I'm analyzing a very large, but not finite, set of survey data. I've assembled a time series crossing gender (2 categories, if there was uncertainty) with age (in 6 categories) and wish to test whether the demographic representation has changed over time. Using the chi-squared test for independence, I understand that failing to reject the null hypothesis means the gender-age distribution between two data points is unrelated statistically, and I can confirm this at points where we know a substantial change was made (for example, an increase in total surveys taken). There is zero expected overlap of respondents in two different time periods (a control in the survey process itself).
My question is, since this is a repeated test (23 adjacent month-to-month pairs of distributions from 24 months of observed distributions), am I amplifying the type II error, and if so is there a better test?