The residential division of Jojo Inc. high speed Internet service uses one advertising agency while its commercial division uses another. Two analysts, A and B are asked to test evaluate effectiveness of the two agencies. Howard proposes an H/C test that compares the click-through rates per ad for the two agencies. B proposes a difference-in-difference test in which the budgets for both agencies are increased by 50% and the percent change in click-through rates are compared.

What might be the sources of selection bias for the two proposals?

Which is likely to be smaller?