Gatekeeping or Not, Sample Selection in the Roll Call Agenda Matters
Bibliographic Data
| ID | 6232525 |
|---|---|
| Authors | Jane Mcintosh Snyder (0000-0002-1386-2844, corresponding author), James M Snyder |
| Year | 1992 |
| Volume | 36 |
| Issue | 1 |
| Pages | 36 |
| Publication date | 1992-02-01 |
| Peer Reviewed | Yes |
| Open Access | No |
| Type | ARTICLE |
| Venue | American Journal of Political Science (JOURNAL) |
| Journal identifiers | ISSN: 0092-5853 • E-ISSN: 1540-5907 |
| Publisher | JSTOR (PUBLISHER) |
| DOI | 10.2307/2111423 |
| OpenAlex | W2322137797 |
| Language | EN |
| Citations received | 6 |
| References cited | 3 |
In his comment on the paper Committee Power, Structure-Induced Equilibria, and Roll Call Votes, Howard Rosenthal makes four points. The first three points have to do with the details of scaling, and while important for distinguishing between good and bad scaling techniques, they do not really affect the main conclusions one should draw from the paper. This is because the artificial unidimensionality produced by committee gatekeeping power is independent of the scaling technique used. To the extent that gatekeeping is important, it puts a unidimensional bias in the roll call data themselves; thus, any good scaling technique applied to the data should find a high degree of unidimensionality because the unidimensionality is really there. The problem is, the roll call data are too unidimensional relative to the distribution of legislators' true ideal points. Rosenthal's fourth point is more crucial, since it questions the importance of gatekeeping itself, and I shall therefore devote a bit more discussion to it. I have no major disagreements with any of the first three points. It is surely inappropriate to apply the theorist's assumption of errorless voting to actual data, just as it is inappropriate to apply any purely deterministic model in the social sciences to actual data-one always needs an error term. Also, the percentage of votes correctly classified may not be an appropriate maximand for scaling techniques, since it often leads to ugly scalings in which voting errors may be far from the cutting lines (this leads one to wonder, however, whether scholars should pay as much attention to the percentage of votes correctly classified as they do). The natural conclusion to draw from these points is clear: theorists who wish to explore the nature of roll call voting by building models that generate artificial roll call data should use a procedure such as DNOMINATE to scale the artificial data generated by the model. This conclusion cannot be taken too literally for the time being, however, since there is currently no publicly available software to run D-NOMINATE (or any similar statistical model). Rosenthal's fourth point is that the model in my essay is too stylized to be applied to the U.S. Congress. While the list of important details left out of the model is obviously long, the main intuitions developed are compelling enough that they should continue to hold in more elaborate settings. Empirically, the issue is straightforward: What fraction of the unidimensionality found in roll call voting is due to committee gatekeeping? While I do not have the answer, I can
Econometrics · Gatekeeping · Point (geometry · Political science · Politics · Power (physics · Sample (material · Scaling · Voting · Wonder · Computer Science · Electoral Systems and Political Participation · Game Theory and Voting Systems · Law · Mathematics · Psychology · Social Psychology
| Unique citing works | 6 |
|---|---|
| Citations per year | 0,18 |
| Citation span | 1992 - 2021 (30) |
| Citation velocity | historical |
| Highly cited | No |
| Citation types | Neutral: 6 |