In a survey using response-based sampling, the probability of selecting an individual $i$ is proportional to the probability that the individual responds. If $R_i$ is the indicator variable for response and $\pi_i$ is the inclusion probability, how does this affect the estimator $\hat{Y} = \sum_{i=1}^n \frac{y_i}{\pi_i \cdot P(R_i=1)}$?