[Anagaw Derseh]

In the example provided by @Enrique Pinzon the treatment variable has more than two categories while the outcome is binary variable. But I am now clear about how to interpret the coefficients for each category of the treatment variable. Normally people do not prefer to interpret estimates for covariates generated using logit/probit model. Some researchers interpret them as "log odd". So, in the result below, how can I interpret the coefficient for "6-10 daily vs 0 daily"? Is this in terms of unit, percent or percentage points? Can we also predict "marginal effects" for the treatment variable having more than two categories as in our example (i.e 0.0656168). Marginal effects are easy to interpret in percentage points by multiplying the estimates by 100%.

[Anagaw Derseh]

.