# How predictable: A few quick and dirty pointers about regression analysis

Quantitative Results
Statistical Analysis

Regression analysis is not difficult. If you repeat it enough times you will believe it, and believing it will make it much less daunting. Right? Well, if that did not minimize your fear related to regression analyses, hopefully these quick and dirty pointers will help you out!

Why should I conduct a regression? Why not stick with my plain old Pearson correlations?

Pearson correlations are easy to conduct and interpret, making them a preferred analysis to conduct among many researchers. However, sometimes the Pearson correlation does not give you the depth of information that you need. Pearson correlations allow you to assess how two variables change together. You can use correlation analyses to determine if an association exists between two variables, as well as the strength and direction of the association. However, in a correlation analysis you are not analyzing how much a set of variables contributes to the prediction of another variable. For example, suppose we think that socio-economic status, community resources, and educational level predict the academic achievement of students. While a Pearson correlation analysis will allow us to analyze if academic achievement varies in conjunction with these variables individually, it does not allow us to determine if these variables together contribute to the variation in academic achievement. Enter: regression analysis.

What type of regression?

What does the regression tell me?

When we conduct the regression we want to determine (1) if the model that contains your independent variables explains a statistically significant amount of the variation in the dependent variable, and (2) how much each individual predictor contributes to the variation of the dependent variable. If (1) is true and the model does explain a statistically significant amount of the variation in the dependent variable, you will report the p and F values (the statistics that indicate if we have statistical significance), and the R2 value (the statistic that indicates the percentage of variation in the dependent variable explained by the model). Additionally, if (1) is true and the model is statistically significant, we will then need to make a statement regarding (2) how much each individual predictor contributes to the variation of the dependent variable. For each independent variable, we will report the p and t values (the statistics that indicate if we have statistical significance) and the unstandardized beta (B; indicates how many units the dependent variable changes with every one unit change in the independent variable).

This is just an overview and does not comprise an in-depth explanation of regression analyses. For additional information on regressions and other analyses, please visit our directory of statistical analyses or call to consult with one of our specialists!

Get Your Dissertation Approved

We work with graduate students every day and know what it takes to get your research approved.

• Address committee feedback
• Roadmap to completion
• Understand your needs and timeframe