Doing Archival Research in Political Science By Scott A. Frisch, ...

Reasons Not to Do Archival Research

In a book partly aimed at encouraging archival research, it may seem counterintuitive to provide a section on the reasons to avoid it. We do not mean to discourage archival research but to promote it by addressing the common sources of hesitation that many readers may have about undertaking such work.

Reason One: This May Not Generate Publishable Findings

Archival research is time-consuming. Many hours and even days will be invested traveling to archives and collecting documents. Many more hours may be spent putting together a data set or constructing a usable interpretation of events based on the documents. What if, after all of this effort, one is confronted with a computer screen full of null findings? What if one does not find anything?

This is a fair concern. But, it is equally likely that if one invests considerable time constructing a data set of similar scope from easily available, previously published data, one faces the same prospect: having spent many hours and days constructing a data set, only to be left with a heap of equally difficult-to-publish findings. Most have heard of—or even experienced themselves––the agony of assembling a data set over a considerable length of time (perhaps years), cleaning that data set of errors, writing code for days, only to write a logit routine, press “return,” and presto! Nothing. Zippo. Nada. Not a single significant coefficient. In any case, quantitative, qualitative, statistical or not, one can still end up with research that belongs in the venerated Journal of Null Results.

So, we doubt that archival research is any more likely to yield null findings than traditional political science research built on quantitative data sets. In fact, we boldly argue that one is less likely to have null findings when using archival data (we will even italicize that statement). There are two reasons for this. First, research that uses archival data relies on data that are appropriate for testing political science theories. For instance,