trend over years for categorical variable

[Reema lutfi]

Hi I am doing repeat cross sectional analysis over 10 years (discharge_year) for of a categorical measure (oc_mix) which have 3 categories(weak, strong, both). I started doing just a descriptive analysis and plotted stacked bar, and calculated the numbers and proportion for each of the category each year. I want now to test if the upward, downward or fluctuating trend seen in each category is statistically significant or not. not sure which test should I use. is it chi square for trend or ologit or some thing else and how to use it in STATA.

discharge year oc_mix
strong week both
2010 1,390 500 200
2011 1450 600 300
2012 1500 679 400
2013 1600 800 100

[Will Fleischman]

If you're looking just for significance of a trend (e.g. year to year), you're probably looking for this:
https://www.statalist.org/forums/for...tata-tutorials

If you're looking for significance in time series data that has an 'intervention' at a specific time, check out the ITSA command.
https://www.stata-journal.com/articl...article=st0389

[Rich Goldstein]

take a look at the following FAQ: https://www.stata.com/support/faqs/s...end/index.html

[Reema lutfi]

Thanks