2. Does
Money
Buy
Influence?
Problem:Widespread perception in the media, among political
scientists, and in the general public that political contributions drive
political influence.
“Americans believe there are continued transparency and corruption
issues in local, state and national government” (Transparency
International, 2012)
Hypothesis:The level of political contributions to state political
campaigns increases the number of bills signed into law at the state
level.
Ed
5. Data sets from the Sunlight Foundation:
Influence Explorer: State campaign finance data.
Open States: Information on state bills, the legislators and activities of all 50 state legislatures.
Supplementary data on state legislatures from NationalConference of State Legislatures,Wikipedia.
Scope
Originally looked at 10 states but expanded due to limited data.
Final analysis included all 50 states.
Performed correlation analysis on:
Number of bills signed (dependent)
Contributions, number of legislators, average amount of contributions per legislator, length of legislative
session (independent).
Performed multiple linear regression on:
Number of bills signed (dependent) and
Contributions, number of legislators, length of legislative session, and state population (independent).
Methodology:
Mai
6. Results
Initial analysis of aggregate variables with the ten-state data set:
No significant correlation between bills signed and contributions at the gross aggregate
level (across all states, years).
No significant correlation between the number of bills signed and state legislators or $
per legislator (undifferentiated average).
Strong and significant correlated between number of bills signed and the length of the
legislative session (p < .001, R2 = 0.84).
At regional level, notable relationship between bills signed and contributions in four
Southern states (Florida, Georgia,Texas, and North Carolina).
Broader analysis of fifty-state set less conclusive:
No significant correlation between bills signed and contributions, state legislators, $ per
legislator, length of the legislative session.
Regional relationship had weakened to insignificance.
Multiple linear regression no more productive.
Best linear regression model generated an R-squared of 0.15
Political contributions made up only 0.03.
State population made up 0.12.
JanOriginal Ten-State Data Set
Ten-State Data Set: Southern Region
Fifty-State Data Set
7. Discussion
Problems with the data sets:
Completeness of data on bills signed varied from state to state, year to year.
State-level political contributions data even more incomplete.
Problem multiplied by legislation/contribution match-up (i.e., state X data present for
bills but missing for contributions).
Only 89 data points out of possible 250 state-year pairs.
Some state legislatures are not in session every year.
Very difficult to properly compare cycles of contributions and legislation across different
states at aggregate level without more detailed assignment work.
Factors unaccounted for by hypothesis could influence number of bills passed.
Contributions to stop proposed bills from passing might offset effect of contributions to
support legislation.
Party balance of state legislature, degree of bipartisan vs. contentious politics in state,
and of party affiliation of legislature vs. governor will impact number of bills passed
compared to the number signed.
Mai
8. Conclusion
Campaign contributions do not appear to drive the volume of
legislation signed at the state level, at least in aggregate.
Our hypothesis, thus, is not proven.
At levels below national aggregate, there seems to be some
relationship.
Analysis more closely tied to specific regions and taking account of
state legislative idiosyncrasies, may prove more productive.
But first step should be to re-evaluate assumptions in light of
actual problem:
Our terms (legislation signed and undifferentiated contributions)
may not be best way to model influence of money on politics.
May need more detailed ground work, connecting data on specific
topics/causes for contributions with legislators’ positions and voting
record.
Ed