Question

1. Each of the following situations describes error in a statistical experiment, where a researcher is trying to find the percentage of American college students who identify as Democrats or Republicans. State whether this is sampling or non-sampling error, and explain why you chose that answer.
   1. The researcher randomly selects 400 college students from around the country and asks them if they are Democrats or Republicans. The researcher finds that 38% of the students are Democrats, when the actual percentage is 41%.
   2. The researcher sets up a table in the cafeteria at their University, a wealthy private school in New England, and asks students who pass by if they identify as a Democrat or Republican. They find that 53% of the surveyed students identify as Republicans, when the actual percentage is 36%.

Answer 1

a) Sampling error happens because the sample of 400 students from whole country cannot be the population of all students in whole country. Hence, we see that the population of students in the whole country who are Democrats is 41% but in his sample he got 36% of Democrats. Naturally this can happen since he is missing out a whole lot of people from the population

b) Non sampling error is happening here. The main problem is with the sampling frame. He selected a sample from a wealthy private school in New England. The sample must be unbiased of the students financial condition as financial condition does influence the political choice to some extent.