Multiple paired ordinal variables
When you have a multiple ordinal variables each measured on the same scale (e.g. fully agree to fully disagree) you might be interested in if any of the variables scored different then the others.
The analysis breaks down into the following parts (for each part in this analysis a separate page was created):
Part 1: Descriptive analysis
Use descriptive statistics to get an impression of the data, using:
1) A cross table with the percentages for each score from each variable
2) A visualisation of the data with a mutiple stacked bar-chart
3) Some statistical measures for central tendency (the median) and dispersion (concensus)
Part 2: Inferential statistics
After the first impression determine what can be said about the population based on your sample data by:
1) Determine if there are any differences between any of the variables in the population (Friedman test)
2) If there are, then determine which variables differ significantly (a post-hoc Dunn, Wilcoxon or Sign test) 3) Determine the effect sizes (Kendall's W)
Part 3: Reporting
As the last step, you will need to write up all the results.
Let's begin with getting an impression of the results in the first section.
3+ Ordinal variables