Hi,
this is likely very simple, but I cannot wrap my head around it. I am running an xtpoisson model with two continuous predictors that are both standardized. How do I interpret the margins?
I constructed a little demo below. The dependent variables is a count (obviously), "number of weeks worked" in the demo data. Nevermind that one would likely not standardize any of the independent variables in this dataset, but in my real data I think it would make sense.
This yields the following result:
How do I interpret the margins in a way that would allow me to interpret them in a "....and this means x more weeks worked" way?
Thank you so much!
Michael
this is likely very simple, but I cannot wrap my head around it. I am running an xtpoisson model with two continuous predictors that are both standardized. How do I interpret the margins?
I constructed a little demo below. The dependent variables is a count (obviously), "number of weeks worked" in the demo data. Nevermind that one would likely not standardize any of the independent variables in this dataset, but in my real data I think it would make sense.
Code:
webuse nlswork.dta, clear xtset idcode year * run regression once to determine sample xtpoisson wks_work c.birth_yr##c.ttl_exp keep if e(sample) * standardize independent variables over full sample egen z_birth_yr = std(birth_yr) egen z_ttl_exp = std(ttl_exp) drop birth_yr ttl_exp rename z_birth_yr birth_yr rename z_ttl_exp ttl_exp * run regression of interest xtpoisson wks_work c.birth_yr##c.ttl_exp * obtain margins margins, at(birth_yr = (-2/1) ttl_exp = (-1 4))
Code:
(omitted)
. * obtain margins
. margins, at(birth_yr = (-2/1) ttl_exp = (-1 4))
Adjusted predictions Number of obs = 27,831
Model VCE: OIM
Expression: Linear prediction, predict()
1._at: birth_yr = -2
ttl_exp = -1
2._at: birth_yr = -2
ttl_exp = 4
3._at: birth_yr = -1
ttl_exp = -1
4._at: birth_yr = -1
ttl_exp = 4
5._at: birth_yr = 0
ttl_exp = -1
6._at: birth_yr = 0
ttl_exp = 4
7._at: birth_yr = 1
ttl_exp = -1
8._at: birth_yr = 1
ttl_exp = 4
------------------------------------------------------------------------------
| Delta-method
| Margin std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
_at |
1 | 3.645507 .0122726 297.04 0.000 3.621453 3.66956
2 | 4.865549 .0147649 329.53 0.000 4.83661 4.894488
3 | 3.630847 .0078433 462.92 0.000 3.615475 3.64622
4 | 4.972935 .0093784 530.25 0.000 4.954554 4.991317
5 | 3.616188 .005467 661.45 0.000 3.605473 3.626903
6 | 5.080322 .0066949 758.83 0.000 5.0672 5.093443
7 | 3.601529 .0074654 482.43 0.000 3.586897 3.616161
8 | 5.187708 .009413 551.12 0.000 5.169259 5.206157
------------------------------------------------------------------------------
Thank you so much!
Michael
