Money, Votes, and the Power Dynamics of Gun Laws
The math of Pennsylvania’s firearm legislation
Co-Authored by Sam Bydlon, Salil Malkan, and Marty Santalucia
In recent years, guns, and the unique political space in which they exist, have increasingly occupied our national attention. Guns are triumphed by some as an embodiment of our constitutional rights and mechanisms of self-defense while others denounce them as the primary culprit in a deadly epidemic responsible for killing tens of thousands of people each year. The issue has become a favorite of our politicians who use it to split voters and foster an “us” versus “them” mentality. Guns have become so divisive and controversial that, well before any of the actual content of the orders was made public, President Barack Obama was roundly criticized by Republicans when he signed 23 executive orders addressing gun safety. The conflict has frozen our legislative process and, at least in congress, inaction on this issue has become the status quo. In contrast, state’s don’t often suffer the same level of gridlock that our federal government does. Thanks to our federalist roots, states have been empowered to act on major issues like immigration, marijuana legalization, and marriage equality years before the federal government takes any meaningful action. In regard to gun policy, state legislatures might be the logical place to start looking for answers. In today’s post, we will explore the legislative dynamics behind this intricate and controversial topic.
To effectively explore the issue in the Pennsylvania legislature, we focused on gun related legislation. This meant stripping our data set of over 600 votes from the 2015–2016 Pennsylvania House down to just 30 pieces of legislation that contained the keywords “gun”, “rifle”, or “firearm”.
Using just 5% of the total available data, we were concerned that not every member took enough votes to produce a statistically significant result. To address this, we calculated that the average house member took 29.72 of the 30 votes, with a standard deviation of 1.96 (the distribution is highly skewed toward 30 votes). Anyone whose vote total was less than 2 standard deviations below the average (26 votes) was removed from the analysis. This removed Representatives Bullock, Krueger-Braneky, McClinton, Neilson, and Rothman from our data set. Relationships between the remaining members were based on how frequently they voted with each other.
Blue lines represent strong relationships; white lines represent moderate relationships; red lines represent weak relationships.
Our resulting network (displayed above) has three separate components. Two of these components represent the major parties; Democrats in the top left corner and Republicans in the bottom right corner. Between them, in the upper right corner, is a small bipartisan component of five members that has four Republicans and one Democrat. The fact that this small component contains the network’s only bipartisan connections makes it interesting and worth investigating. We figured we might be able to use NPAT Common Space Scores to examine this unusual group. These scores are available online and are annually updated by Boris Shor of Georgetown University and Nolan McCarty of Princeton University. The scores allowed us to quantify the general ideology of state legislators by putting them on a scale from high to low. A higher score indicates that a member is more conservative, a lower score indicates that a member is more liberal.
According to the NPAT scores, we were able to find relationships between the most conservative member of the House (Representative Daryl Metcalfe), a liberal Republican, and even a Democrat (Representative Tony DeLuca). This isn’t necessarily an unexpected result though. We limited our network because we wanted to find subtle, issue-specific relationships that might be overshadowed in the full data set. Similarly, NPAT scores are calculated by taking all of a member’s votes into account to quantify an overall ideology, not an issue-specific ideology. If these members are ideologically similar on gun issues, that subtlety might be lost under all the other data used to create the score. Basically, our network and the NPAT scores are measuring two different things so we can’t compare the two. Later in this post, we’ll return to the idea of a gun-specific score.
Our current model suggests that the parties are almost completely isolated from each other on the gun issues. A bill typically requires 102 votes to pass, more than 50% of the 203 total members. Legislation can only move through the House if at least some of the 119 Republicans support it. Additionally, it is hard to see how a bill introduced by a member of the bipartisan component would be able to get any traction given the relative isolation of that component. Votes are obviously only part of the story, so we wanted to build and explore a more robust data set.
There is often a financial component where donors (both individuals and organizations) will contribute to the campaigns of elected officials they agree with, especially with controversial issues like gun control. Using data from FollowTheMoney.org and the Pennsylvania Department of State, we pulled all available contributions for 2014 and 2015 to current legislators from gun control and advocacy groups. Gun control and gun advocacy groups contributed a total of $23,265. Gun control contributions come from only one group, Ceasefire Pennsylvania. Ceasefire Pennsylvania is an organization that works with voters and elected officials across Pennsylvania to address gun violence issues through education and gun control legislation. Gun advocacy contributions come from the National Rifle Association (NRA) and Firearm Owners Against Crime (FOAC). Both of these groups are dedicated to the preservation of Americans’ second amendment rights. Both groups are very active in backing candidates who also support the rights of gun owners. FOAC is a Pennsylvania specific organization based in Western Pennsylvania, while the NRA has a national scope but is active on the state level.
One noteworthy assumption we made was that donors contribute to elected officials as a way of supporting those who are already ideologically in-line with them. There is a notion that campaign contributions are used by wealthy individuals and organizations to incentivize a member to adjust their position on a given issue. We haven’t done any hard data analysis to say whether or not this is the case. We do, however, acknowledge that fully exploring whether certain money moves politicians, or whether politicians attract certain money could be a post by itself. For now, we assume that money helps win elections by enabling candidates to better communicate with voters.
The table below shows the breakdown of these gun-related contributions by the recipient’s party and their stance on gun issues:
Gun advocacy groups were responsible for 82% of contributions. Additionally, of the $19,000+ in contributions from gun advocacy groups, 73% went to Republicans. Gun advocacy groups also spread their money among more candidates by contributing to 32 candidates, at an average of $594 each. Comparatively, gun control groups contributed to just 8 candidates, at an average of $321 each.
The Pennsylvania state financial landscape echos the deep partisan divides we see on the national level. Compare the table above with the table below; the table below breaks down contributions made directly to federal candidates across the country from gun control and gun advocacy groups during the 2014 election. The pattern, in which Republican gun advocates receive the overwhelming majority of contributions, is similar. At the federal level, however, the discrepancy is much larger.
To flesh out the data a little more, and make it comparable to our voting data, we created a financial relationship network. Relationships in the network are based on members taking money from the same groups. For example, Representative Daryl Metcalfe and Representative Harry Readshaw received $2,000 and $1,000 from FOAC, respectively. Because both members took money from the same group, we draw a relationship between them. The weight of the relationship is determined by how much money they received. The average of their contributions is $1,500 which is the value we assign to the weight of the relationship. When we draw the network blue lines are stronger relationships and red lines are weaker.
Representative Krueger-Braneky (the recipient of $2,000 from Ceasefire Pennsylvania) was excluded from our vote network for not having enough voting data to include. For consistency, we will also be excluding her from the financial contribution network.
On the left are the members who received contributions from the gun control group Ceasefire Pennsylvania, and on the right are the members who received contributions from the gun advocacy groups FOAC and the NRA. Comparing this network to our vote network, we noticed that three of the five members in our small, bipartisan component have taken money from gun advocacy groups. Representatives Dan Truitt and Warren Kampf, who are missing from the financial contribution network, vote frequently with gun advocates but do not take money from gun advocacy groups. If we were on either of their campaign finance teams, we’d be emailing these Representatives the phone numbers for the NRA and FOAC as soon as possible.
The vote and financial contribution data sets complement each other by independently confirming the divisive, yet simultaneously bipartisan, environment we regularly observe around this issue. Our analysis is limited if we only consider each model in a vacuum. We wanted our network to reflect the real complexities of the issue, which we decided would be best accomplished by merging our vote data and our financial contribution data into a single data set.
The first step in this process was validating that our financial contribution network was actually measuring the ideology of any given member. Without the use of the NPAT scores, we needed a way to quantify gun-specific ideology. We accomplished this by creating a gun advocacy score using FOAC’s public position statements on 75 bills in the Pennsylvania General Assembly. A member was given 1 point for voting with FOAC’s position or sponsoring a bill that FOAC supported. A member lost 1 point for voting against FOAC’s position or sponsoring a bill they opposed. After calculating scores for every member in our financial contribution network, we used a two-way ANOVA analysis to compare the average scores of members who took gun control money versus those who took gun advocacy money. We found a statistically significant difference between the average scores of the groups (P = 0.001). Those who received money from the gun control group had an average score of -5. Those who took money from gun advocacy groups had an average score of 9.4. Further, a member that scored higher on our gun advocacy score, typically received more money from gun advocacy groups (P = 0.02). In short, we can use financial contributions to measure a member’s ideology on guns.
For both of these tests, our confidence interval was 95% with P < 0.05.
We wanted to merge the data sets by averaging together corresponding relationship weights in each set. If two members voted together 75% of the time and had a financial relationship of $500, we wanted to simply average these values to get the relationship weight in our composite network. This raised some compatibility issues, because the network relationships were on two very different scales. While vote similarity was capped at 100%, financial contributions could theoretically be infinite. To ensure that both data sets were on the same scale, we normalized the financial contributions such that every contribution value was between 0 and 100. A member with a normalized financial contribution value of 0 received no money from any of the three firearm-related groups. A member with a normalized financial contribution value of 100 received one of the largest contributions in the data set.
In our data set, Representative Daryl Metcalfe (R) received the largest total amount of money from these groups ($2,000 from FOAC), so he has a normalized financial contribution of 100. If other members had also received a contribution of $2,000, their normalized financial contribution would also be 100; however, Representative Metcalfe was the only member with a normalized financial contribution of 100.
It’s worth noting that no member took contributions from both gun control and gun advocacy groups. Only one member, Representative Rick Saccone (R), received money from both of the gun advocacy groups.
Finally, we averaged the voting and financial contribution relationships between members to form a new composite relationship network.
When we merge the data sets, we get a graph that resembles our first network, with some significant additions. Most notably, unlike the vote network, none of the components are isolated from each other. Our inclusion of financial contributions has allowed us to identify many bridges between the Democrats and Republicans. Careful readers will also notice that all of the new connections involve at least one member who has accepted gun advocacy money. The bipartisan relationships formed by gun control contributions simply weren’t strong enough to be maintained in the composite network.
These relationships are the pathways that allow legislators to work together on gun-related legislation. Our model now has a mechanism to accommodate through which bipartisan action. Where these connections occur is very important to how the model functions. If, as the case is here, the only path between any Democratic member and any Republican member is through a gun advocate, the model is going to have a gun advocacy bias. We use betweenness centrality to identify members who are ideologically representative of the dominant position within the legislative body. The most central member in this composite network is Democratic Representative Ted Harhai who was given a score of 93% by the NRA in 2014.
Remember that there are still gun control supporters in both the Democratic and Republican parties that might serve as a backstop to legislation, but they are weaker than the gun advocates within the network. We think this is reflective of reality, which might be best summed up in Guns and Ammo’s July 2015 “Best States for Gun Owners” article:
20. Pennsylvania
Pennsylvania’s biggest gun issue has always been the battle between gun owners and municipalities. Last fall, outgoing-Gov. Tom Corbett signed a strong preemption bill to ensure that gun laws are consistent across the Keystone State and allows groups such as the National Rifle Association to sue cities that defy state gun laws. A lawsuit against gun control ordinances in Philadelphia, Lancaster and Pittsburgh was filed by the NRA in January. Preemption battles aside, Pennsylvania has a strong Castle Doctrine law, it doesn’t restrict tactical firearms, and all [National Firearms Act] items are available.
Pennsylvania isn’t likely to move up this list in the near future, due to the election of Governor Tom Wolf in 2014. Governor Wolf is ranked as one of the most liberal governors in the country, strongly supports gun control, and it is highly unlikely he would sign any gun advocacy legislation that makes it to his desk. This alone might stop the House from even introducing a piece of legislation that expands gun rights since it would be dead no matter how quickly or decisively they pass it. The Senate, which we did not examine, also factors heavily into what a final piece of gun legislation looks like. Regardless, if they chose to, the House could continue to try and push the state toward a more open gun culture.
This analysis resulted in two primary conclusions. First — money matters. This may seem obvious, but here we’ve quantified the political role of financial contributions with some creative network analysis. Analyzing which members get money from what organizations can add a substantial degree of depth to understanding how a legislative body will approach a specific issue. As we said, we don’t believe this is because money buys influence. We believe it is simply another quantifiable expression of the member’s ideology.
Second, we identified network dynamics which were unique to our issue-specific model and distinct from our general model. This included different ideological centers and different bipartisan relationships. Being able to isolate a legislative body’s position on an issue was a huge step toward our ongoing goal of shining light on how our legislative process works. We are looking forward to digging into many issues like this in the future.
As we explore more questions like the one in this post, we’d love to continue to share our work with you. Please take a second to look at the links below where you can stay up to date with VoteLab by following us here on Medium, liking us on Facebook, and signing up to get our articles via email. Please also share this article with your friends online through Twitter and Facebook so we know you like reading our posts as much as we like writing them. We really enjoy talking to people about the work we’re doing here, so feel free to reach out to us on Twitter or through email if you have any comments or questions. Thank you!
— Sam, Salil, and Marty
Stay up to date with VoteLab
Like us on Facebook
Follow us on Medium
Subscribe to our mailing list
Contact the authors
Sam Bydlon: email / twitter
Salil Malkan: email / twitter
Marty Santalucia: email / twitter
