Question

What is the specific contribution of the Zero-Inflated models in the history of modeling count outcomes?

Answer 1

Answer -

Zero inflated model assumes that some zeros are created by Poisson process but others were not possible or eligible to event occur.

Because of this there are 2 processes.

1.At work when one determines that if the individuals is capable for a non zero response.

2.That determines the count of that response for eligible individual.

The important part is either process can result in a zero count., since we can't tell which zeros were available for a non zero count ,we can't tell which zeros were result of which process.

This zero inflation model also have two separate regression models.

1.Logistic models that models the probability of being elegible for a non zero count.

2.The other includes the size of that count.

Both models use same predictor variables but separately calculate their coefficients.So there will be vast different effects on the two processes.

Zero inflated models also assumes that some zeros are observed as a result of some specific structure in data.