Conversation

I'm trying to check for the average sex rate in a relationship, grouped by how poly they are, controlling for age. Using chatgpt, I have this code. It's giving me a reasonable-looking result, and I think I did it right, but I'd like some human eyeballs to verify this is not dum

159

Replying to

Statistician here - looks right to me. You could try allow poly and selfage to interact in the model - using "sexrate ~ poly * selfage" will get you that.

Aella

@Aella_Girl

Replying to

@emilybartha

And what does interacting in the model mean? How do I know if this is a right choice?

9:45 PM · Jan 11, 2023

5,599

Views

Replying to

Interaction means that you are allowing for the fact that age may have a different effect on sex rate for poly couples than for nonpoly. Maybe age is associated with a decrease in sex for nonpoly couples but age isn't a major factor for poly. Your model doesn't allow for that.

Replying to

and

Basically: What if older mono couples fuck less but older poly couples actually fuck more? That interaction allows you to investigate that.

Just Contemplating Life and Such

Replying to

and

Since you are doing data exploration it can be very powerful to look at plots of data points for the two interesting variables, like sexrate by selfage graphed separately by poly status. If the plotted relationship look different for poly status you have an interaction.

Just Contemplating Life and Such

Replying to

and

Visuals tell you so much more than the stats, and let you interpret the stats the model spits out.

Replying to

and

be careful when interpreting interactions though. not sure about R code, but in stata it’s like this to interpret if interactions are significant: reg Y X1 X2 margins, dydx(X1) at(X2=(values)) see this study for more: jstor.org/stable/25791835 good luck!

Replying to

and

my tip is to not listen to too many stats people. no matter how rigorous someone will say “but what about…”