Political science graduate students Joshua Kalla and David Broockman, in collaboration with the liberal organization CREDO Action, ran an experiment to see whether donors really get preferential access to members of Congress:

In the experiment, CREDO Action requested meetings with 191 Congressional members to talk about a pending bill. Though all of the requests were on behalf of CREDO members who had made political donations, the organization randomly selected whether to tell the elected official that they were meeting with donors or ordinary constituents.

A total of 86 congressional offices agreed to meetings. Senior staffers, such as chiefs of staff or deputy chiefs of staff, showed up to meet identified donors at 19% of those meetings, with actual members of Congress attending 8%. But only 5% of the meetings with ordinary constituents were with senior staffers, including a mere 2% with actual members. The majority of meetings, whether with donors or constituents, were with Washington D.C.-based legislative assistants or local district directors.

John Sides interviews Kalla and Broockman about their findings:

Q: What does your research tell us about the quality of American democracy? Should it make us more concerned?

A: The results are clearly concerning. Most Americans can’t afford to contribute to campaigns in meaningful amounts, while those who can have very different priorities than the broader public. Concern that campaign donations facilitate the wealthy’s well-documented greater influence with legislators has long inspired reformers to make changes to the system of campaign finance. Our results support their concerns. If legislators are surrounding themselves with individuals who can afford to donate, they’re going to receive a distorted portrait of the public’s priorities and hear a distorted set of arguments about what is best for the country.

Jennifer Victor is skeptical of the results:

[T]his experiment has not yet been published in a peer-reviewed outlet and its findings contradict those of published research. Chin, Bond, and Geva (JOP 2000) also used an experimental design to determine whether contributing groups received more access than constituents, and they did not. Again, Chin 2005 finds that staffers grant meetings based on contextual attributes about groups and constituents, rather than contribution history. The fact that the recently reported research depends on variation within a single group, rather than the more advantageous across-group design of the published works in this area, hinders its ability to offer a generalizable finding.