A survey is carried out by the finance department to determine the distribution of household size in a certain city. They draw a simple random sample of 1,000 households. After several visits, the interviewers find people at home in only 653 of the sample households. Rather than face such a high non-response rate, the department draws a second batch of households, and uses the first 347 completed interviews in the second batch to bring the sample up to its planned strength of 1,000 households. The department counts 3,087 people in these 1,000 households and estimates the average household size in the city to be about 3.1 persons. Is this estimate likely too low, too high, or about right

Right now, is almost certain that the normal tenants per family unit is overestimated. Note that you can't state this without a doubt. This is just 1 example.

The explanation is: If the family has less tenants, there is a higher likelihood of the house being vacant, accepting that everybody is similarly liable to remain at home or to go out.

Thus, on the off chance that they just tally family units with individuals at home, they are probably going to pass up families with less tenants.