**Observational cross-sectional epidemiological study**that basically has two parts:

**Sample:**all patients diagnosed with one disease during a period of time (from 1 health care center). All data will be cross sectional from one moment.

**Descriptive part:**aims to describe various epidemiological characteristics of the sample. Ex: categorical variables such as gender, presence/absence of some things, and other categorical variables that assume more than 2 results, for example staging (1,2,3 or 4)). It also considers two continuous variables such as age for example.

**Analytical part:**aims to describe some of the above variables related to another one (for example gender distribution on each stage 1, 2 3 and 4). Then I will use chi square tests (one for each variable pair) to check if there is a statistically significant association between those variables. I expect to meet the chi square requirements (no cells with value=0, 80% or more of the cells with value>5, sample>40)

Now, theoretically I should be able to calculate sample sizes needed for all variables of the descriptive part and for all chi square tests of the analytical part, right? Then I would choose the largest sample calculated to use as reference for the whole study. (Because its a relatively not frequent disease, If the sample that I can recruit is below the calculated sample, I will calculate power

*post hoc*based on the actual study sample used)

**My question is:**

Is there a way or is it usually necessary to calculate the sample for the descriptive part? If yes, how?

How to calculate the sample size for the chi square association tests?

Is there a way or is it usually necessary to calculate the sample for the descriptive part? If yes, how?

How to calculate the sample size for the chi square association tests?

I’ve checked many websites on the internet with different calculators that require different data and now I’m confused. Particularly for the chi square I found this one ( https://www.masc.org.au/stats/PowerCalculator/PowerChiSquare) that requires only the effect size that I don’t have (does it work for me? can I estimate the effect size?). I also found this website that I find very good https://www.openepi.com/Menu/OE_Menu.htm, but I'm not sure which calculator to use.

Any help would be greatly appreciated

Thanks!