This chapter reviews the statistical methods used to explore causal (directional) and noncausal research questions. Researchers have often regarded some statistical procedures as experimental and other statistical procedures as nonexperimental (e.g., correlation, structural equation modeling). The chapter discusses practical issues that influence the implementation of the analysis and design features of the study relevant to that statistical procedure. The most widely used noncausal analysis in social-personality psychology is exploratory factor analysis (EFA). There are two major issues that should be taken into account when designing studies to be analyzed using EFA: selection of measured variables and selection of sample. The chapter overviews the major types of causal hypotheses. It explains the conditions necessary for establishing causal relations and comments on study design features and statistical procedures that assist in establishing these conditions. The chapter also reviews the statistical procedures used to test different types of causal hypotheses.

This chapter focuses on what has become known as the psychometric or nonrepresentational approach to measurement. It considers the issues traditionally discussed under the heading of reliability, and reviews several still persistent definitions or types of reliability coefficients. The chapter discusses the problems and misuses of coefficient alpha, the most commonly used psychometric index in social-personality psychology. It suggests the generalizability theory as a broader and more heuristic perspective. The chapter examines issues related to construct validation. It also considers the construct validation as the crucial issue in psychological measurement and includes a broad range of validity evidence, focusing on convergent and discriminant aspects. The chapter describes model testing in construct validation and scale construction. It also focuses on the measurement models in structural equation modeling. The chapter reviews three classical strategies (external criterion, rational-intuitive, and internal-factor analytic) and also suggests an integrated model adopting the construct-oriented approach.