T-TEST and ANOVA--Test whether the difference(s) between the means of two (T-TEST) or more samples (ANOVA) could be due to random sampling error alone.  The T-TEST option and the ANOVA option are almost identical in MicroCase.  Use the T-TEST option when you are comparing two samples (e.g., men versus women) or when your independent variable is dichotomized.  Use ANOVA when you are comparing more than two samples (e.g., Anthropology majors versus Economics majors versus History majors versus Political Science majors versus Sociology majors) or when your independent variable has more than two attributes.

1.  When you choose either the T-TEST or the ANOVA option, the first window asks you to indicate your dependent variable (remember it should be intervally measured) and your independent variable.  You may also indicate subset variables or control variables.  When you have entered the variable names or numbers, click on OK in the upper right hand corner of the window.

2.  The first screen in the analysis provides a ABox-and-Whisker@ Diagram which graphically represents the distribution of dependent variable scores within each sample/attribute of the independent variable.  The boxes demarcate one standard deviation above and one standard deviation below the mean.  The horizontal plot lines show the relationship between the sample means as well as the overall mean.

3.  Note the options in the gray menu on the left side of the screen.
A.  MEANS gives you the sample means and standard deviations, and if you are using the T-TEST option, tells you the probability that the difference between two means could be due to random sampling error alone.
B.  ANOVA@ gives you the ANOVA summary table and probability values to assess statistical significance.  In addition, the ANOVA screen provides Eta-squared, an asymmetrical measure of association for non-linear relationships between one categorical independent variable and one intervally (or ordinally) measured dependent variable.  Eta-squared can be interpreted as one would interpret r-squared.  For information about interpreting these correlation coefficients, consult the MicroCase correlation page.
C.  With more than two samples, the ANOVA screen also reports an R-squared test for linearity.  Although the symbols are the same, this R-squared should not be confused with the multiple correlation coefficient (for a discussion of the multiple correlation coefficient, consult the MicroCase regression page).  The R-squared in the test of linearity on the ANOVA screen tells you the amount of variation in the dependent variable can be explained by the linear effect of the independent variable.  If the line which connects the means in the box and whisker diagram is approximately straight, R-squared will be close to 1.0.  If R-squared in close to zero, you can assume that the linear imapct independent variable does not account for much of the variation in the dependent variable.  If Eta-squared is also close to zero, the interpretation would be that the independent variable has neither a linear nor a non-linear contribution to the variation on the dependent variable.  When Eta-squared is closer to 1.0 than R-squared, you can conclude that there the independent variable has some non-linear effect on the dependent variable.  The F value (and its corresponding probability) on the same line as the R-squared indicates whether Eta-squared and R-squared are significantly different.  If the F value is statistically significant, it indicates that there is a significant non-linear relationship between the independent and dependent variable.
D.  Caution: There are two F values and corresponding probability values on the ANOVA screen, and it is very important that you report the one that is appropriate for your research question.  The F value in the Between row of the ANOVA summary table tests whether the differences between the means of the sample groups could be due to sampling error; the probability value by which you assess the significance of this between groups F is found in the ANOVA summary table between the Total row and the Eta-squared row.  The F value on the R-squared row tests whether Eta-squared and R-squared are significantly different, and thus whether the relationship between the independent and dependent variables is best described as linear or non-linear; the probability value for this F value is on the same row.