Voter Registration and Trump's Victory in 2016

• In Event: Donald Trump and Voters

Abstract

In terms of quality of electoral administration, the US is an outlier, lagging significantly behind most liberal democracies. For instance, if we look at Pippa Norris’ “Electoral Integrity” index, US’ score in 2016 was 70 (on a 0-100 scale), while Finland’s was 86. Moreover, there is a lot of variability across states. At state level, the scores for the same index ranged from 75 (Vermont) to 54 (Wisconsin). Even though a poor quality of electoral administration is unfortunate, it does not necessarily have a partisan bias – poor design of ballots (e.g., the infamous “butterfly ballot”) or voting machines malfunction are not inherently advantageous for a specific party. However, we do have a specific dimension of electoral administration – i.e., voter registration -, where we do find a clear partisan divide, with the Democratic party favouring laxer provisions, while the Republican party takes the opposite stance. Thus, rather than looking at the impact of overall electoral administration on electoral outcomes (Norris’ approach), I only focus on the quality of voter registration. In one analysis, using electoral integrity as a predictor of vote for Trump in 2016, Norris found that electoral integrity had a large and negative impact on Trump’s share of the vote (b = -0.54; adjusted R2 = 0.09; n = 50). Norris also provides scores for voter registration. When using these (voter registration as a predictor of Trump vote), I get b = -0.72 and adjusted R2 = 0.25. All states that were crucial in Trump’s 2016 victory had low scores on this dimension, ranging from low 50’s (Florida, Michigan, Pennsylvania) to mid-40’s (Wisconsin). In theory, a 20-points increase on this dimension (say, from 50 to 70 – the scores of Vermont, Oregon and Hawaii) would have resulted in a massive drop (around ten percentage points) in Trump’s share of the vote in the former group of states, presumably because of the effective enfranchisement of minorities, a core Democratic constituency. It is also worth noting that, even if the scores for voter registration are not an accurate measure of reality (say, because of a liberal bias among political scientists providing these scores, computed based on the assessment/perception of specialists), it is still conceivable that they did have an effect. If scholars believe that voter registration rules are biased against minorities (think Duverger’s “mechanical effect”), perhaps this perception, accurate or not, is shares by minorities themselves, leading to a lower turnout among minority voters (Duverger’s “psychological effect”).

Author

• Florin Nicolae Fesnic, Babes-Bolyai University Cluj