Statistical Analysis and SPSS Guide

This set of vocabulary flashcards covers statistical test selection, normality criteria, and reporting requirements based on variable types and distribution characteristics.

Independent variable

Also known as the predictor variable.

Dependent variable

Also known as the response variable.

Chi-square test

Used when the independent variable is Categorical and the dependent variable is Numeric count (raw number/frequency of occurrence). Requires reporting the $X^2$-statistic, $df$, and $p$-value (asymp. sig.).

Linear regression

Used when both independent and dependent variables are Numeric. Requires reporting $R^2$, $F$-statistic [regression $df$, residual $df$], and $p$-value.

Shapiro-Wilk test

A test used to determine if data are normally distributed. Data are normal if the $p$-value (sig.) > 0.05; data are not normal if p < 0.05.

t-test

A means test used for 2 categorical groups when the data are normally distributed.

Levene’s test

A test for equality of variances used during a t-test. A p > 0.05 indicates variances are equal.

Equal Variances t-test reporting

If Levene’s test p > 0.05, report the $t$-statistic, $df$, and 2-sided $p$-value from the top row of the SPSS output table.

Unequal Variances t-test reporting

If Levene’s test p < 0.05, report the $t$-statistic, $df$, and 2-sided $p$-value from the bottom row of the SPSS output table.

Anova

A means test for 3+ categorical groups with normally distributed data. It assumes equal variances and requires reporting the $F$-statistic [between group $df$, within group $df$] and $p$-value.

Tukey posthoc pairwise comparisons

Additional comparisons run specifically after performing an Anova for 3+ groups.

Mann-Whitney U test

A non-parametric means test for 2 groups used when data are not normally distributed (p < 0.05 for Shapiro-Wilk). Requires reporting the $U$-statistic and 2-sided $p$-value (asymp. sig.).

Kruskal-Wallis

A non-parametric means test for 3+ groups used when data are not normally distributed (p < 0.05 for Shapiro-Wilk). Requires reporting the test-statistic, $df$, and $p$-value (asymptotic sig.) along with posthoc pairwise comparisons.

Numeric count

The raw number or frequency of occurrence, used as the dependent variable in a Chi-square test.

Asymp. sig.

An abbreviation for asymptotic significance, reported as the $p$-value in tests like Chi-square, Mann-Whitney U, and Kruskal-Wallis.