Plotting interactions - please help with the commands

[Anton Ivanov]

Hello dear forum members,

I am interested in examining the impact of the moderator variable on a relationship b/w a number of continuous predictors and continuous DV. There is also a squared term included in the model.

The estimated model is (properly specified, OLS assumptions are met):

reg y x1 x2 x2^2 x3 x4 x5 x2x3 x2x4, vce(hc3)

What would be the proper commands to plot and examine the relationships b/w the predicted y and x3 and x4 depending on x2 (at say -1sd, mean, and +1sd)?

Both interactions included in the model have significant coefficients and also significant R^2 change.

Thank you in advance for help with this.

[Maarten Buis]

http://maartenbuis.nl/wp/inter_quadr/inter_quadr.html

[Anton Ivanov]

http://maartenbuis.nl/wp/inter_quadr/inter_quadr.html

Thank you Mr. Buis, great "tip" paper.

[daniel klein]

Maarten perfectly demonstrates how to do things from first principles. With margins and marginsplot things might be a little easier for starters, though what happens behind the doors has more of a "black-box" character.

sysuse auto ,clear

reg price c.mpg##c.weight##c.weight i.foreign

su mpg
loc mean = r(mean)
loc plus_sd = r(mean) + r(sd)
loc minus_sd = r(mean) - r(sd)

su weight
loc min = r(min)
loc max = r(max)

margins ///
,at(weight = (`min'(100)`max') ///
mpg = (`minus_sd' `mean' `plus_sd'))

marginsplot ,noci

Best
Daniel

[Anton Ivanov]

I have created a do-file following exactly the way Buis specified it in his article, but I get this error that I don't understand

reg [my varlist]
qui sum ln_likes_c if e(sample)
local m = r(mean)
local sd = r(sd)
preserve
qui replace hsa_c = 0
forvalues i = -2/2 {
qui replace ln_likes_c = `m ́ + `i ́*`sd ́
local j=`i ́+2
qui predict yhat`j ́
}
format yhat* %8.0gc
sort weight

when I run it, I get the following error. Why is m an invalid name if I specified it earlier?
`m ́ invalid name
r(198);

Please help

[Svend Juul]

I think you have the right single quote wrong. It is the simple apostrophe:

`m'

not

`m´

[Anton Ivanov]

I think you have the right single quote wrong. It is the simple apostrophe:

`m'

not

`m´

Thank you, Sir. Worked right away.