Different housing policies have different effects on student life. It would be irresponsible to implement anchor housing without taking a data-driven look at the experiences of other colleges. Consider a specific example. The CUL's argument asserts that anchor housing would benefit sophomores because it would allow for ``deeper connections'' to other classes. This is an empirical claim. Why not gather some data and examine the issue?
First, we would need to operationalize the notion of ``deeper connections.'' What does this mean and how can we measure it? We could ask each sophomore:
Much of survey research is devoted to figuring out the best way of eliciting correct information. There are several Williams professors (Marcus, De Veaux, Klingenberg, Sheppard, Zimmerman, et al.) with the requisite expertise as well as statistician alumni more than willing to help out.
Second, we need to see how these measures of ``deeper connections'' vary across time and space. It is a shame that the CUL, or some other body, does not design a thorough survey of undergraduate life at Williams and administer it every year. If participation in room draw were contingent upon completing the form, response rates would be close to 100 percent. If the CUL had been doing this for the last 20 years, it would be much easier to examine how life at Williams has changed and to speculate on the causes of those changes as well as the likely effects of alternate policies. Although this hasn't been done, there is no reason not to start now.
But, since we don't have access to historical data -- and since doing surveys of alumni is problematic on several levels -- there is nothing to be done but look at other schools. Luckily, other schools are also interested in these questions. They want better information on student life and would likely be willing to administer an identical survey to their students, assuming that there would be data sharing all around. The quality and quantity of sophomore-senior interaction probably varies across schools, but we need to know where and how much it varies in order to develop policies for improving it.
Third, we need to analyze the data to see if there is a connection between different housing arrangements -- either at Williams in the past or at other colleges in the present -- and the various results that the CUL seeks to achieve. We need to see some statistics. Do colleges with anchor-like housing policies have ``deeper connections'' between sophomores and seniors than colleges with policies more like the current all-campus lottery? How much deeper are those connections, and what aspects of the housing policies seem most important in creating them?
The CUL has made a series of empirical claims (about house unity, party quality, cross-class connections and so on) concerning how different student life would be under anchor housing. Are those empirical claims correct? The CUL has provided no data to back up its forecasts. If the necessary data were impossible to obtain, that would be fine. We would have to make do with intuition. But we can get the data. There is no excuse for not doing so. What is the rush?
Policy forecasts are, of course, only one half of the puzzle. Even if we could all agree that the average sophomore would go from eating meals with two seniors to four seniors over a typical two day period, we might still disagree about the desirability of the anchor proposal. Even when we agree on what the costs and benefits will be, we may weigh them differently. Yet first we need to analyze those costs and benefits. As Deming's quote reminds us, we cannot take these claims and generalizations as given until we have numbers. Data first, analysis second, policy last.