If you are trying to check the relationship between two categorical variables (and not scatter plot), you can create 'Highlight Tables'. You need a measure to compare the relationship, such as, count of records, etc.
You can create numerical equivalents from the categorical variables (male = 1, female = 2), etc. and plot them to see the relationship.
Not a statistician, but my 2 cents..
Hope this helps..
I may be hijacking this thread a bit but I have a similar question in producing correlation comparisons between search terms based on a time series for the count of each individually search query.
I have read Jonathan Drummey and Bora Beran posts to get very close. The issue I appear to be having is the re-factoring into a time series. I used a string interpretation of the date field for the variable dimension but since I have to filter by index to get one value per cell, it is using the first date.
In reality my first objective is to find the best matching search trend to bacon. Then my final objective is to phase shift bacon forward in time to once again find the best matching trend as a predictor item to bacon. I have attached a packaged workbook (see sheet "Correlation Matrix" with a subset of data to work with easier.
Doc-SearchQuery-Analysis.twbx 66.3 KB