When social psychologists think of multiple methods, we are perhaps just as likely to focus on the independent variable as the dependent variable. There is clear virtue even in an exact or strict replication of a finding (e.g., Hendrick, 1990), but social psychologists prefer a conceptual replication, using a different operationalization of the independent variable (e.g., Gerard & Mathewson, 1966; Pratka-nis, Greenwald, Leippe, & Baumgardner, 1988). If alternative manipulations replicate the effect, then we have greater confidence that it is the supposed conceptual variable producing the effect.
Conceptual replication is valuable because manipulations are usually translations of an idea and thus are indirect representations of a manipulated variable. There are any number of ways of operationalizing an independent variable, and the more complex the manipulation the more one or more aspects of the manipulations may introduce an unwanted confound. Manipulations in social psychology are often more likely to introduce this sort of problem because in efforts to achieve sufficient realism and impact (Aronson, Ellsworth, Carlsmith, & Gonzales, 1990), some aspect of the manipulation can create unforeseeable effects, making interpretation unclear or baffling.
Researchers have various ways of categorizing methods of manipulating independent variables, but one broad distinction that is especially relevant for social psychologists, and that often translates into using different operationalizations, is between-field studies and laboratory studies. Social psychologists are most likely to use lab experiments to test their theories largely because of the high degree of control that lab experiment can more easily provide and the yield in terms of clear causal inferences. However, for all the virtues of the highly controlled lab experiment, there are also well-known ethical and practical drawbacks, especially for studying certain phenomena (e.g., reactions to terrorist attacks or natural disasters), that often make lab experiments unfeasible and make field studies desirable.
Field studies are useful even if ethical and practical challenges are absent in the laboratory. For one thing, much social psychological phenomena can be altered by participants' knowledge of their being in an experiment, despite the cleverest cover story. Field experiments in which participants are unaware of their participation eliminate this problem (although they introduce their own ethical problems in terms of lack of informed consent). Also, from a multiple method point of view, field and lab studies are far from unequal partners in the research process. A field study may add support to a hypothesized relationship through conceptual replication and produce unexpected associations that then receive further attention. Testing an idea in a field setting, once it has already been isolated in a lab setting, not only addresses issues of generaliz-ability, but may also suggest moderating variables.
We should also note that experimentation itself need not be seen as the be-all and end-all as general research strategy. A weakness of experimentation, especially in laboratory settings, is that it can fail to capture the complex interaction of variables that are sometimes the major determinants of behavior. Indeed, many social psychological research traditions, such as those that focus on cross-cultural issues (e.g., Diener & Suh, 2000), use correlational methods more typically. Correlational research strategies seem to address the inherent complexities of social interactions in a more satisfying way than experimental approaches and represent a varied set of effective, alternative methodological approaches to tackling social psychological phenomena.
A commonly cited example (Aronson et al., 1990) of the interplay between the laboratory and the field in social psychology is the research done on the effects of mood and helping. Laboratory experiments using mood induction techniques ranging from reading positive or negative texts (e.g., Aderman, 1972) to remembering happy or sad experiences (e.g., Moore, Underwood, & Rosenhan, 1973) have found fairly consistent increased helping effect for positive mood. However, these studies suffer from potential experimenter demand problems, and as well as the sense the helping settings used are unnatural. But, the basic finding has been replicated in field settings (e.g., Isen & Levin, 1972; Underwood, Froming, & Moore, 1977). In a now-classic series of studies, Isen and Levin (1972) found that undergraduates who received cookies while studying in a library were more likely to volunteer in response to a student's request, and adults who found a dime left in a public telephone were more likely to pick up papers that were dropped in front of them. This research actually provided two conceptual replications, at both the independent (between-method replication) and dependent variable level (multimethod assessment). As Aronson et al. (1990) pointed out, the "conver gence of results across methods and across settings greatly enhances our confidence in both sets of findings" (p. 181). Field studies seem especially valuable in assessing the generality of an effect and in uncovering "new variables that must be brought under control if the research is to have widespread applicability" (p. 181).
Although there is considerable precedent for between-method replication using field and lab studies, most programs of social psychological research involve multiple laboratory methods. As an example of between-method replication in laboratory settings, commentaries on social psychological methods (e.g., Aronson et al., 1990; Hendrick, 1990) often cite Gerard and Mathewson's (1966) conceptual replication of the classic dissonance study by Aronson and Mills (1959) showing the effects of the severity of initiation on liking for a group. In the original study, Aronson and Mills operationalized severity by varying whether female participants had to recite obscene words to gain group membership. As compelling as this study was, one could easily generate alternative explanations for the finding and wonder about its general-izability. The Gerard and Mathewson study used shock instead of the reciting of obscene words and found the same core finding, therefore enhancing the generality and reliability of the effect and as well as the dissonance interpretation.
Hendrick (1990), in a thorough analysis of the virtues of different types of replications, offers a few suggestions for how to go about doing replications in a systematic manner. First, clarify what procedural aspects of the original study, together with other features of the experimental context (e.g., participant characteristics, mode of participant recruitment, historical and cultural context, physical setting, experimenter attributes, formatting of materials), would produce a "strict" replication if duplicated. Second, alter the original procedural details such that some variations should produce results similar to the original study and that other variations should produce results different from the original study (all decisions guided by the conceptual similarity or dissimilarity of the variations to the original manipulation). Third, resources permitting, further replicate the original finding by vary ing contextual variables of such as participant type, cultural setting, mode of participant recruitment, and so on.
Social psychologists actually use systematic replication much less than its scientific yield would call for (Aronson et al., 1990; Hendrick, 1990). Apart from the often forbidding additional resources required for doing systematic replication, the discipline of psychology provides little reward for the effort in terms of journal space and respect (Hendrick, 1990). Nonetheless, there seems no question that systematic replication, or some approximation of this approach, has great potential to establish the reliability, construct validity, and generalizability of an effect (also see Rosenthal, 1990). What is more, as researchers probe the range of the effect, new hypotheses often emerge, and new research directions beckon. Thus, social psychologists should embrace more consistently and vigorously this facet of multi-method research.
Was this article helpful?