Path analysis relies on diagrams that resemble flow charts, making this the most visual structural equation modeling technique. Path diagrams enable researchers to depict how a set of independent variables affect the values of an outcome, or dependent variable. An education researcher, for example, could employ a path diagram to illustrate the extent to which teacher experience, class sizes, neighborhood poverty, course grades and attendance rates influence student academic performance on standardized tests. A path diagram depicts the variables as rectangular boxes connected by arrows that show the direction of correlation. Subsequent statistical analysis will provide correlation scores that quantify the level of association between two variables. Correlation can be positive or negative, ranging in value from zero, meaning no correlation, to 1, indicating perfect correlation.
Factor analysis is a popular technique for analyzing survey data and is based on the idea that certain unmeasured, or latent, variables influence patterns of relationships among observed, or measured variables. Data analysts use exploratory factor analysis to extract these factors, then employ confirmatory factor analysis to test specific hypotheses. For example, a sociological researcher may test a hypothesis that latent racial bias among survey respondents influences their responses to a series of survey items measuring political, social and cultural attitudes.
Multivariate, or multiple, regression is a popular research technique in economics and other social sciences. Researchers use regression techniques to model relationships among many social and economic phenomena, such as the influence of education, experience, gender, ethnicity, geographic location and other variables on average wages for manufacturing workers. Statistical software packages such as SPSS and SAS help researchers analyze the extensive data sets used in multiple regression.
Structural equation modeling often employs multiple techniques, according to professors Randall E. Schumacker and Richard G. Lomax, authors of "A Beginner's Guide to Structural Equation Modeling." For example, some analysts diagram structural equation models using path analysis techniques that depict underlying factors as latent variables, then conduct factor or regression analyses to test the patterns of association among the variables in the models.