Virata manual

poisson

Explains a count — how many times something happened.

What we have

  • the same factor notation as regress

Coming later

  • irr, to print incidence rate ratios
  • exposure() and offset()
  • predict and margins afterwards

How you write it

poisson depvar indepvars [if]

Square brackets mean optional. A vertical bar means choose one.

What it does

For outcomes that are whole numbers starting at zero: household size, hospital visits, accidents per month. Coefficients are on a multiplying scale, so exponentiating one gives the factor by which the count changes (an incidence rate ratio).

It rests on one strong assumption: that the spread of the counts equals their average. Real counts are usually more spread out than that, which is what nbreg is for.

Try it

Every example below was run after sysuse vira.sample02, clear.

. poisson hhsize age
Poisson regression                           Number of obs      =        6,000
                                             LR chi2            =         0.09
                                             Prob > chi2        =       0.7626
                                             Pseudo R2          =       0.0000

------------------------------------------------------------------------------
      hhsize | Coefficient  Std. err.      z   P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
         age |   9.752e-05  .00032286    0.30   0.763    -.0005353   .00073031
       _cons |    1.508891  .01635485   92.26   0.000     1.476836    1.540946
------------------------------------------------------------------------------
The same thing in pandas and in R

pandas

import statsmodels.formula.api as smf

model = smf.poisson("hhsize ~ age", data=df).fit()

R

model <- glm(hhsize ~ age, family = poisson(link = "log"), data = df)
summary(model)

# This R was run in a real R (4.6.1) and gave the same numbers.

Nearby

regress logit probit nbreg ologit oprobit