Straight Party Voting and Down Ballot Outcomes: The Impact of Indiana s Public Law 21-2016 Jay S. Bagga, Ph.D. & Bryan D. Byers, Ph.D. VSTOP, Ball State University With Special Assistance From: Joseph Losco, Ph.D., Consultant to VSTOP James A. Jones, Ph.D., Statistical Consultant Director of Research and Academic Effectiveness Ball State University Aniketh Ramname, VSTOP Project Manger Manikantesh Kilaru, VSTOP IT Specialist
Straight Party Voting (SPV) States Indiana is one of 9 states which allows SPV: Alabama Indiana Pennsylvania South Carolina Kentucky Texas (ending in 2020) Michigan Utah Oklahoma Source: National Conference of State Legislatures (ncsl.org)
Ten States have Abolished SPV since 1994 Georgia (1994) New Mexico (2001, 2012) Illinois (1997) North Carolina (2014) Iowa (2017) South Dakota (1996) Missouri (2006) Wisconsin (2012 except UOCAVA) New Hampshire (2007) West Virginia (2015) Source: National Conference of State Legislatures (ncsl.org)
The Issue in Indiana Some contests create issues with SPV: At-Large (N of M) races which are partisan Down ballot straight party voting cannot be used due to limits on choices For example, if 4 persons can be elected to an office and there are only 3 candidates for a given party X, then a straight party vote for X causes an undervote, unless the voter makes some individual choices. In combination of choices from different parties, the voter s intent needs to be clear School Board Races are non-partisan Down ballot straight party does not apply because School Board Candidates are not identified by political party, thus straight party/ticket selector does not apply We were also interested in the down ballot impact on School Board contests
Example Straight party voted: R At Large Office (Vote for three) Candidates Candidate 1 (R) Candidate 2 (R) Candidate 1 (D) Candidate 2 (D) Ind. Candidate Voted X X X
Public Law 21-2016 Summary of Instructions on the ballot: Straight party votes do not apply to at-large races (county council, city common council, town council, or township board) To vote for any candidate for an at-large office you must make another voting mark for each candidate you wish to vote for Your straight party vote will not count as a vote for any candidate for that office Source: 2016 Indiana Election Legislation Summary, Prepared by the Indiana Election Division
Study Request and Scope The request for the study of the impact of PL 21-2016 came from the Indiana Secretary of State With the new law, the Secretary wanted to know if voters would follow instructions and go down ballot and make choices This is a particularly important issue when it comes to SPVs VSTOP designed a state-wide voting study: Aggregate level voting data for 2008, 2012, 2016 Ballot level data from 10 counties (only 2016 available) Best Practices from counties for implementing the new law in 2016 General Election The study was approved by the Secretary and carried out during the spring of 2017
Methodology Developed data collection templates to ensure that all data formats were consistent for ease of data analysis Collection of data: Aggregate Level: From counties as well as web posted election results for 81 counties in the state (data available for all three years) Ballot Level: From Vendors who were able to retrieve vote data for 10 counties for the 2016 General Election Best Practices: Contacted all 92 County Clerks for information (68 responses received)
Analysis Aggregate Level data was organized and analyzed in Excel Ballot Level data was obtained in Excel and converted to SPSS for analysis Statistical Analyses: Descriptive (e.g., frequencies, percentages, ratios, means, SDs) Inferential (e.g., cross tabulations with chi-square, ANOVAs) Comparability of Aggregate and Ballot Level Data (Mann-Whitney U Test) When appropriate, data were subjected to tests of statistical significance
Findings Ballot Level Compared straight party voters and non-straight party voters on down ballot falloff for at-large and school board contests SPV Average Non-SPV Average At-large (Vote for 3) 1.83 2.31 School Board (Vote for k) 1.57 2.00
Findings Ballot Level These differences held across all 10 counties Held for election day and absentee Held for DRE versus OPSCAN (however, not all 10 counties employed both systems)
Findings Ballot Level Partisan Comparisons: In 6 of the 10 counties, the difference between R and D voters who voted for no at-large candidates was no more than 10% Republican straight ticket voters were more likely to vote for at least one atlarge candidate There was no consistent pattern for school board voting based on voting straight party R or D Partisan differences were not as significant as were the overall straight party selection impact on down ballot outcomes
Findings Aggregate Level At-Large Contests Down ballot drop off was found in results based on 81 Indiana counties We used an undervote ratio. For instance, for a vote for 3 office, with n voters and no undervotes should yield 3n votes. If there are m votes tabulated for this office, the undervote ratio is 1 (m/3n). Year Undervote Ratio Average Undervote Per Voter 2008 0.19 0.57 2012 0.22 0.66 2016 0.31 0.93
Findings Aggregate Level School Board Contests Vote for k (n=44) Down ballot drop off found Voting increased between 2008 and 2012 and remained relatively level in 2016 Data limitations: Fewer than half of the counties had data k varied by school district Therefore, the school board data in the aggregate do not carry the same level of confidence
Findings Comparability of Aggregate and Ballot Level Data Did the incidence of straight party voting and average number of undervotes differ significantly in the more granular ballot level data versus the aggregate level data or do these figures represent similar patterns? The ballot and aggregate level data were subjected to the Independent Samples Mann-Whitney U Test.
Findings Comparability of Ballot Level and Aggregate Data SPV Undervoting SPV Undervoting Population (n=81) Sample (n=10)
Mann-Whitney U Test The ballot and aggregate level data were subjected to the Independent Samples Mann-Whitney U Test For each comparison, we failed to reject the Null Hypothesis, meaning that the two data sets are similar statistically This gives us confidence that ballot level data was comparable to the aggregate
Findings Best Practices Education and information through media at State and local levels Orientation at Conferences Poll worker Training Materials at SOS website Compliance with Indiana Election Code Posted Instructions Public Tests (Cover Straight Party Scenarios)
Key Take-Aways Down Ballot drop off was found in both the Ballot and Aggregate Level data In the Ballot Level data, the drop off was statistically significant and in the aggregate level data there was drop off evident as well
Key Take-Aways While Down Ballot Drop off occurred in the data, this could be due to: The impact of PL 21-2016 or Those running for office in the contests or A combination of both We tend to agree with the first reason given the pattern emerged across counties and the data at the ballot and aggregate levels were comparable based on the Mann-Whitney U Test