Version info: Code for this page was tested in SAS 9.3
Zero-inflated Poisson regression is used to model count data that has an excess of zero counts. Further, theory suggests that the excess zeros are generated by a separate process from the count values and that the excess zeros can be modeled independently. Thus, the zip model has two parts, a Poisson count model and the logit model for predicting excess zeros. You may want to review these Data Analysis Example pages, Poisson Regression and Logit Regression.
Please Note: The purpose of this page is to show how to use various data analysis commands. It does not cover all aspects of the research process which researchers are expected to do. In particular, it does not cover data cleaning and verification, verification of assumptions, model diagnostics and potential follow-up analyses.
Examples of zero-inflated Poisson regression
Example 1. School administrators study the attendance behavior of high school juniors over one semester at two schools. Attendance is measured by number of days of absent and is predicted by gender of the student and standardized test scores in math and language arts. Many students have no absences during the semester.
Example 2. The state wildlife biologists want to model how many fish are being caught by fishermen at a state park. Visitors are asked whether or not they have a camper, how many people were in the group, were there children in the group and how many fish were caught. Some visitors do not fish, but there is no data on whether a person fished or not. Some visitors who did fish did not catch any fish so there are excess zeros in the data because of the people that did not fish.
Description of the data
Let’s pursue Example 2 from above using the dataset fish.
We have data on 250 groups that went to a park. Each group was questioned about how many fish they caught (count), how many children were in the group (child), how many people were in the group (persons), and whether or not they brought a camper to the park (camper).
In addition to predicting the number of fish caught, there is interest in predicting the existence of excess zeros, i.e., the zeroes that were not simply a result of bad luck fishing. We will use the variables child, persons, and camper in our model. Let’s look at the data.
proc means data = fish mean std min max var; var count child persons; run; The MEANS Procedure Variable Mean Std Dev Minimum Maximum Variance ---------------------------------------------------------------------------------------- count 3.2960000 11.6350281 0 149.0000000 135.3738795 child 0.6840000 0.8503153 0 3.0000000 0.7230361 persons 2.5280000 1.1127303 1.0000000 4.0000000 1.2381687 ---------------------------------------------------------------------------------------- proc univariate data = fish noprint; histogram count / midpoints = 0 to 50 by 1 vscale = count ; run; proc freq data = fish; tables camper; run; The FREQ Procedure Cumulative Cumulative camper Frequency Percent Frequency Percent ----------------------------------------------------------- 0 103 41.20 103 41.20 1 147 58.80 250 100.00
Analysis methods you might consider
Below is a list of some analysis methods you may have encountered. Some of the methods listed are quite reasonable while others have either fallen out of favor or have limitations.
- Zero-inflated Poisson Regression – The focus of this web page.
- Zero-inflated Negative Binomial Regression – Negative binomial regression does better with over dispersed data, i.e., variance much larger than the mean.
- Ordinary Count Models – Poisson or negative binomial models might be more appropriate if there are no excess zeros.
- OLS Regression – You could try to analyze these data using OLS regression. However, count data are highly non-normal and are not well estimated by OLS regression.
SAS zero-inflated Poisson regression analysis using proc genmod
If you are using SAS version 9.2 or higher, you can run a zero-inflated Poisson model using proc genmod.
proc genmod data = fish; class camper; model count = child camper /dist=zip; zeromodel persons /link = logit; run; The GENMOD Procedure Model Information Data Set WORK.FISH Written by SAS Distribution Zero Inflated Poisson Link Function Log Dependent Variable count Number of Observations Read 250 Number of Observations Used 250 Class Level Information Class Levels Values camper 2 0 1 Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF Deviance 2063.2168 Scaled Deviance 2063.2168 Pearson Chi-Square 245 1543.4597 6.2998 Scaled Pearson X2 245 1543.4597 6.2998 Log Likelihood 774.8999 Full Log Likelihood -1031.6084 AIC (smaller is better) 2073.2168 AICC (smaller is better) 2073.4627 BIC (smaller is better) 2090.8241 Algorithm converged. Analysis Of Maximum Likelihood Parameter Estimates Standard Wald 95% Confidence Wald Parameter DF Estimate Error Limits Chi-Square Pr > ChiSq Intercept 1 2.4319 0.0413 2.3510 2.5128 3472.23 ChiSq Intercept 1 1.2974 0.3739 0.5647 2.0302 12.04 0.0005 persons 1 -0.5643 0.1630 -0.8838 -0.2449 11.99 0.0005
The output begins with a summary of the model and the data. This is followed by a list of goodness of fit statistics.
The next block of output includes parameter estimates from the count portion of the model. It also includes the standard errors, Wald 95% confidence intervals, Wald Chi-square statistics, and p-values for the parameter estimates.
The last block of output corresponds to the zero-inflation portion of the model. This is a logistic model predicting the zeroes. The output includes parameter estimates for the inflation model predictors and their standard errors, Wald 95% confidence intervals, Wald Chi-square statistics, and p-values.
All of the predictors in both the count and inflation portions of the model are statistically significant. This model fits the data significantly better than the null model, i.e., the intercept-only model. To show that this is the case, we can run the null model (a model without any predictors) and compare the null model with the current model using chi-squared test on the difference of log likelihoods.
proc genmod data = fish; model count = /dist=zip; zeromodel / link = logit ; run; The GENMOD Procedure Model Information Data Set WORK.FISH Written by SAS Distribution Zero Inflated Poisson Link Function Log Dependent Variable count Number of Observations Read 250 Number of Observations Used 250 Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF Deviance 2254.0459 Scaled Deviance 2254.0459 Pearson Chi-Square 248 1918.7890 7.7371 Scaled Pearson X2 248 1918.7890 7.7371 Log Likelihood 679.4854 Full Log Likelihood -1127.0229 AIC (smaller is better) 2258.0459 AICC (smaller is better) 2258.0945 BIC (smaller is better) 2265.0888 Algorithm converged. Analysis Of Maximum Likelihood Parameter Estimates Standard Wald 95% Confidence Wald Parameter DF Estimate Error Limits Chi-Square Pr > ChiSq Intercept 1 2.0316 0.0349 1.9631 2.1000 3388.16 ChiSq Intercept 1 0.2728 0.1277 0.0225 0.5232 4.56 0.0327
The log likelihoods for the full model and null mode are -1031.6084 and -1127.0229, respectively. The chi-squared value is 2*( -1031.6084 – -1127.0229) = 190.829. Since we have three predictor variables in the full model, the degrees of freedom for the chi-squared test is 3. This yields a p-value <.0001. Thus, our overall model is statistically significant.
We may want to compare the current zero-inflated Poisson model with the plain poisson model, which can be done with the Vuong test. Currently, the Vuong test is not a standard part of proc genmod, but a macro program that performs the Vuong test is available from SAS here. Usage of the macro program requires the %include statement, in which we list the location of the macro. This macro program takes quite a few arguments, as shown below. We rerun the models to get produce these required input arguments. With the zero-inflated Poisson model, there are total of five regression parameters which includes the intercept, the regression coefficients for child and camper for the Poisson portion of the model as well as the intercept and regression coefficient for persons. The plain Poisson regression model has a total of three regression parameters. The scale parameter is the dispersion parameter from each corresponding model, and for our poisson model is fixed at 1.
%inc "C:UsersALINDocumentsMy SAS Filesvuong.sas"; proc genmod data = fish order=data; class camper; model count = child camper /dist=zip; zeromodel persons; output out=outzip pred=predzip pzero=p0; store m1; run; proc genmod data = outzip order=data; class camper; model count = child camper /dist=poi; output out=out pred=predpoi; run; %vuong(data=out, response=count, model1=zip, p1=predzip, dist1=zip, scale1=1.00, pzero1=p0, model2=poi, p2=predpoi, dist2=poi, scale2=1.00, nparm1=3, nparm2=2) The Vuong Macro Model Information Data Set out Response count Number of Observations Used 250 Model 1 zip Distribution ZIP Predicted Variable predzip Number of Parameters 3 Scale 1.00 Zero-inflation Probability p0 Log Likelihood -1031.6084 Model 2 poi Distribution POI Predicted Variable predpoi Number of Parameters 2 Scale 1.00 Log Likelihood -1358.5929 Vuong Test H0: models are equally close to the true model Ha: one of the models is closer to the true model Preferred Vuong Statistic Z Pr>|Z| Model Unadjusted 3.5814 0.0003 zip Akaike Adjusted 3.5705 0.0004 zip Schwarz Adjusted 3.5512 0.0004 zip Clarke Sign Test H0: models are equally close to the true model Ha: one of the models is closer to the true model Preferred Clarke Statistic M Pr>=|M| Model Unadjusted 13.0000 0.1137 zip Akaike Adjusted 2.0000 0.8496 zip Schwarz Adjusted 2.0000 0.8496 zip
For the Vuong test, a significant z indicates that the zero-inflated model is better. Here we see that the preferred model is a zero-inflated Poisson model over a regular Poisson model. The positive values of the z statistics for Vuong test indicate that it is the first model, the zero-inflated poisson model, which is closer to the true model.
We can use the estimate statement to help understand our model. We will compute the expected counts for the categorical variable camper while holding the continuous variable child at its mean value using the atmeans option, as well as calculate the predicted probability that an observation came from the zero-generating process. In the estimate statement, we provide values at which to evaluate each coefficient for both the Poisson model and the zero-inflation model. The sets of coefficients of the two models are separated by the @ZERO keyword.
proc genmod data = fish; class camper; model count = child camper /dist=zip; zeromodel persons /link = logit ; estimate "camper = 0" intercept 1 child .684 camper 1 0 @ZERO intercept 1 persons 2.528; estimate "camper = 1" intercept 1 child .684 camper 0 1 @ZERO intercept 1 persons 2.528; run; Contrast Estimate Results Mean Mean L'Beta Standard Label Estimate Confidence Limits Estimate Error Alpha camper = 0 2.4220 1.9724 2.9741 0.8846 0.1048 0.05 camper = 0 (Zero Inflation) 0.4677 0.3838 0.5536 -0.1292 0.1756 0.05 camper = 1 5.5768 4.8823 6.3701 1.7186 0.0679 0.05 camper = 1 (Zero Inflation) 0.4677 0.3838 0.5536 -0.1292 0.1756 0.05 Contrast Estimate Results L'Beta Chi- Label Confidence Limits Square Pr > ChiSq camper = 0 0.6792 1.0899 71.28 <.0001 camper = 0 (Zero Inflation) -0.4735 0.2150 0.54 0.4619 camper = 1 1.5856 1.8516 641.42 <.0001 camper = 1 (Zero Inflation) -0.4735 0.2150 0.54 0.4619
In the Mean Estimate column, we find predicted counts of fish from the Poisson model, ignoring the zero-inflation model, for both camper = 0 and camper = 1, as well as the predicted probability of belonging to the zero-generating process from the zero-inflation model. The zero-inflation model does not include camper as a predictor, so the probability of zero for both zero-inflation models is the same. To get the expected counts of fish from the mixture of the two models, simply multiply the expected counts from the Poisson model by the probability of getting a non-zero from the zero-inflation model (1 – p(zero)). Thus, the expected counts of fish for camper = 0 and camper = 1 including zero-inflation are 2.422*(1-0.4677) = 1.289 and 5.5768*(1-0.4677) = 2.968, respectively.
SAS zero-inflated Poisson analysis using proc countreg
Proc countreg is another option for running a zero-inflated Poisson regression in SAS (again, version 9.2 or higher). This procedure allows for a few more options specific to count outcomes than proc genmod. The proc countreg code for the original model run on this page appears below. We indicate method = qn to specify the quasi-Newton optimization process that matches the proc genmod results.
proc countreg data = fish method = qn; class camper; model count = child camper / dist= zip; zeromodel count ~ persons; run; The COUNTREG Procedure Class Level Information Class Levels Values camper 2 0 1 Model Fit Summary Dependent Variable count Number of Observations 250 Data Set MYLIB.FISH Model ZIP ZI Link Function Logistic Log Likelihood -1032 Maximum Absolute Gradient 3.69075E-7 Number of Iterations 13 Optimization Method Quasi-Newton AIC 2075 SBC 2096 Algorithm converged. Parameter Estimates Standard Approx Parameter DF Estimate Error t Value Pr > |t| Intercept 1 2.431911 0.041271 58.93 <.0001 child 1 -1.042838 0.099988 -10.43 <.0001 camper 0 1 -0.834022 0.093627 -8.91 <.0001 camper 1 0 0 . . . Inf_Intercept 1 1.297439 0.373850 3.47 0.0005 Inf_persons 1 -0.564347 0.162962 -3.46 0.0005
SAS Zero-inflated Poisson analysis using proc nlmixed
For those using a version of SAS prior to 9.2, a zero-inflated negative binomial model is doable, though significantly more difficult. Please see this code fragment: Zero-inflated Poisson and Negative Binomial Using Proc Nlmixed.
Things to consider
- Since zip has both a count model and a logit model, each of the two models should have good predictors. The two models do not necessarily need to use the same predictors.
- Problems of perfect prediction, separation or partial separation can occur in the logistic part of the zero-inflated model.
- Count data often use exposure variables to indicate the number of times the event could have happened. You can incorporate exposure into your model by using the exposure() option.
- It is not recommended that zero-inflated poisson models be applied to small samples. What constitutes a small sample does not seem to be clearly defined in the literature.
- Pseudo-R-squared values differ from OLS R-squareds, please see FAQ: What are pseudo R-squareds? for a discussion on this issue.
- Cameron, A. Colin and Trivedi, P.K. (2009) Microeconometrics using stata. College Station, TX: Stata Press.
- Long, J. Scott (1997). Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks, CA: Sage Publications.