Estimating Nonresponse Bias in a Telephone-Based Health Surveillance Survey in New York City