ABSTRACT

In this chapter you will:

Learn how to determine empirically the best-fit line (the line that, when used as a basis for predicting one variable from another, minimizes the error sum of squares).

Learn what the phrase accounting for variance means.

Learn to distinguish measures that indicate the exact nature of the relation between two variables from measures that indicate the strength of their relation.

In this chapter we begin laying the groundwork required for most of the analytic techniques described in the remainder of this book. Many attributes of interest to behavioral scientists are measured with interval, ratio, or at least ordinal scales. In such cases, almost all specific research questions can be reduced to a single general form. The general question is, if the values of some quantifiable attribute (a dependent or criterion variable measured on a quantitative scale) vary from subject to subject, can that variability be explained, or accounted for, by research factors identified by the investigator (independent variables measured on either qualitative or quantitative scales)? In other words, does the independent variable have an effect on the dependent variable? Is the number of lies the expert detects affected by the subject's mood, for example, or by whether or not the subject received a drug? The techniques described in this chapter explain how such questions can be answered.