A Introduction

At this point you might be wondering about another prominent kind of hypo­thesis that I seem to be ignoring: the statistical one. This is not because scientific and statistical hypotheses are more or less interchangeable but because they are so different that the statistical hypothesis merits its own coverage.

I have three interrelated goals in this chapter: to distinguish between statistical and scientific hypotheses, to show how the principle of falsification coexists with probability and statistics, and, finally, to review statistical ideas that will come up again later in the book. I will focus on general concepts that few scientists were exposed to in their introductory statistics class, not on the computational details of the methods that we learned, which, as I'll show, are a mash-up of contradic­tory philosophies that have divergent implications for science.

Why do we have to get into statistics at all if we're interested in the scientific hypothesis? First, I've argued (Chapter 2) that basic science is not ultimately inter­ested in “probable truth.” To appreciate why it isn't, we need to understand the var­ious meanings of “probability,” a subj ect that recurs in several chapters and is within the purview of statistics. Second, if you think about the pervasiveness of uncer­tainty, you'll see there could be a problem in the scientific process involving falsifi­cation of scientific hypotheses. Thus far, I've implied that test outcomes are neatly cut and dried, with tests either confirming or falsifying a prediction. Science is al­most never so straightforward, however. Variability resulting from sampling error, instrument resolution, or random noise, constantly rears its ugly head. Change is constant, and no two things are the same (in the macro world; all electrons may be identical). We need statistics to help us cope with variability. Third, several fundamental statistical matters arise in the context of the Reproducibility Crisis (Chapter 7), and we'll need to be familiar with them in order to evaluate the claims of crisis. Let's start with the comparison of scientific and statistical hypotheses.

5.