I analyzed roll call data for the vote in the House Financial Services Committee on the Campbell amendment to the Consumer Financial Protection Act (H.R. 3126). The amendment, which exempts auto dealers from certain oversight provisions, and roll call data on the committee vote are discussed in the maplight.org web page here. While Republican committee members voted in a block to exempt the auto dealers, the Democrats’ voting pattern suggests that they were influenced by campaign contributions from the auto dealer industry.

This figure shows the separate histograms for the yes and no vote contributions to Democrats on the Committee. Even ignoring the outlier, the distribution of contributions to Democrats who voted yes is weighted towards higher contributions compared to the distribution for those who voted no. The distributions overlap — there is no gap between the two distributions, i.e. there is no definite contribution amount that guarantees a "yes" vote.

This situation is ideally suited to analysis using logistic regression which is able to model how the probability of voting "yes" depends on contribution amount. For a discussion of logistic regression see the wikipedia page here. The model yields a forecast of the probability of voting yes as a function of contribution amount.

Here is a graph showing the fit from the data. This shows that $10,000 corresponds to an increase in the probability by 40% that a Democratic Party committee member would vote for the amendment, from about 30% given no money [i.e. probably voting against] to 70% when receiving $10,000. Break even occurs at about $5,000 — for which contribution a Democratic Party committee member is equally likely to vote yes or no. Here is a summary of the regression coefficients and their significance. For the regression jargon literate, the coefficient on the contribution amount is significant with a p-value of 4.4%.

Note, for this analysis I ignored the 2 members who did not vote.

Here is the data file, and here is the R code which performed the analysis.