##### "The purpose of computation is insight, not numbers."

#### Richard Hamming,

discoverer of the first useful automatic error-correction system for electronic communication

# SPSS/PASW Guides

Following are links to step-by-step guides I've written to help you use SPSS, a.k.a. PASW. These guides are not fancy at all, and none of them have any graphics. They have been written for my FDMath 221, FDMath 222, FDMath 223, and Math 224 students. The instructions given below seem to work fine for versions 14 through 19. (Exception: The graphs menu in older versions doesn't have a "Legacy Dialogs" entry.) If you find any that don't, please email me.

If what you really need is a quick guide to statistical things, try the Statistics reference pages

## Basics

(Old and clunky, but still possibly useful)

## Opening and importing data files

## Creating your own data files

## Entering two-way tables

## Weighting the cases

(Coming someday! Meanwhile, the "Frequency tables" documents---found below---address this, as dos the "Two-Way Tables" document---also below---and the "Entering two-way tables" document, above.)

## Probability

Note: Many students find the applet doesn't work on their computer. There is a quick and painless fix for this at the Help Desk. Getting this applet running is worth a try, because it's very visual and very pleasant to use, especially when compared to SPSS.

## Statistics for univariate data

## Creating graphs with SPSS/PASW

## Frequency tables and graphs for nominal data

## Frequency tables and graphs for ordinal data

## Frequency tables and histograms for "scale" (i.e., interval or ratio) data

## Descriptive statistics for univariate numeric data (the "Explore" tool)

## The Q-Q plot: Assessing normality with SPSS/PASW

## Confidence intervals for the mean

## Testing normality

## One-sample t-tests

## Matched pairs t-test

## Statistics for bivariate data

## Creating graphs with SPSS/PASW

## Nominal vs. nominal data

## Two-Way Tables

## Chi-squared tests of homogeneity

(Several proportions, or one-way tables)

## Chi-squared tests of independence or association

(Contignency tables, or two-way tables)

## Numeric vs. nominal data

## Two-sample t-tests

## One-way ANOVA

## One-way ANOVA with post-hoc tests

(e.g., Tukey or Bonferroni)

## Numeric vs. numeric data

## Association and correlation

## Linear regression and the coefficient of determination

Note: I have yet to find anything like a consensus on what to do when one of the variables is ordinal, so I haven't posted anything about that here, nor do I plan to, anytime soon. But you can always treat your ordinal data as though they were nominal.