The second general type of multiplicity in method involving the independent variable is replication within a specific method, usually within a particular experiment. This is an important consideration for much social psychological experimentation, although, as Wells and Windschitl (1999) showed in a recent review of this research strategy, it is underappreciated in its importance and underused. Stimulus sampling, in which multiple instances of a stimulus category are used in a particular experiment, is the prime example of this type of replication. Wells and Windschitl (1999) emphasized that stimulus sampling is especially needed "whenever individual instances in the category potentially vary from one another in ways that might be relevant to the dependent measure" (pp. 1115-1116). An example would be of using photographs to examine the effects of physical attractiveness on person perception (e.g., Alicke, Smith, & Klotz, 1986). Using only one instance to represent levels of attractiveness, given the huge variability in physical attributes potentially contributing to attractiveness, would be methodologically imprudent.
Stimulus sampling furthers external validity by enhancing the sense that any effect can be generalized across other similar stimuli. Wells and Winds-chitl (1999) also emphasized the role stimulus sampling plays in assessing construct validity. Only using a single instance of category to represent the category "can confound the unique characteristics of the selected stimulus with the category. What may be portrayed as a category effect could in fact be due to the unique characteristics of the stimulus selected to represent the category" (p. 1116).
Stimulus sampling may be a particularly important issue in social psychology because many social psychological experiments, all too often, end up involving a single individual to represent a category of people. To illustrate the problem Wells and Windschitl (1999) gave the hypothetical research example in which it is proposed that people give more personal space to males compared to females. The idea is tested by having a male or female confederate stand at a place in a mall by which many patrons must pass. Suppose the researcher finds, on average, that the male confederate is given 12 centimeters greater distance than the female confederate. This may seem to be good evidence of a "gender" effect, but, in fact, one is really comparing confederates who are very different in ways besides their gender. The researcher may appear to be manipulating gender, but a better way to capture what is happening is to label each condition with the name of each experimenter. If results show that "Stan" produces greater personal space than "Mary," then the ambiguity inherent in the manipulation is plain to see, and the need for multiple male and female confederates is obvious.
Interestingly, although most social psychologists are well aware of the need for multiple sampling of stimuli, in any one experiment the implementation of such sampling is often disregarded, even though the "use of one stimulus to represent a category can be construed as functionally equivalent to conducting an experiment with a sample size of n = 1" (Wells & Windschitl, 1999, p. 1123). Wells and Windschitl (1999) urged social psychologists to recognize the need for within-method replication, although they also acknowledged that practical considerations often preclude its extensive use. In addi tion, it is often unclear how one would best select a set of exemplars for a construct despite the best intentions. Even so, researchers benefit from appreciating the potential problems associated with neglecting within-method replication strategies, and they should use their common sense in taking steps to avoid these problems. Simply being aware of the tendency to assume incorrectly that the exemplars chosen to represent a category are in fact representative encourages a more careful selection process.
Was this article helpful?