Hi all,

I am hoping someone can clear something up for me.

I want to run a poisson regression model - with a continuous dependent variable and with sex and age as factor variables.

glm cpssdx ib0.female##ib5.age_factor, fam(poisson) link(log) nolog vce(robust) eform baselevels

I want to report the interaction term but can't determine whether this would be from either of the following codes:

contrast female@age_factor

Contrasts of marginal linear predictions

Margins : asbalanced

-----------------------------------------------------
| df chi2 P>chi2
------------------+----------------------------------
female@age_factor |
13 | 1 1.05 0.3066
14 | 1 0.43 0.5096
15 | 1 0.55 0.4572
16 | 1 1.39 0.2376
17 | 1 10.57 0.0011
Joint | 5 14.00 0.0156
-----------------------------------------------------

(i.e [F(5,707) = 14.00, p = 0.0156) - so a significant interaction.

OR

margins ib0.female#ib5.age_factor

Adjusted predictions Number of obs = 717

Expression : Linear prediction, predict()

-----------------------------------------------------------------------------------
| Delta-method
| Margin Std. Err. t P>|t| [95% Conf. Interval]
------------------+----------------------------------------------------------------
female#age_factor |
MALE#13 | .1111111 .1042772 1.07 0.287 -.0936188 .3158411
MALE#14 | .2272727 .0666959 3.41 0.001 .096327 .3582184
MALE#15 | .2714286 .0528782 5.13 0.000 .1676115 .3752456
MALE#16 | .25 .0638565 3.92 0.000 .124629 .375371
MALE#17 | .1190476 .0482709 2.47 0.014 .0242761 .2138192
FEMALE#13 | .2325581 .067467 3.45 0.001 .1000986 .3650177
FEMALE#14 | .2803738 .0427694 6.56 0.000 .1964035 .3643441
FEMALE#15 | .326087 .0461245 7.07 0.000 .2355296 .4166443
FEMALE#16 | .35 .044241 7.91 0.000 .2631404 .4368596
FEMALE#17 | .3423423 .0419917 8.15 0.000 .2598989 .4247858
-----------------------------------------------------------------------------------

I specifically want to look at sex differences at each age - but there appears to be numerous ways to go about this.

Anyone have any thoughts on this?

Thanks!