Psychology Will Be a Much Better Science When We Change the Way We Analyze Data

because I believed that within it dwelt some of the most fundamental and challenging problems of the extant sciences. Who could not be intrigued, for example, by the relation between consciousness and behavior, or the rules guiding interactions in social situations, or the processes that underlie development from infancy to maturity? Today, in 1996, my fascination with these problems is undiminished. But Ive developed a certain angst over the intervening thirtysomething years—a constant, nagging feeling that our field spends a lot of time spinning its wheels without really making all that much progress. This problem shows up in obvious ways—for instance, in the regularity with which findings seem not to replicate. It also shows up in subtler ways—for instance, one doesnt often hear Psychologists saying, Well this problem is solved now; lets move on to the next one (as, for example, Johannes Kepler must have said over three centuries ago, after he had cracked the problem of describing planetary motion). Ive come to believe that at least part of this problem revolves around our tools—particularly the tools that we use in the critical domains of data analysis and data interpretation. What we do, I sometimes feel, is akin to trying to build a violin using a stone mallet and a chain-saw. The tool-to-task fit is not all that good, and as a result, we wind up building a lot of poor-quality violins. My purpose here is to elaborate on these issues. In what follows, I will summarize our major data-analysis and data-interpretation tools, and describe what I believe to be amiss with them. I will then offer some suggestions for change.