The influence of experimental design: The example of the Brown-Peterson paradigm.

Abstract

Abstract The results that are found in an experiment may depend upon whether the independent variable is manipulated in a between - subjects or a within - subjects fashion. This is illustrated here using the Brown - Peterson short - term temporory distractor paradigm. The traditional finding of rapid memory loss as a function of retention interval was found only when retention interval was manipulated within subjects. No effects of retention interval was found when retention interval was manipulated between subjects. Perhaps the most basic distinction among kinds of experimental design is that between within - subject (repeated - measures) and between - subject (separate - group) designs. Within - subjects designs have long been popular in experimental psychology, and they have become even more widely used in recent decades (Poulton, 1982). In some domains, this may reflect greater ecological validity for within - subjects designs insofar as people may experience a number of levels of a particular variable in everyday life (Greenwald, 1976). However, a major reason for the popularity of within - subjects designs is that they are more statistically powerful than between - subjects designs. In memory research, several variables of current interest have been shown to yield different patterns of effects in within - subjects and between - subjects experiments. The effects of stimulus generation (Begg & Snider, 1987; Slamecka & Katsaiti, 1987), imagery bizarreness (McDaniel & Einstein, 1986), levels of processing (Challis & Brodbeck, 1992; Thapar & Greene, 1994), and presentation modality (Greene, 1989) may each be modified by the nature of the experimental design. The purpose of this study is to demonstrate that even the most classic findings of memory research are not necessarily immune to the effects of design. One of the most heavily studied memory paradigms is the Brown - Peterson (or distractor) short - term memory task (for a review, see Greene, 1992). On each trial, subjects are given a short list of items and then engage in a rehearsal - preventing activity during the retention interval. At the conclusion of the retention interval, subjects are asked for recall of the items (usually in order). The classic finding (first discovered by Daniels, 1895, and later reported by Peterson & Peterson, 1959) is that recall is quite accurate at short retention intervals but drops rapidly, reaching asymptote when the retention interval is approximately 15 - 25 s. Two aspects of the procedure of Daniels (1895) and Peterson and Peterson (1959) that have been followed by most later researchers are that each subject goes through multiple trials and that the duration of the retention interval is varied randomly within subjects. A rare exception to this pattern is an experiment by Turvey, Brick, and Osborn (1970), who performed an experiment in which, for the first four trials, each subject experienced only one duration of the retention interval. Different groups underwent 5, 10, 15, 20, or 25 s of distraction on each trial. An examination of Turvey et al.'s (1970) data reveals no tendency for longer retention intervals to lead to poorer memory; in fact, there was a slight tendency in the opposite direction. Perhaps because this experiment was not carried out to study the effects of retention interval, this aspect of the data was given little attention by Turvey et al. (1970) and has been rarely discussed in subsequent studies. Turvey et al.'s (1970) data are reminiscent of Keppel and Underwood's (1962) finding that performance on the first trial of a Brown - Peterson task was unaffected by the duration of the retention interval. Keppel and Underwood's (1962) data are often dismissed as being uninterpretable due to ceiling effects on first - trial performance (see, e.g., Baddeley, 1990, p. 48; Crowder, 1989, p. 278). Turvey et al.'s (1970) results suggest that null effects of retention interval are not necessarily always due to ceiling effects. …