Hi. As far as the sample size is concerned, if you know the size of the universe (how many potential respondents) you could opt for the sample determination formula developed by Krejcie and Morgan

Article Determining Sample Size for Research Activities

To make it even simpler, there is a table that has been derived using this formula that can help you estimate your sample size based on the size of the universe.

https://www.kenpro.org/sample-size-determination-using-krejcie-and-morgan-table/

With regard to securing statistical significance and the ideal sample size, the simple answer is that there is no ideal sample size and it depends on the size of the universe. Generally, a sample of 30 is considered to be the minimum sample size required for preforming any kind of inferential statistics that can be used to reach a significant conclusion.

I have a video on choosing the right statistical test based on various factors. That might be helpful. https://www.youtube.com/watch?v=F4AKZ9-u_ic&ab_channel=Dr.S.RamaGokulaKrishnan

All the best.

Yeah, you to figure out what it is you're trying to compare between the two groups...

It's possible you want to use analysis of covariance (ancova). Imagine this plot ( https://dzchilds.github.io/stats-for-bio/stats-for-bio_files/figure-html/unnamed-chunk-282-1.png ). Where, Location would be your two generation groups, and the x and y axes would be your two scale variables. (I don't know which would be x and which would be y). From there, what is it that would be meaningful ? A difference in slopes between the two groups ?

You could also look at the correlations in each group. From there, you can compare the correlation coefficients, r. Does that have meaning ? If your two variables are more correlated in one group than the other ?

Also, importantly, what kind of variables are porn use and relationship satisfaction ? Are they scales based on several Likert items ? Are they each a single Likert-type question ?

Considering my first comment, are you interested in just comparing two correlations, or something more like the ancova ?