Calculate the mean of the data values for the dependent variable in the experimental groups and the control group if there is one. For example, if you are examining the influence of caffeine on test performance, calculate the mean test scores of students in the caffeine groups and no-caffeine group.
Choose a statistical test based on the number and type of independent and dependent variables and the type of study. You need to use a specific statistical test for single versus multiple variables, nominal versus scale variables and experimental versus correlation designs. For example, if you have one nominal independent variable with three levels and one scale dependent variable for an experimental design, the proper statistical test is a one-way ANOVA.
Calculate the statistical value for the test by hand, using a calculator or with the aid of a computer program such as SPSS. Many statistical tests are quite sophisticated; calculating the value by hand may require dozens of computations and take an hour or more. The one-way ANOVA is a test that compares within-groups variance and between-groups variance. The resulting F-value is higher the more the between-groups variance outweighs the within-groups variance.
Look up the corresponding p-value for the statistical test value on a chart or find the p-value on the output table on SPSS. Higher F-values (or t-values or chi-square values) mean the variance between groups is more significant, which in turn means the experimental manipulation was more successful. For basic statistical tests like t-tests and chi-square tests, you can find the p-value on a chart or set of charts. More complex tests like the ANOVA require a program such as SPSS to find the p-value.