Parametric tests require data to be "interval" or "ratio" and normally distributed. If you apply them to ordinal scales: Outliers can heavily skew the mean.
| Research Question | Parametric Test (Avoid) | Nonparametric Test (Use) | SPSS Menu Path | | :--- | :--- | :--- | :--- | | Differences between 2 independent groups | Independent t-test | | Analyze > Nonparametric > 2 Independent Samples | | Differences between 2 paired groups | Paired t-test | Wilcoxon Signed-Rank | Analyze > Nonparametric > 2 Related Samples | | Differences among 3+ independent groups | One-Way ANOVA | Kruskal-Wallis H | Analyze > Nonparametric > K Independent Samples | Parametric tests require data to be "interval" or
In social science research, medicine, and psychology, data does not always conform to the rigid requirements of interval or ratio scales. Frequently, researchers deal with —variables where the order of values is significant, but the distance between them is unknown. Common examples include Likert scales (e.g., "Strongly Disagree" to "Strongly Agree"), pain scales, and military ranks. If the p-value is significant (e
The output will provide the test statistic, degrees of freedom, and p-value. If the p-value is significant (e.g., < 0.05), we can conclude that there are significant differences in satisfaction ratings among the three products. If the p-value is significant (e.g.
Below are the three most common scenarios involving ordinal data and their execution in SPSS.
Would you like an example dataset with step-by-step screenshots for any specific test?
Nonparametric tests are "distribution-free." They transform raw scores into ranks, making them robust against outliers and non-normal distributions. Core Nonparametric Tests and Their SPSS Applications 1. Comparing Two Independent Groups: Mann-Whitney U Test