Sampling

To distribute the paper-based survey, a list of names of language department chairs was rented from the Modern Language Association. The entire list consisted of 2589 names. From the outset, we were not planning to have a truly representative survey. If we were, we would ideally have to sample from the entire universe of colleges in the US. Instead, we were mainly interested in getting an idea of the types of placement procedures used in various schools around the country. For this more limited purpose, we decided to shoot for approximately 200 total responses (either paper or electronic) as a goal. Initially assuming a 20% response rate, this meant that we would need to send questionnaires to approximately 1,000 programs in total. To conserve resources should the response rate prove higher than anticipated, we decided perform a smaller mailing first with the option of a second mailing should we not reach our goal of 200 responses. This meant that two lists of 500 names each would be needed.

Choosing the 500 names for each mailing was done through a semi-random process in which every 5th name from the MLA list was chosen starting from a randomly generated starting point. This procedure was repeated to generate the second list as well. It should be noted that the list itself is not a random sample in that non-MLA members are excluded and because MLA membership is not necessarily strictly proportional to various geographical populations or program types. We intentionally did not keep track of to which programs questionnaires would be sent because that would create a great administrative burden with no real benefit. The reader is referred to Table 2 for information on which geographical locations are represented by survey respondents.