LOURNAL
OF EXPERIMENTAL
SOCIAL
Misconceptions
PSYCHOLOGY
Revisited: MARTIN
University
12,
591-593
(1976)
A Final Comment
FISHBEIN
of Illinois,
Urbuna-Chnmpaign
AND ICEK University
AJZEN
of Massachusetts,
Amherst
Received July 8, 1976 Although Songer-Neck’s reanalysis of her data is consistent with our earlier comments, her reply indicates that some of her misconceptions persist.
We regret that once again we feel compelled to comment on SongerNecks’s misrepresentation of our theoretical position. In her reply, Songer-Necks (1976b) attempted to justify her use of behavior, rather than intention, as her dependent variable. She argued that “it is of empirical importance to specify factors which affect the BI-B relationship or, equivalently, the relationship between [(Aact)w, + (NBs)w,] and B” (p. 586). Although we concur in the importance of studying the relation between intention and behavior, we strongly disagree that this is equivalent to investigating the relationship between the model’s two components and behavior. If Songer-Necks had truly been interested in the intention-behavior relationship, it would not have been necessary for us to suggest that she conduct regression analyses substituting intentions for attitudes and normative beliefs. In fact, as she herself admits, her real interest was (and unfortunately still is) in “assessing the relationship between attitudinal components and actual behavior” (p. 585). In our initial comments (Fishbein & Ajzen, 1976) we tried to make it clear that it is inappropriate to go directly from attitudes and subjective norms to behavior. Although it may occasionally be of interest to examine the empirical relation between the model’s two components and actual behavior, such an analysis is theoretically meaningless unless: (1) the relationship between the model’s two components and intention and (2) the relationship between intention and behavior Send reprint requests to Martin Fishbein. Department Illinois, Champaign, Illinois 61820. 591 Copyright All rights
0 1976 by Academic
Press, Inc.
of reproductionin anyformreserved.
of Psychology, University
of
592
FISHBEIN
AND AJZEN
are known. Evidently, Songer-Necks still fails to appreciate the importance of this fact. Her misunderstanding of this issue is reflected in her continued interest in her finding that “under certain conditions neither the attitudinal nor the normative component predicts actual behavior” (p. 587). Apparently, she still fails to realize that this finding follows directly from her finding that, under these same conditions, BI contributes relatively little to the prediction of behavior. Clearly, if the model leads to accurate prediction of intentions, but the intentions do not predict behavior, it follows that the model cannot predict behavior. We are grateful to Songer-Necks for reanalyzing her data, and we are gratified to note that her results provide complete confirmation of all the points we made in our original comments. Variables external to our model were found to affect intentions only indirectly by influencing either of the two components or their relative weights. That is, the experimental manipulations had no independent effects on intentions and, as expected, the relative weight of the normative component varied as a function of motivational orientation. The latter effect is evidenced by the significant normative beliefs by motivation interaction. Given this finding, it becomes clear that the independent effects of the external variables on behavior reported in Songer-Necks’s ( 1976a) original study were due to the influence of those variables on the intentionbehavior relation. Empirical confirmation of this is provided by SongerNecks’s analysis regressing behavior on intentions and external variables. Although we are heartened that Songer-Necks now agrees that her data do not question the generality or utility of our model of intentions or of our more general theory of behavior, we are disappointed that she is still encouraging people to investigate the relations between the models two components and behavior without pointing out that this is only meaningful if the relations between the components and intention and between intention and behavior are known. In addition, it should be noted that Songer-Necks’s data do not “indicate that the prediction of behavior from attitudinal components has certain limitations which are potentially specifiable” (p. 589). In marked contrast, what her data do indicate is that external variables influence the strength of the intention-behavior relationships but not the strength of the relation between the model’s components and intentions. Thus, consistent with our theory, the prediction of behavior from the attitudinal and normative components was limited only by the strength of the intention-behavior relationship. Clearly then what does need to be further specified are those factors that limit the prediction of behavior from intentions.
FINALCOMMENT
593
REFERENCES Fishbein. M.. & Ajzen, I. Misconceptions about the Fishbein model: Reflections on a study by Songer-Necks. Journul of Experimenini Socicrl Psychology. 1976. 12, 579-584. Songer-Necks, E. Situational factors affecting the weighting of predictor components in the Fishbein model. Jo~wnc~l of Experimen[ol Socirrl Psychology, 1976. 12, 56-69. (a) Songer-Necks. E. Reply to Fishbein and Ajzen. Journul of Experimetrrcrl Socio/ Psychology, 1976, 12, X35-590. (b)