In conclusion, the multimethod approach is ideal for the study of developmental psychology because, as illustrated in this chapter, developmental constructs are often difficult to tap and require assessment in multiple contexts. Moreover, one single perfect method does not exist. Self-reports may be biased by social desirability and age constraints; other-report methods may not be useful for constructs that are difficult to observe; and observational methods may lack generalizability when performed in a laboratory setting. Psychophysiological methods are promising, but it is often difficult to control and interpret measurement of certain biological states. Nevertheless, each method has something unique to offer that may compensate for a weakness of another method: self-reports can tap internal states, other reports may be more objective, observational methods result in extremely rich data, and psychophysiological methods tap internal states that cannot be verbalized. Thus, when used in combination, these methods can represent the intended constructs more accurately than can any single measure and can broaden our theoretical scope.
Therefore we have chosen to end with a summary of a few practical tips for planning a multimethod study for developmental research, in addition to the tips provided throughout the chapter. First, start with a strong theoretical model and justification for the choice of certain types of methods. Second, choose measures with moderate to high convergent validity so you are relatively certain that your methods are measuring the same constructs, and use statistical techniques to examine relations among multiple measures. Third, be certain that your measures are appropriate for the age, ethnicity, and SES group you are investigating. Finally, do not try to do too much (although what is "too much" depends partly on your theory and resources). Choose only a few constructs and the design that best answers your research question in the most parsimonious way. The multimethod approach may be time consuming and expensive because of the resulting abundance of data, so planning accordingly is crucial to a successful program of research.
Chapter 2 6
MULTIMETHOD APPROACHES IN SOCIAL PSYCHOLOGY: BETWEEN- AND WITH1N-METHOD REPLICATION AND MULTI-METHOD ASSESSMENT
As social psychologists we pursue the goal of explaining social behavior by using the best methods we can muster. We usually admit the limitations of any one method. Even if a particular method seems unassailable, few social psychologists would feel confident about the validity of a finding unless it had been replicated, for example. Typically, such replication goes beyond merely repeating the same experiment. Rather, we choose a different type of method to operationalize our independent variable with the aim of conceptually replicating the effect. We might also choose alternative ways of measuring our dependent measure that match the original dependent measure as a construct but are distinct in measurement type. We might go further still and use a participant population that is distinctive to generalize the effect across variations in type of participant and culture. If we achieve a consistent, converging pattern of findings across all these variations in method, then we may hail our effect as reliable, valid, and uncontaminated by variance because of method. Even if we do not achieve consistent findings, such surprising patterns of data are a benefit, as they can actually sharpen our sense of the phenomenon at hand and generate new ideas.
The purpose of our chapter is to present a broad picture of the ways in which social psychologists use, or could use, multiple methods. Although we will largely emphasize the assessment end of research, we will do so in the context of the wider sense in which social psychologists can use multiple methods. Our conception of multimethod approaches makes rough distinctions between three basic ways that social psychologists use multiple methods, which we label (a) between-method replication, (b) within-method replication, and (c) multimethod assessment. We categorize the first two types as focused largely on the independent variable side of research. Between-method replication refers to using different methods of manipulating an independent variable (such as varying the way an independent variable is opera-tionalized), whereas within-method replication refers to using variations of the same method (such as using stimulus sampling). The third type, multi-method assessment, varies the method used to tap the dependent variable. The general aim across the three types of replication is to vary some aspect of the independent or dependent variable while preserving the conceptual meaning of the construct being manipulated or assessed. Usually, we seek consistency of results across replication strategies. We will emphasize, however, that inconsistency often goes with the territory and is a potentially useful feature of any ongoing research program (Eid & Diener, chap. 1, this volume).
Was this article helpful?