best test for categories over time?

I have a data set that includes 1300 patients with brain aneurysms that was collected over 25 years. We have some demographic variables and categories for aneurysm size. What is the most appropriate statistical test for:

- continuous variable (age) over time categories
- categorical variable without rank (race) over time categories
- categorical variable with rank (aneurysm size) over time categories

Really appreciate the insight. Thanks!
I want to know, for example, whether there is a change in the proportion of aneurysms in each of five size categories over five-year time intervals.

I want to know whether the gender is significantly different over five-year time intervals.

I want to know whether there is a significant difference in age over five-year time intervals.

does that make sense?


Omega Contributor
Write more about this dataset. Is it nursing home data, insurance data or hospital data. So you only have data for people who had an aneurysm?

If I only had car accident data, could I say more older people are getting into car wrecks in the last period if I don't know the underlying ages of those not getting into car wrecks. So for one period it could look like the age decrease, but could it be because that period had few older people and more younger people but the age rates were the same as historically?
It's hospital data, pretty complete. I'm not looking to do a super-sophisticated investigation of confounders. We have age, gender, race, aneurysm location, aneurysm size, etc. for all patients.

Looks like Pearson is best for two ~continuous variables like age and year, potentially could also be used for size and year?

But what about year and a categorical variable like aneurysm location (internal carotid, anterior cerebral, etc)?
Last edited:


Omega Contributor
Is data for years, or do you have their date of service?

For continuous variables you could plot value versus time period and get a correlation. You could also compare means between periods using ANOVA, but would want to correct your level of significant to minimize false discovery.

The other variables could be looked at with bar graphs and Pearson's Chi-Square test, plot time groups against binary variable.