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

problem is to carefully review congressional staff directories, develop a list of key staff members, search the member’s and other collections for staff member papers, and—if the staff member is still living—contact the member directly in order to gain access to these papers.

Although bias can certainly be introduced into congressional collections, understanding the potential sources of those biases can help mitigate their effects. Archival collections do not provide a complete account of a member’s career or a particular political moment or process, but neither do other sources of data. Paying careful attention to what an archival collection contains and what it does not pushes an investigator to consider other sources of information that may be available to fill in the information gaps.

Coping with “Missing Data”

There is another saying in politics: “Don’t let the perfect be the enemy of the good.” Would it be “perfect” if any and every discussion and decision in a political setting was documented? Sure. Would it be perfect if every repository could keep every last shred of paper? Of course. Does any potential shortcoming in the archival record mean that these data should be disregarded? No. That would be throwing the baby out with the bath water. That would make the perfect the enemy of the good. Archival resources offer, in many cases, a qualitative improvement over many current measures of political behavior (e.g., committee assignments as a measure of committee assignment “preference” or press releases as a measure of a member’s governing style).

Researchers certainly do not hold other measures of political behavior to such a high standard. Congressional scholars routinely employ roll call votes to measure individual preferences. Of course they are imperfect. For instance, many votes are never taken because party leaders prefer to avoid votes that will cause their members to take positions on controversial issues. How would roll call data be different if these “missing data” were included? By their nature, roll call votes do not reflect a continuum of choice on the part of voters. Roll call data would be more perfect