Question

Billy is being asked by his local city council to predict who will be elected mayor among candidate A and B. The electorate population of this city is composed of 80% of people living in urban areas and of 20% living in rural areas. Moreover, previous polls and past election records indicate that candidate B is much more popular than candidate A in rural neighborhoods, while the inverse trend can be seen in urban neighborhoods. Nonetheless, Billy decides to sample as many people in rural areas than in urban areas. Assuming the response rate is the same everywhere in the city, what can be expected from Billy’s prediction? How should he change his sampling scheme to minimize sampling bias

Answer 1

Assuming the response rate is the same, Billy's prediction is going to be more biased towards the voice of the rural areas, that is, candidate B is gonna be more preferred. However, in a city like such where the population proportion is very contrasting (80% urban and 20% rural), prediction is that united voice of urban people will always win.

Since Billy's task is to predict the results, he should present the true picture of the day (of casting votes). He should sample from the two populations proportionately, that is 80% of the total sample should come from urban areas and the rest 20% should come from rural areas. This is the true thing that is gonna happen on the actual day of elections. Hence this approach should be followed to eliminate the sampling bias.