impl ies that

the causal relation between two variables changes as a function

of he mod erato r variable. The statistical analysis mus t measure

and test the differential effect of the indepen dent variable on the

dependent variable as a funct ion o f the m oderator. The way to

measure and test the differential effects depends in par t o n the

level of measuremen t o f the independent variable and the mo d-

erator variable. We will consider four eases: In Case 1, both

moderator and independent variables are categorical variables;

in Case 2, the moderator is a categorical variable and the inde-

pendent variable a continuous variable; in Case 3, the modera-

1At a conceptual level, a m oderator m ay be m ore impressive if we gofrom a strong to a weak relation or to no relation at all as opposed tofinding a crossover interaction. T hat is, although crossover interactions

are stronger statistically, as they are no t accompanied by residual m aineffects, conceptually no effect shifts may be m ore impressive.



T H E M ODE RA TOR-M EDIA T OR DIS T IN C T ION 1 1 7 5

t o r i s a c o n t i n u o u s v a r i a b l e a n d t h e i n d e p e n d e n t v a r i a b l e i s a

c a t e g o r i c a l v a r ia b l e ; a n d i n C a s e 4 , b o t h v a r i a b l e s a r e c o n t i n u -

o u s v a r i a b le s . T o e a s e o u r d i s c u s s i o n , w e w i l l a s s u m e t h a t a l l t h e

c a t e g o r i c a l v a r ia b l e s a r e d i c h o t o m i e s .

Case 1

T h i s i s t h e s i m p l e s t c a s e . F o r t h i s c a s e , a d i c h o t o m o u s i n d e -p e n d e n t v a r i a b l e ' s e f f e c t o n t h e d e p e n d e n t v a r i a b l e v a r i e s a s a

f u n c t i o n o f a n o t h e r d i c h o t o m y . T h e a n a l y s i s i s a 2 • 2 A N OV A ,

a n d m o d e r a t i o n i s i n d i c a t e d b y a n i n t e r a c t i o n . W e m a y w i s h t o

m e a s u r e t h e s i m p l e e f f e c ts o f t h e i n d e p e n d e n t v a r i a b l e a c r o s s

t h e l e v el s o f t h e m o d e r a t o r ( W i n e r , 1 9 7 1 , p p . 4 3 5 - 4 3 6 ) , b u t

t h e se s h o u l d b e m e a s u r e d o n l y i f t h e m o d e r a t o r a n d t h e i n d e -

p e n d e n t v a r i a b l e i n t e r a c t t o c a u s e t h e d e p e n d e n t v a r i a b le .

Case 2

H e r e t h e m o d e r a t o r i s a d i c h o t o m y a n d t h e i n d e p e n d e n t v a ri -

a b l e i s a c o n t i n u o u s v a r i a b l e . F o r i n s t a n c e , g e n d e r m i g h t m o d e r -

a t e t h e e f fe c t o f i n t e n t i o n s o n b e h a v io r . T h e t y p i c a l w a y t o m e a -

s u r e t h i s t y p e o f m o d e r a t o r e f f ec t is t o c o r r e l a t e i n t e n t i o n s w i t h

b e h a v i o r s e p a r a t e l y f o r e a c h g e n d e r a n d t h e n t e s t t h e d i ff e r en c e .

F o r i n s t a n c e , v i r t u a l l y a l l s tu d i e s o f m o d e r a t o r s o f t h e a t t i t u d e -

b e h a v i o r r e l a t i o n u s e a c o r r e l a t i o n a l t e s t .

T h e c o r r e l a t i o n a l m e t h o d h a s t w o s e r i o u s d e fi c ie n c ie s . F i rs t ,

i t p r e s u m e s t h a t t h e i n d e p e n d e n t v a r i a b l e h a s e q u a l v a r ia n c e a t

e a c h l e v e l o f t h e m o d e r a t o r . F o r i n s t a n c e , t h e v a r i a n c e o f i n t e n -

t i o n m u s t b e t h e s a m e f o r t h e g e n de r s . I f v a r i a n c e s d i f fe r a c r o s s

l e v el s o f t h e m o d e r a t o r , t h e n f o r le v e ls o f th e m o d e r a t o r w i t h

l e ss v a r i a n c e , t h e c o r r e l a t i o n o f t h e i n d e p e n d e n t v a r i a b l e w i t h

t h e d e p e n d e n t v a r i a b l e t e n d s t o b e l e s s t h a n f o r le v e ls o f th e

m o d e r a t o r w i t h m o r e v a r i a n c e . T h e s o u r c e o f t h i s d if f e r e nc e is

r e f e r r e d t o a s a r e s t r ic t i o n i n r a n g e ( M c N e m a r , 1 96 9) . S e c o n d ,

i f t h e a m o u n t o f m e a s u r e m e n t e r r o r i n t h e d e p e n d e n t v a ri a b lev a r i e s a s a f u n c t i o n o f t h e m o d e r a t o r , t h e n t h e c o r r e l a t i o n s b e -

t w e e n t h e i n d e p e n d e n t a n d d e p e n d e n t v a r i a b l e s w il l d i ff e r s p u r i -

ously .

T h e s e p r o b l e m s i l lu s t r a te t h a t c o r r e l a t i o n s a r e i n f lu e n c e d b y

ch an g es i n v a r i an ces . Ho wev er , r eg res s io n co e f f i c i en t s a r e n o t

a f f e c t e d b y d i f f e r en c e s i n t h e v a r i a n c e s o f t h e i n d e p e n d e n t v a r i -

a b l e o r d i f f er e n c e s i n m e a s u r e m e n t e r r o r i n th e d e p e n d e n t v a r i -

a b l e . I t i s a l m o s t a l w a y s p r e f e r a b le t o m e a s u r e t h e e f f e ct o f th e

i n d e p e n d e n t v a r i a b l e o n t h e d e p e n d e n t v a r i a b l e n o t b y c o r r e l a -

t i o n c o e f f i c ie n t s b u t b y u n s t a n d a r d i z e d ( n o t b e t a s ) r e gr e s s i o n

co e f f i c i en t s ( Du n ca n , 1 9 75 ). T es t s o f t h e d i f f e ren ce b e tween r e -

g r e s s i o n c o e f f i c i e n t s a r e g i v e n i n C o h e n a n d C o h e n ( 1 9 8 3 , p .

5 6 ) . T h i s t e s t sh o u ld b e p e r fo rm ed f i r s t , b e fo re t h e two s lo p esa re i n d iv id u a l ly t e s t ed .

I f t h e r e i s d i f fe r e n t ia l m e a s u r e m e n t e r r o r i n t h e i n d e p e n d e n t

v a r i ab l e ac ro ss l ev e l s o f t h e m o d era to r , b i a s r e su l ts . R e l i ab i l it i e s

w o u l d t h e n n e e d t o b e e s t i m a t e d f o r t h e d i ff e r e n t le v e ls o f t h e

m o d e r a t o r , a n d s l o p e s w o u l d h a v e t o b e d i s a t te n u a t e d . T h i s c a n

b e a c c o m p l i s h e d w i t h i n t h e c o m p u t e r p r o g r a m L I S R E L - V I

( J 6 r e s k o g & S 6 r b o m , 1 9 8 4 ) b y u s e o f t h e m u l t i p l e - g r o u p o p -

t i o n . T h e l e ve l s o f th e m o d e r a t o r a r e t r e a t e d a s d i f fe r e n t g r o u p s.

Case 3

I n t h i s c a s e , t h e m o d e r a t o r i s a c o n t i n u o u s v a r i a b l e a n d t h e

i n d e p e n d e n t v a r i a b l e i s a d i c h o t o m y . F o r i n s ta n c e , t h e i n d e p e n -

Figure 2.Three different ways n which the moderator changes the effectof he independent variable on the dependent variable: linear ( top), qua-dratic (middle) , and step (bottom).

d e n t v a r i a b l e m i g h t b e a r a t i o n a l v e r s u s f e a r -a r o u s i n g a t t i tu d e -

c h a n g e m e s s a g e a n d t h e m o d e r a t o r m i g h t b e i n t e l l i g e n c e a s

m e a s u r e d b y a n I Q t e s t . T h e f e a r - a r o u s i n g m e s s a g e m a y b e

m o r e e f f e c ti v e fo r l o w - I Q s u bj e c ts , w h e r e a s t h e r a t i o n a l m e s s a g e

m a y b e m o r e e f f e c ti v e f o r h i g h - I Q s u b j ec t s . T o m e a s u r e m o d e r a -

t o r e f f ec t s i n t h i s c a s e , w e m u s t k n o w a p r i o r i h o w t h e e f f ec t o f

t h e i n d e p e n d e n t v a r i a b l e v a r i e s a s a f u n c t i o n o f th e m o d e r a t o r .

I t i s im p o s s i b l e t o e v a l u a t e t h e g e n e r a l h y p o t h e s i s t h a t t h e e f f ec t

o f th e i n d e p e n d e n t v a r i a b l e c h a n g e s a s a f u n c t i o n o f t h e m o d e r -

a t o r b e c a u s e t h e m o d e r a t o r h a s m a n y l e ve ls .

F i g u r e 2 p r e s e n t s t h r e e i d e a l iz e d w a y s i n w h i c h t h e m o d e r a -

t o r a l t e rs t h e e f f e c t o f th e i n d e p e n d e n t v a r i a b l e o n t h e d e p e n d e n t

v a r i a b le . F i r s t , t h e e f f e ct o f th e i n d e p e n d e n t v a r i a b l e o n t h e d e -

p e n d e n t v a r i a b l e c h a n g e s l in e a r l y w i t h r e s p e c t t o t h e m o d e r a t o r .

T h e l i n e a r h y p o t h e s i s r e p r e s e n t s a g r a d u a l , s t e a d y c h a n g e i n t h e

e f fe c t o f t h e i n d e p e n d e n t v a r i a b l e o n t h e d e p e n d e n t v a r i a b l e a s

t h e m o d e r a t o r c h a n g e s . I t is th i s f o r m o f m o d e r a t i o n t h a t i s g e n -

e r a l l y a s s u m e d . T h e s e c o n d f u n c t i o n i n t h e f i g u r e i s a q u a d r a t i c

f u n c t i o n . F o r i n s t a n c e , t h e f e a r - a r o u s in g m e s s a g e m a y b e m o r eg en era l ly e f fec t iv e t h a n th e r a t i o n a l m essag e fo r a l l l o w- I Q su b -

j ec t s , b u t a s IQ in c reases , t h e f ea r - a ro u s in g m essag e lo ses it s ad -

v an tag e an d th e r a t i o n a l m essag e i s m o re e f fec tive .

T h e t h i r d f u n c t i o n i n F i g u r e 2 i s a s t e p f u n c t i o n . A t s o m e

cr i t i ca l IQ l ev e l , t h e r a t i o n a l m essag e b eco m es m o re e f f ec t iv e

t h a n t h e f e a r - a r o u s i n g m e s s a g e . T h i s p a t t e r n i s t e s t e d b y d i c h o t -

o m i z i n g t h e m o d e r a t o r a t t h e p o i n t w h e r e t h e s t e p i s s u p p o s e d

t o o c c u r a n d p r o c e e d i n g a s i n C a s e 1 . U n f o r t u n a t e l y , t h e o r i e s

i n s o c i a l p s y c h o l o g y a r e u s u a l l y n o t p r e c i s e e n o u g h t o s p e c i f y

t h e e x a c t p o i n t a t w h i c h t h e s t e p i n t h e f u n c t i o n o c c u r s .

T h e l i n e a r h y p o t h e s i s is te s t e d b y a d d i n g t h e p r o d u c t o f t h e

m o d e r a t o r a n d t h e d i c h o t o m o u s i n d e p e n d e n t v a r i a b le t o t h e r e -



1176 REUBEN M. BARON AND DAVID A. KENNY

gression equasion, as described by Cohen and Cohen (1983) and

Cleary and Kessler (1982). So if the independent variable is de-

noted as X, the moderator as Z, and the dependent variable as

Y, Y is regressed on X, Z, and XZ. Moderator effects are indi-

cated by the significant effect of X Z while X and Z are con-

trolled. The simple effects of the independent variable for

different levels of the moderator can be measured and tested by

procedures described by Aiken and West (1986). (Measurement

error in the moderator requires the same remedies as measure-

ment error in the independent variable under Case 2.)

The quadratic moderation effect can be tested by dichotomiz-

ing the moderator at the point at which the function is pre-

sumed to accelerate. If the function is quadratic, as in Figure 2,

the effect of the independent variable should be greatest for

those who are high on the moderator. Alternatively, quadratic

moderation can be tested by hierarchical regression procedures

described by Cohen and Cohen (1983). Using the same notat ion

as in the previous paragraph, Y is regressed on X, Z, XZ, Z 2,

and XZ 2. The test of quadratic moderation is given by the test

of XZ 2. The interpretation of this complicated regression equa-tion can be aided by graphing or tabling the predicted values

for various values ofXa nd Z.

Case 4

In this case both the moderator variable and the independent

variable are continuous. If one believes that the moderator al-

ters the independent-dependent variable relation in a step func-

tion (the bottom diagram in Figure 2), one can dichotomize the

moderator at the point where the step takes place. After dichot-

omizing the moderator, the pattern becomes Case 2. The mea-

sure of the effect of the independent variable is a regression co-

efficient.If one presumes that the effect of the independent variable

(X) on the dependent variable (Y) varies linearly or quadrati-

cally with respect to the moderator (Z), the product variable

approach described in Case 3 should be used. For quadrat ic

moderation, the moderator squared must be introduced. One

should consult Cohen and Cohen (1983) and Cleary and Kessler

(1982) for assistance in setting up and interpreting these regres-

sions.

The presence of measurement error in either the moderator

or the independent variable under Case 4 greatly complicates

the analysis. Busemeyer and Jones (1983) assumed that the

moderation is linear and so can be captured by an X Z product

term. They showed that measuring multiplicative interactions

when one of the variables has measurement error results in low

power in the test of interactive effects. Methods presented by

Kenny and Judd (1984) can be used to make adjustments for

measurement error in the variables, resulting in proper esti-

mates of interactive effects. However, these methods require

that the variables from which the product variable is formed

have normal distributions.

The Nat ure of Mediator Variables

Although the systematic search for moderator variables is rel-

atively recent, psychologists have long recognized the impor-

lance o f mediating variables. Woodworth's (1928) S-O-R

model, which recognizes that an active organism intervenes be-

tween stimulus and response, is perhaps the most generic for-

mulation of a mediation hypothesis. The central idea in this

model is that the effects of stimuli on behavior are mediated

by various transformation processes internal to the organism.

Theorists as diverse as Hull, Tolman, and Lewin shared a belief

in the importance of postulating entities or processes that inter-

vene between input and output. (Skinner's blackbox approach

represents the notable exception.)

General A nalyt ic C onsiderations

In general, a given variable may be said to funct ion as a medi-

ator to the extent that it accounts for the relation between the

predictor and the criterion. Mediators explain how external

physical events take on internal psychological significance.

Whereas moderator variables specify when certain effects will

hold, mediators speak to how or why such effects occur. For

example, choice may moderate the impact o f incentive on atti-

tude change induced by discrepant action, and this effect is in

turn mediated by a dissonance arousal-reduction sequence (of.

Brehm & Cohen, 1962).

To clarify the meaning of mediation, we now introduce a path

diagram as a model for depicting a causal chain. The basiccausal chain involved in mediation is diagrammed in Figure 3.

This model assumes a three-variable system such that there are

two causal paths feeding into the outcome variable: the direct

impact o f the independent variable (Path c) and the impact of

the mediator (Path b). There is also a path from the independent

variable to the mediator (Path a).

A variable functions as a mediator when it meets the follow-

ing conditions: (a) variations in levels of the independent vari-

able significantly account for variations in the presumed media-

tor (i.e., Path a), (b) variations in the mediator significantly ac-

count for variations in the dependent variable (i.e., Path b), and

(c) when Paths a and b are controlled, a previously significant

relation between the independent and dependent variables is no

longer significant, with the strongest demonstrat ion of media-

tion occurring when Path c is zero. In regard to the last condi-

tion we may envisage a continuum. When Path c is reduced to

zero, we have strong evidence for a single, dominant mediator.

If tbe residual Path c is not zero, this indicates the operation of

multiple mediating factors. Because most areas of psychology,

including social, treat phenomena that have multiple causes, a

more realistic goal may be to seek mediators that significantly

decrease Path c rather than eliminating the relation between the

independent and dependent variables altogether. From a theo-

retical perspective, a significant reduction demonstrates that a

given mediator is indeed potent, albeit not both a necessary and

a sufficient condition for an effect to occur.



THE MODERATOR-MEDIATOR DISTINCTION 117 7

T e s t i ng M e d i a t i o n

An ANOVAprovides a limi ted tes t ofa m ediat ional hypothes is

as extensively discussed in Fiske, Kenny, and Taylor (1982).

Rather, as reco mm ende d by Ju dd and Ken ny (1981 b) , a ser ies

of regress ion models should be es t imated. To test for mediat ion,

one sho uld estimate the thre e following regression equations:first , regressing the mediator on the independent variable; sec-

ond, regressing the dependent variable on the independent vari -

able ; and thi rd, regressing the dependent variable on both the

independent variable an d o n the mediator. Separate coeffic ients

for each equat ion should be es t imated an d tes ted. There is no

need for hierarchical or s tepwise regress ion or the c omp utat ion

of any part ia l or semipart ia l corre la t ions .

These three regress ion equat ions provide the tes ts o f the l ink-

ages of the mediat ional model . To es tablish mediat ion, the fol -

lowing condi t ions must hold: Firs t , the independent variable

must affect the mediator in the first equation; second, the inde-

pendent variable m ust be shown to affect the dependent variable

in the second equat ion; an d thi rd, the mediator must affect thedependent variable in the thi rd equat ion. If these condi t ions a l l

hold in the predicted di rect ion, then the effect of the indepen-

dent variable on the dependent variable must be less in the thi rd

equat ion than in the second. Perfect mediat ion holds i f the inde-

pendent variable h as no effect when the med iator is control led.

Because the independent variable is assumed to cause the me-

diator, these two variables should be correlated. The presence

of such a corre la t ion results in mul t icol l ineari ty when the

effects of independent variable and m ediator o n the dependent

variable are estimated. This results in reduce d power in the test

of the coefficients in the th ird equation . It is then critical tha t

the inves t igator examine not only the s ignificance of the co-

efficients but also their absolu te size. For instance, i t is possiblefor the independent variable to have a smaller coefficient when

i t a lone predicts the dependent variable than when i t and the

mediator are in the equat ion but the larger coeff ic ient i s not

significant and the smaller one is .

Sobel (1982) provided an approximate s ignif icance tes t for

the indirect effect of the independen t variable on the dependent

variable via the mediator . As in Figure 3, the path from the

independent variable to the mediator is denoted as a and i ts

s tandard error is sa ; the path from the mediator to the depen-

dent variable is denoted as b and its standard error is sb. The

exact formula , given mul t ivaria te no rmal i ty for the s tandard er-

ror o f the indirect effect or ab, is this:

Vb2sa2 q- a2Sb2 d- Sa2Sb

Sobel ' s method omits the term Sa2Sb , but that term ordinari ly

is small . His approxim ate metho d can be used for mor e compli -

cated models .

The use of mul t iple regress ion to es t imate a m ediat ional

mode l requires the two fol lowing assumptions : that there be no

measu remen t error in the mediator and that the dependent vari-

able not cause the mediator .

The mediator, because it is often an internal, psychological

variable, is l ikely to be measured with error. The presence of

measurement e r ror in the media tor t ends to p roduce an under -

es t imate o f the effect of the m ediator and an overes timate of

the effect of the independ ent variable o n the dependent variable

when all coefficients are positive (Judd & Kenny, 198 la). Ob vi-

ously this is not a desirable out com e, becau se successful med ia-

tors may be overlooked.

General ly the effect of measurem ent error is to a t tenuate the

size of measures of association, the resulting estim ate being

closer to zero than i t would be i f there were no measu remen t

error (Judd & Kenny, 198 la). Addi t ional ly, meas urem ent error

in the m edia tor is l ikely to result in a n overe stimate in the effect

of the independent variable on the dependent variable . Because

of measurem ent error in the m ediator, effects of the mediator

on the dependent variable can not tota l ly be control led for when

measuring the effects of the independen t variable on the depen-

dent variable.

The overes t imat ion o f the effects of the indepen dent variable

on the dependent variable is enhanced to the extent that the

independent variable causes the mediator and the mediator

causes the dependent variable. Because a successful mediator is

caused by the independent variable and causes the dependent

variable , successful mediators measured with error are most

subject to this overestimation bias.The common approach to unrel iabi l i ty is to have mul t iple

operat ions or indicators of the cons truct . Su ch an ap proac h re-

quires two or m ore operat ional izat ions or indicators of each

cons truct . One can use the mul t iple indicator appro ach and es-

t imate mediat ion paths by la tent-variable s t ructural model ing

methods . Th e majo r advantages of s t ructural mod el ing tech-

niques are the following: First , although these techniques were

developed for the analysis of nonexpe rimen tal data (e.g., field-

correlational studies), the experimental context actually

strengthens the use of the technique s. S econd, all the relevant

paths are di rect ly tes ted and none are omit ted as in ANOVA.

Third, com plicat ions of measu remen t error , corre la ted mea-

surement error , and even feedback are incorporated di rect lyin to the mode l. The mos t com mo n compute r p rogram used to

es t imate s t ructural eq uat ion m odels is LISR EL- VI (JSreskog

& S6rbom, 1984). Also available is the program EQS (Bentler,

1982).

We now turn our a t tent ion to the second source of bias in

the mediat ional chain: feedback. The use of mul t iple regression

analys is presumes that the m ediator is not caused by the depen-

dent variable . I t may be poss ible that we are mis taken about

which variable is the mediator and wh ich is the dependent vari -

able.

Smith (1982) has proposed an ingenious solut ion to the prob -

lem o f feedback in mediat ional chains . His m ethod involves the

manipulat ion of two variables, one p resume d to cause only themediator and not the dependent variable and the other pre-

sumed to cause the dependent variable and not the mediator .

Mode ls of this type are estimated by two-stage least squares or

a re la ted technique. Int roduc t ions to two-s tage leas t squares are

in James and Singh (1978), Duncan (1975), and Judd and

Kenny (1981a). The earl ier-ment ioned s t ructural model ing

procedures can a lso be used to es t imate feedback models .

O v e r v i e w o f C o n c e p t u a l D i s t i n c ti o n s

B e t w e e n M o d e r a t o r s a n d M e d i a t o r s

As shown in the previous section, to demonstra te mediat ion

one must establish strong relations between (a) the predictor




and the mediating variable and (b) the mediating variable and

some distal endogenous or criterion variable. For research ori-

ented towar d psycholog ical levels of explanation (i.e., where th e

individual is the relevant unit of analysis), mediators rep resent

properties of the person that t ransform the predictor or input

variables in some way. In this regard the typical mediator in

cognitive social psychology elaborates or constru cts the vario us

meanings that go "beyond the informat ion given" (Bruner ,

1957). However, this formulation in no way presupposes that

mediato rs in social psychology are limited to individualistic or

" in the head" mechanisms. Group-level mediator cons tructs

such as role conflict, norms , grou pthink , and cohesiveness have

long played a role in social psychology. Moreover, with the in-

creasing interest in applied areas, there is likely to be a n increas-

ing use of mediators form ulated at a broad er level of analysis.

For example , in the area of environmen tal psychology, terr ito-

rial constructs such as defensible space (Newman , 1972) or the

role of sociopetal versus sociofugal sit t ing patterns (Sommer,

1969) c learly take the m ediator co ncept beyon d the int raorga-

nismic level. Despite this range of application of the mediato rconcep t, i t is in principle cap able of rigorous tests at the group

level. For example, Zaccaro (1981) has attempted to support a

med iator interp retation of cohesiveness using a strategy com-

bining experimental m anipulat ions wi th causal modeling.

In addition, whereas mediator-oriented research is more in-

terested in the m echanism than in the exogenous variable i t se l f

(e.g., dissonance and per sonal-con trol mediators have been im-

pl icated as explaining an a lmost unen ding varie ty of predic-

tors), moderator research typically has a greater interest in the

predic tor variable p er se. However, whether a given moderator -

oriented investigation is strongly com mit ted to a particular p re-

dictor is l ikely to v ary widely. Althou gh a pr agmatic- predicto r

orientation is typical in industrial psychology, where the pred ic-tor is often a test, in social psychology mod erators are often as

theoretically derived as mediators.

Strategic Considerations

Mode rator variables are typical ly int roduced when there is

an unexpectedly weak or inconsistent relation between a pre-

dictor an d a criterion variable (e.g., a relation holds in one set-

r ing but not in another , or for one subpopulat ion but not for

another). T he re cent use by Snyd er (1983) and others (ef. Sher-

ma n & Fazio, 1983) of the self-monitoring variable as a mean s

to imp rove the ability of personality traits to predict behavioral

criteria is i l lustrative. M ediation, on the other hand, is best donein the case of a s t rong rela t ion between the predictor and the

criterion variable.

Moderator to m ediator. In addition, th ere m ay be a wide vari-

a t ion in the s t ra tegic funct ions served by mod erators and medi-

a tors . In this regard one m ay begin wi th a mo derator or ienta-

t ion and end up e lucidating a med iator process , or begin wi th a

mediator approach and derive moderator- type intervent ions .

For example , le t us assume that race funct ioned as a mod erator

for the efficacy of certain instruction al techn iques, such tha t a

given techniq ue (e.g., pro gra mm ed instruction) work s better for

one racia l group than for another. O ne couM view such a f inding

as just the first step toward specifying the und erlying dimen -

sion(s) that acco unt fo r the instruc tional effect. Fo r example, i t

could be argued that the real issue is a difference in anxiety

level; that is , when black a nd white childre n are plac ed in m id-

die-class learning environm ents, black children m ay experience

a h igher level of evaluative anxiety. Th erefore, evaluative-anxi-

ety level may be postulated to mediate the differential effec-

tiveness of a given instructiona l techn ique. Thus, here we have

a situation where a moderator variable has been useful in sug-

gesting a possible mediator variable. What is at stake in this

regard is se lecting moderators that do more than improve pre-

dictive power. Fo r example, r ace wou ld be preferre d o ver social

class as a mo dera tor if race was mo re able to tell us somethin g

about the processes underlying test performance.

A s imilar point can be made in regard to the current use of

mod erato r variables in personality research. T hat is, if two vari-

ables have equal power as potent ia l moderators of a t ra i t -behav-

ior relation, one should choose the variable that more readily

lends itself to a specification of a mediation al m echa nism. For

example, the self-monitoring variable both improves predictive

efficacy and suggests mediational processes involving atten tion

deploy ment. Inde ed, such a strategy of selection points to o neway to c i rcumve nt the oft -made cr i t ic ism of modera tor vari -

ables that we have no principled p rocedure for reducing their

proliferation (of. Epstein, 1983).

Mediator to moderator. The re la t ion may also work in the

opposite direction. Differences in perceived control may be

found to mediate the re la t ion between social dens i ty and decre-

ments in task performance. In this s i tuat ion a mediator may

suggest an environmental intervent ion to prevent dens i ty from

having adverse effects. For example, wha t appears to b e need ed

is an intervention that wo uld serve to increase the con trollabil-

i ty of socia l encounters . This m ight take the form of archi tec-

tural variation, for example, suite versus corridor dorm ar-

rangements, o r involve various types of restrictions on cha ngeor unpredictable social encounters, fo r example, institution of

quiet hours. W hat is at stake here is the choice o f mediato rs that

point to the poss ibi l ity of environmen tal intervent ion.

Thus, at t imes moderator effects may suggest a mediator to

be tested at a mo re adv anced stage of research in a given area.

Conversely, mediators may be used to derive interventions to

serve applied goals.

Operat ional Implicat ions

There a re a n umber of impli cat ions of the m odera tor -media -

tor distinction at the level of the choice of research o perations.

Firs t , the mod erator interpreta t ion of the re la t ion between thestressor and co ntrol typically entails an experim ental man ipula-

t ion o f control as a m eans of es tabl ishing independence between

the s t ressor and control as a feature of the environm ent separate

from the stressor. W hen co ntrol is experimen tally man ipula ted

in service of a moderator funct ion, one need not m easure per-

ceived control, which is the cognitive intraorganismic concept.

If i t is measured, perceived control serves as a manip ulatio n

check.

A theory that assigns a mediator role to the control cons truct ,

however, is only secondarily concerned with the independent

manipulat ion of control . The most essent ia l feature of the hy-

pothes is i s that perceived control i s the mechanism through

which the stressor affects the o utco me variable. F or such a the-




Figure 4. Path diagram com bining mediation and moderation.

o ry , an indepen den t a s ses sm ent o f pe rce ived con t ro l i s e s sen t i a l

fo r concep tua l reasons , a s opposed to m e thodolog ica l reasons

as in the m od era to r case . Because o f the concep tua l s t a tus o f

th i s a s ses sm ent in the m edia to r case , one ' s m a in conce rn i s the

dem o ns t ra t io n o f cons t ruc t va l id ity , a s i tua t ion tha t idea l ly re -

qu i re s m ul t ip le independen t and conve rg ing m easurem ents

(Cam pbe l l & F i ske , 1959) . Thus , when m edi a t ion i s a t i ssue we

need to inc rease bo th the qua l i ty and the quan t i ty o f the da ta .

A F ramework or C ombin ing Med ia tion and Modera t ion

F i g u r e 4 p r e s e n t s a c o m b i n e d m o d e l w i t h b o t h m e d i a t i o n

a n d m o d e r a t i o n . T h e v a r i a b l e c o n tr o l h a s b o t h m e d i a t o r a n dm od era to r s t a tus in the m ode l . T he s t re ssor in the f igure i s the

indepen den t va r iab le , and the dependen t va r iab le i s l abe led the

ou tcom e . We denote m anipu la ted con t ro l a s C , the s t re s sor a s

S , the C • S in te rac t ion a s CS , m easured pe rce ived con t ro l as

E the P X S in te rac t ion a s P S, the C x P in te rac t ion a s CE the

C X P X S in te rac t ion a s CP S , and the ou tcom e as O. We as sum e

tha t bo th the m an ipu la t ion o f con t ro l and the s t re s sor a re d i -

cho tom ies a nd tha t a l l m odera to r e f fec t s a re l inear .

The analys is proceed s in three s teps . In S tep 1, the effects of

the m anipu la ted va r iab le s on O a re a s ses sed . In S tep 2 , the

effects to an d fr om P are assessed. In S tep 3, the effect from PS

is assessed.

Step 1. The S tep 1 regres s ion i s i l lus t ra ted in F igure 1 . Th i ss tep is a s imple 2 X 2 ANOVA on th e ou tcom e varia ble . If C

has a s ign i f i can t e f fec t on O, then con t ro l m ay be a m ed ia t ing

varia ble of the s t ressor effect on th e outc ome. If S affects O,

then i t i s s ens ib le to eva lua te the m edia t ing e f fec ts o f pe rce ived

cont ro l . These two e f fec ts a re suppor t ive o f the m edia t io n hy-

po thes i s , bu t d i rec t ev idence fo r m edia t ion i s p rov ided in the

next s tep. F inal ly, the CS effect indicat es moder at ion .

Step 2. The S tep 2 regres s ions a re i l lus t ra ted in F igure 4 . In

this s tep, two equat ions are es t imated. F irs t , P is regressed on

C, S , and CS . Th is can be m or e eas i ly accom pl i shed by a 2 • 2

ANOVA. Seco nd, O is regr essed o n C, S, P, and CS. For P to

mediate the S to O re la t ion, S must affect P and P must affect

O. I f the re i s com ple te m ed ia t ion , then S does no t a f fec t O when

P i s con t ro l l ed . To s t reng then the c la im tha t i t i s pe rce ived con-

t ro l tha t m edia te s the re la t ion , C should s t rong ly a f fec t P b u t

should no t a f fec t O . I fC a f fec t s O, then i t i s ind ica ted tha t som e

aspec t the con t ro l m an ipu la t io n i s d if fe ren t f rom pe rce ived con-

trol .

The re a re two rem a in ing pa ths in S tep 2 . They a re the pa ths

f rom CS to P and to O. I fCS a f fec t s P , then the con t ro l m an ipu-

la t ion i s no t equa l ly e ffect ive in de te rm in ing pe rce ived con t ro l

across the levels of the s t ressor. The s tressor mod erat es the

e f fec t ivenes s o f the m an ipu la t ion . The f ina l S tep 2 pa th i s the

one f rom CS to O. Le t us a s sum e tha t CS a f fec t s O in the S tep

1 regress ion, and in S tep 2 CS has a w eaker effect on O. Then

the in te rpre ta t ion i s tha t P has m edia ted the CS e f fec t on O.

W e h a ve w h a t m i g h t b e t e r m e d media ted moderation. M e d i a t e d

m ode ra t ion w ould be ind ica ted by CS a f fec t ing O in S tep 1 , and

in S te p 2 CS affect ing P and P affect ing C. So i t is poss ible for P

to m e dia te bo th the e f fec t o f S on O and the e f fec t o f CS on O.

Step 3. In th i s s t ep, one equ a t ion i s e s t im a ted . The va r iab le

O is regressed on C, S , P, CS, and PS. This equ at ion is ident ic al

to the s econd S tep 2 equa t ion , bu t the P S te rm has been added .The key ques t ion i s the ex ten t to which the CS e f fect on O i s

reduced in m oving f rom S tep 2 to S tep 3 . I f i t has been , then we

can say tha t P an d no t C m odera te s the S to O re la t ion . In a

sense , P me diates the mo dera t ing effects of C on S . For this to

happen , CS m us t have l es s o f an e f fec t on O a t S tep 3 than a t

S tep 2, and PS must affect O. F inal ly in S tep 2, C shou ld affect

P , which w i l l re su l t in CS and P S be ing cor re la ted .

The re a re then two ways in which the C S e f fec t on O can be

expla ined by E I t can be exp la ined by P because the con t ro l

m anip u la t ion i s d i f fe ren t i a l ly a f fec t ing pe rce ived con t ro l fo r the

levels of the s t ressor. Or, the CS inter act io n can be funn el led

through the P S in te rac t ion . The fo rm er exp lana t ion would

change wha t was a m odera to r e f fec t in to a m edia to r e f fec t , andthe l a t t e r would keep the m odera to r exp lana t ion bu t enhance

t h e m e a n i n g o f t he m o d e r a t o r c o n s t r u c t.

We presen t the th ree s t ep hypotheses because they represen t

a s e r i e s o f reasonab le hypotheses . I f one wished , fu r the r m ode ls

cou ld be e s t im a ted . F or ins tance , one cou ld regress O on C , S ,

E CS , and CE The presence o f the CP e f fect , a s we ll as m edia -

t iona l e f fec ts by P o f the S to O re la t ion , would be ind ica t ive o f

moderated mediation (Jame s & Bret t , 1984). That is , the med i-

a t iona l e f fec t s o f P va ry ac ros s the l eve l s o f C . The s econd-orde r

in te rac t ion e f fect , CP S, cou ld a l so be e s t im a ted an d t e s ted .

I m p l i c a t i o n s a n d A p p l i c a t i o n s o f t he

M o d e r a t o r - M e d i a t o r D i s t i n c t i o n

In th i s s ec t ion , we t ake the them es deve loped in the th ree p re -

v ious s ec t ions and app ly the m to th ree a reas o f soc ia l psycholog-

ica l re sea rch . These a reas a re pe rsona l con t ro l , the beha v ior -

in ten t ion re la t ion , and l ink ing t ra i t s and a t t i tudes to behav ior .

Clar i fying the Meanin g o f Control

Man y inves t iga t ions o f the im pac t o f pe rsona l con t ro l in so -

c ia l and env i ronm enta l psychology have been m e thodo log ica l ly

(bu t no t theore t i ca l ly ) am biva len t wi th re spec t to the con t ro l

va r iab le ' s causa l s t a tus. Inves t iga to rs have t ended to use expe r i -

m enta l m anip u la t ions o f pe rsona l con t ro l a long wi th ANOVA-




type analyses. This practice leads to serious difficulties of inter-

pretation when a researcher intends to investigate one function

of control but s tudies only the other funct ion. For example ,

Langer and Saegert (1977) and Rodin, Solomon, and Metcalf

(1978) sought to examin e the m ediat ional role of lessened con-

trol for crowding. Given this mediator interpretation, i t is not

enough to dem onstra te by use of an experimental m anipulat ion

that high dens i ty creates more perceived crowding than does

low density only when there is a low availability o f control, for

example, the ability to escape from the high-density situation.

To provide s t ronger evidence of mediat ion, an independent as-

sessment of the im pact of the s t ressor on some index o f organis-

mic co ntrol is required. O nly when this is don e can we establish

the crucial l ink between perceived control and the criterion. Be-

cause the Langer and Saegert and Rodin, So lomon, and Metcalf

studies failed to provide an inde penden t assessment o f control,

they lack the requisite info rma tion to establish a strong case for

control as a mediator. Moreover, because I_anger and Saegert

failed to find differential effects for density un der varying levels

of heir control manipulat ion, that i s, a Control • Dens i ty inter-

act ion, they are not even in a pos i t ion to m ake m oderator-vari -

able claims.

Final ly, there is another impo rtant role that the present mod-

erator-m ediator dis t inct ion can play in the domain o f crowding

theory and research. Al though a control-mediat ion model of

crowd ing is generally accepted (e.g., Bar on & R odin, 1978; Sto-

kols, 1976), there are significant dissenters such as Freedman

(1975). Given the presen t status o f the evidence, i t appears

muc h eas ier to suppo rt the c la im that co ntrol moderates , as op-

pose d to mediates, the density-crowd ing relation. Such an inter-

pretation would leave open the possibili ty that other factors,

such as an arousal- labeling or an arousal-ampli f icat ion mecha-

nism, media te the effects of density (i .e., Fre edm an, 1975, Wor-chel & Teddlie, 1976).

Behavior Intent ion-Behavior Relation

Because Fishbein and Ajzen's (1975; Ajzen & Fishbein,

1980) atti tude th eory o f reasone d action is in general highly so-

phisticated at both the conceptual and quantitative levels, i t

provides a goo d example of the extent of confus ion regarding

mediators and moderators. Moreover, this model, as Bentler

and Speckart (197 9) have demonstr ated, readily lends itself to

a causal modeling ap proa ch. Specifically, behavioral inten tion

(BI) is a clear-cut example of a med iator c once pt in social psy-

chology. Fishbein and A jzen assumed that the im pact of bothattitudes and normative factors on behavior (B) is mediated

through behavioral intent ions. Al though one can disagree wi th

Fishbein and Ajzen's assert ion that a t t itudes and n orms can in-

f luence behavior only indirect ly through behavioral intent ion

(see Bentler & Speckart, 1979; Songer-Nocks, 1976), their for-

mulat ion represents a correct s ta tement of a s t rong mediator

position.

Surprisingly, however, given the elegance o f their gen eral

model , s imilar care was not taken regarding the nature of the

BI- B l ink. For example , Fishbein and Ajzen's treatment of this

relation failed to distinguish between variables that are likely to

mod erate a nd those likely to med iate this relation. S uch diverse

variables as gender, t ime delay, perceived likelihood o f co-work-

ers complying, skill , and resources were all treated as m ediating

factors (Fishbein & Ajzen, 1975, pp. 377-3 81).

From the present perspect ive , such an a pproa ch ignores the

possibility that som e of these factors are best concep tualized

and treated statistically as moderators whereas others are best

viewed as mediators. For example, gender of subjects is bestviewed as a mod erator of the BI- B re lat ion. Given this dis t inc-

tion, different analysis strategies are entailed at the statistical

level. Specifically, Fishbein a nd Ajzen tested the im por tanc e of

given factors by looking at the impact on the multiple correla-

t i6n o f dropping or adding a variable . Th is type o f s tra tegy,

which is analogous to treating a covariate as a potential m edia-

tor, is best used to infer mediation as opposed to moderation.

For testing a moderator interpretation, what is required is a

term involving the produ ct of BI and the hypothes ized modera-

tor; for example , one would cons truct a Gender • BI interact ion

term to test for gender as a mo dera tor variable.

Finally, although Fishbein's (1966) finding that inte ntions are

better predictors for wom en than for men is in itself best viewed

as a mo dera tor effect, a sensitivity to the pre sent set of issues

prom pts further analyses . For example , i f we ask why gender

has such effects on sexual intentions, i t is possible that we will

be led to postulate a media tor that transc ends gender. Fo r exam-

ple, i t might be argued that intent ions predict bet ter for wom en

because women are less impuls ive than men in regard to the

timing of sexual behavior.

Linking Global Dispositions to Behavior:Attitudes and Traits

Of all the cu rren t areas in social psychology, the on e where

the use of what we have referred to as the combine d m odel (see

Figure 4) is perhaps the strongest is the prediction of social be-havior from global dispositional variables. In this regard, the

t ra i t -behavior an d the a t t i tude-behavior re la tions have recently

been expl ic i t ly approached from the moderator-variable per-

spective. For example, the predictive efficacy of bot h traits and

attitudes have improved when self-monitoring (Snyder, 1983)

and self-consciou sness (Scheier, 1980), respectively, have been

used as moderator variables. Moreover, investigators such as

Snyder and Ickes (1985) and Sherm an and Fazio (1983, p. 327)

have asked the following questions: By what process or pro-

cesses do atti tudes toward an object affect behavior toward the

object? Likewise, what conceivable processes link traits to be-

havior?

Wh at su ch suggestions lack is precisely the kin d o f unifiedconceptual and analyt ic framework presented in our com bined

moderator-mediator example (see Figure 4) . By us ing such a

path analytic framework, one could take a variable such as

differences in self-monitoring orientation and simultaneously

es tabl ish both i t s role as a mo derator an d the natu re of the me-

diat ion process through which i t has an im pact o n a given class

of behavior. At an operationa l level, such a strategy comp els one

to go beyond merely measuring differences in se l f-moni toring

(the moderator paths) to operat ional iz ing a mediator mecha-

nism, for example, pro viding some mea sure o f differential at-

tent ion or variables in impress ion m anagement .

Further , placing both moderator and mediator variables

within the same causal system helps to make salient the more




dynamic role played by mediators as opposed to moderators

(Fi nne y, Mitche ll, Cronki te, & Moos, 1984). Specifically, in tro -

ducing a moderator variable merely involves a relatively static

classification procedure. For example, self-monitoring as a

moderator sets up a parti tion of people holding a given personal-

ity trait into subgroups of those mo re or less likely to translate

their psychological dispositions into overt actions; that is, the

emphasis is on who does what. O n the other hand, linki ng the

Self-Monitoring • Trait relation to a specific mediating mecha-

nism implies that variations in self-monitoring elicit or insti-

gate different patterns o f coping or in form ation processing that

cause people to become more or less consistent with their atti-

tudes in their behavior. Here the p rior co ndition allows us to

discover different states that cause ind ividu als to act differ-

ent ly- -a more dynami c conception of how third variables op-

erate.

S u m m a r y

In this article we have attemp ted to achieve three goals. First,by carefully elaborating the m any ways in which moderators

and mediators differ, we have tried to make theorists and re-

searchers aware of the imp ortanc e of not using the terms m o d -

erator an d media tor interchangeably. We then went b eyond this

largely pedagogical functio n and delineated the conceptual a nd

strategic implications of making use of this distin ction with re-

gard to a wide range of pheno mena, includ ing control and

stress, attitudes, and personality traits. We have also provided

the first specific compen diu m of analytic procedures appropri-

ate for makin g the most effective use of the moder ator-m edia-

tor distinctio n both separately and in terms of a broader causal

system that includes both moderators and mediators.

References

Aiken, L. S., & West, S. G. (1986). Use and interpretation of regression

analysis models containing interactions an d power polynomials. Un-

published manuscript, Arizona State University,Tempe.Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predict-

ing social behavior. Englewood Cliffs, NJ: Prentice-Hall.

Baron, R. M., & Rodin, J. (1978). Personal control as a mediator ofcrowding. In A. Baum, J. Singer, & S. Valins (Eds.),Advances in envi-

ronmentalpsychology (pp. 145-190). Hillsdale, NJ: Erlbaum.

Bentler, P. M. (1982). Theory and implementation of EQS: A structuralequations program. Unpublished manuscript, University of Califor-nia, Los Angeles.

Bentler, P. M., & Speckart, G. (1979). Models of attitude-behavior re-

lations. Psychological Review, 86, 452-464.Brehm, J. W., & Cohen, A. R. (1962). Exploration s in cognitive disso-

nance. New York: Wiley.

Bruner, J. S. (1957). Going beyond the informationgiven. In H. Gruber,G. Terrell, & M. Wertheimer (Eds.), Contemporary approaches to

cognition (pp. 258-290). Cambridge, MA: Harvard University Press.

Busemeyer, J. R., & Jones, L. E. (1983). Analysis of multiplicative com-

bination ules when the causal variables are measured with error. Psy-

chological Bulletin, 93, 549-563.

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminantvalidation by the multitr ait-multimethodmatrix. Psychological Bul-

letin, 56, 81-105.Cleary, P. D., & Kessler, R. C. (1982). The estimation and interpretation

of modified effects.Journ al o f Health and Social Behavior, 23, 159-

169.

Cohen, J., & Cohen, P. (1983). Applie d m ultiple regression~correlation

analysis for the behavioralsciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Collins, B. E., & Hoyt, M. G. (1972). Personal responsibility for conse-

quences: An integration and extension of the forced compliance liter-

ature. Journal of Experim ental S ocial Psychology, 8, 558-593.

Duncan, O. D. (1975).Introduction to structural equation m odels. New

York: Academic Press.Epstein, S. (1983). Aggregation and beyond: Some basic issues in the

prediction of behavior.Journ al o f Personality, 51, 360-392.

Findley, M. J., & Cooper, H. M. (1983). Locus of control and academic

achievement: A literature review.Journal o f Personality a nd So cialPsychology, 44, 419-427.

Finney, J. W., Mitchell, R. E., Cronkite, R., & Moos, R. H. (1984).

Methodological issues in estimating main and interactive effects: Ex-amples from the coping/social support and stress field. Journal of

Health and So cial Behavior, 10, 85-98.

Fishbein, M. (1966, November). Sexual behavior and propositional

control. Paper presented at the meeting of the Psychonomic Society,Chicago.

Fishbein, M., & Ajzen, I. (1975). Belief attitude, intention and behav-ior. Reading, MA: Addison-Wesley.

Fiske, S. T., Kenny, D. A., & Taylor, S. E. (1982). Structural models for

the mediation of salience effects on attribution. Journal of Experi-

men tal Socia l Psychology, 18, 105-127.

Freedman, J. L. (1975). Crowding and behavior. San Francisco: Free-man.

Glass, D., & Singer, J. (1972). Urban stress: Experiments on noise and

social stressors. New York: Academic Press.

Harkins, S. G., Latan6, B., & Williams, K. (1980). Social loafing: Allo-

cating effort or taking it easy? Journal o f Experime ntal Social Psy-

chology, 16, 457-465.

James, L. R., & Brett, J. M. (1984). Mediators, moderators, and testsfor mediation. Journal ofApplie d Psychology, 69, 307-321.

James, L. R., & Singh, B. K. (1978). An introduction o logic, assump-

tions, and basic analytic procedures of two-stage least squares. Psy-

chological Bulletin, 85, 1104-1123.J6reskog, K. J., & S6rbom, D. (1984). LISREL-VI--Est imation of in-

ear structural equations by maximum likelihood methods (3rd ed.).Mooresville, IN: Scientific Software.

Judd, C. M., & Kenny, D. A. (1981a). Estimatin g the effects o f social

interventions. New York: Cambridge University Press.

Judd, C. M., & Kenny, D. A. (1981b). Process analysis: Estimating me-

diation in evaluation research. Evaluation Research, 5, 602-619.

Kenny, D. A., & Judd, C. M. (1984). Estimating the nonlinearand inter-active effects of latent variables. Psycholog ical Bulletin, 96, 201-210.

Langer, E. J., & Saegert, S. (1977). Crowding and cognitive control.

Journal of Personality a nd S ocial Psychology, 35, 175-182.

McNemar, Q. (1969). Psychological statistics (4th ed.). New York: Wi-

ley.

Newman, O. (1972). Defensible space. New York: Macmillan.Rodin, J., Solomon, S., & Metcalf, J. (1978). Role of control in mediat-

ing perceptions of density.Journal of Personality and Social Psychol-

ogy, 36, 988-999.

Scheier, M. E (1980). The effect of public and private self-consciousness

on the public expressions of personal beliefs. Journal of Personality

and S ocial Personality, 39, 514-521.

Sherman, S. J., & Fazio, R. H. (1983). Parallels between attitudes and

traits as predictors of behavior.Journal ofPersonality, 51, 308-345.

Smith, E. R. (1982). Beliefs, attributions, and evaluations: Nonhierar-

chical models of mediation in social cognition.Journal ofPersonality

and S ocial Psychology, 43, 248-259.

Snyder, M. (1983). The influence of individuals on situations: Implica-tions for understanding the links between personality and social be-

havior.Journal of Personality, 51, 497-516.




Snyder, M., & Ickes, W. (1985). Personality and social behavior. In G.

Lindzey & E. Aronson (Eds.), Handbook of social psychology (3rded., pp. 883-9 48). Reading, MA" Addison-Wesley.

Sobel, M. E. (198 2). Asymptotic co nfidence intervals for indirect effectsin structural equations models. In S. Leinhart (Ed.), Sociologicalmethodology 1982 (pp. 290-312). San Francisco: Jossey-Bass.

Sommer, R. (1969). Personal space. Englewood Cliffs, NJ: Prentice-Hall.

Songer-Nocks, E. (197 6). Situational factors affecting the w eighting of

predictor components in the Fishbein model. Journal o f Experimen-tal Social Psychology, 12, 56-69.

Stern, G. S., McCants, T. R., & P ettine, P. W. (1982). The relative contri-bution of controllable and unco ntrollable life events to stress and ill-ness. Personality and Social P sychology Bulletin, 8, 140-145.

Stokols, D. (1976). The experience of crowding in prim ary an d second-ary environments. Environment a nd Behavior, 8, 49-86.

Winer, B. J. (1971). Statistical principles in experimental design (2nded.). New Yo rk: McGraw -Hill.

Woodworth, R. S. (1928). Dynam ic psychology. n C. Murchison (Ed.),Psychologies o f 1925. Worcester, MA: C lark Un iversity Press.

Worchel, S., & Teddlie, C. (1976). Th e experience o f crowding. Journalo fPersonality and Social Psychology, 34, 30-40.

Zaccaro, S. J. (1981). The effects o f cohesion source on process loss ingroup performance. Unpublished doctoral dissertation, University ofConnecticut, Storrs.

Received August 7, 1985

Revision received July 14, 1986 9

I n s t r u c t i o n s t o A u t h o r s

Authors should prepare manuscripts according to the Publ icat ion M anua l o f the American Psy-

chological Association (3rd ed.). Articles not prepa red a ccordin g to the guidelines of the Man-

ual will not be reviewed. Al l manuscripts m ust include an abs t ract of 100-150 words typed on

a separate sheet of paper. Typing instructions (all copy must be double-spaced) an d instru ctions

on preparing tables, figures, references, metrics, and abstracts appear in the Manual . Also, all

manu scripts are subject to ed iting for sexist language.

APA pol icy prohibi ts an autho r from subm it t ing the same man uscript for concurrent cons id-

erat ion by two or m ore journals . Also, authors of manuscripts subm it ted to A PA journals are

expected to have available their raw data thr oug hou t the editorial review process and for at least

5 years af ter the date of publ icat ion. For further inform at ion on content , authors should refer

to the editorial in the Marc h 1979 issue of this jour nal (Vol. 37, No. 3, pp. 468 -469 ).

The reference citation for any article in any J P S P section follows APA's standard referencestyle for jour nal articles; that is , authors, yea r of publication, article ti t le, journal t i tle, volu me

number, and page numbers . The c i ta t ion does not include th e section title.

Authors will be required to state in their initial submission letter or sign a statement that they

have complied with APA ethical s tandards in the t reatment of thei r sample , hu man or animal .

A copy of the APA Ethical Principles m ay be obtain ed fr om the A PA Ethics Office, 1200 17th

Street, N.W., Washington, DC 20036.

Ano nym ous reviews are opt ional, and authors who wish anony mou s reviews must specifical ly

reques t them when submit t ing their manuscripts . Each copy of a manuscript to be ano nymo usly

reviewed should include a separate ti t le page with authors ' names and affil iations, and these

should not ap pear anyw here e lse on the manuscript . Footnotes that ident ify the authors should

be typed on a separate page. Authors should ma ke every effort to see that the m anus cript i tself

contain s no clues to their identities.

Manuscripts should be submitted in quadruplicate, and all copies should be clear, readable,and on paper o f good quali ty. A dot m atr ix o r unusual typeface is acceptable only i f i t is c lear

and legible. Dittoed and mimeographed copies will not be considered. Authors should keep a

copy of the manusc ript to g uard ag ainst loss. Mail manuscrip ts to the appro priate section editor.

Edi tors ' addresses appear on the ins ide fro nt cover of the journal .

Section editors reserve the right to redirect papers among themselves as appropriate unless

an au thor specifically requests otherwise. Reje ction by o ne section editor is considered rejection

by all, therefore a manu script rejected by one section editor should no t be submitted t o another.

Articol mediere

Documents

Transcript of Articol mediere