Читать книгу Elementary Regression Modeling - Roger A. Wojtkiewicz - Страница 14

Researchers have a purpose when they estimate regression equations and that purpose is to test research hypotheses. In the social sciences, these hypotheses are usually stated in words. Statistics are then used to test the hypotheses.

Social scientists accept the general scientific idea that hypotheses are not ever proven to be true but at best can only be supported. The argument is that it is always possible that evidence might arise that would contradict a supposedly proven hypothesis. The possibility of contradictory information arising is particularly strong in survey research where research results found with one sample from a population could be contradicted by results from another sample from the same population.

A researcher can use regression analysis to test research hypotheses in several ways, and four of those ways are examined in detail in this book. Research hypotheses can be tested by estimating individual coefficients and by estimating a series of regression equations as part of control modeling, modeling interactions, or in using spline variables to model linearity.

Thus, this book has two main goals. One is to provide an intuitive explanation of regression analysis. The other is to provide thorough guidance on how to use regression analysis to address social science research hypotheses.