Thursday, February 6, 2014

Response to the National University System Institute for Policy Research (NUSIPR) Flawed Mayor's Race Study



Guest Post by Henry Kim Assistant Professor, Political Science University of Arizona
The National University System Institute for Policy Research (NUSIPR) has issued a recent policy brief regarding the San Diego Mayoral election that is largely grounded on potentially fallacious interpretation of pertinent facts.  The only non-trivial claim being made is that “Faulconer leads in early voting,” as the title of the report indicates.  To support this conclusion, NUSIPR draws on two facts: 1) Republican absentee ballots are being returned at a higher rate than Democratic ballots, and 2) turnout in voter precincts that favored Faulconer is higher than in pro-Alvarez precincts.
Difference in the rates of turnout among different parties’ absentee voters provides no useful information about which candidate is leading in early voting, without reference to the difference in their actual numbers. For example, Reform Party absentee voters currently lead both democrats and republicans in early voting with a whopping 56% of ballots returned to date, but no one would seriously argue that the Reform Party’s 266 ballots are “leading in early voting.”  As the Table 3 of the NUSIPR report indicates, the raw number of Democratic absentee votes is substantially larger than the Republican:  despite the lower rate of turnout, Democrats currently lead Republicans by about 3,000 votes in ballots submitted.
By concluding that Faulconer is leading early voting because turnout in voter precincts that favored Faulconer in November is higher than those precincts that favored Alvarez, NUSIPR is ignoring the substantial differences in the electoral landscape between the first and second rounds of the election.  In the first round of the election, Democratic votes were split among Alvarez and two other major candidates.  With only two candidates of different party affiliations remaining, it is improbable that the second round votes will simply replicate the patterns seen in the first.  Furthermore, the electorate of the second round is likely to include a large number of new voters who did not participate in the first.  Over 16% of the ballots cast thus far are from voters who DID NOT vote in the primary.  Their vote choices cannot be reliably predicted simply based on the patterns seen in the first round.  Point 2 thus represents an excellent example of an “ecological fallacy”: NUSIPR infers that the individual general election absentee voters will behave the same as the group (primary polling locations) to which they belong.       
So who is really leading early voting? Based on the relevant facts, the most likely answer is Alvarez. Let us take a closer look at the data.
Table 1 shows an increase in the percentage of Democrats and Latinos and a drop off in the percentage of Republicans since early voting has started. The columns show early voters as of 1/19/14, between 1/19 and 1/30, and as of 1/30.  The rows show totals for party identification and Latino ethnicity.  Table 1 shows clear trend that favors Alvarez: between 1/19 and 1/30, the pool of early voters has grown more Democratic, more Latino, and less Republican.    
As shown in Table 2, the new entrants to the electorate (those who did not vote in the Primary on 11/19/13 and make up about 16% of the ballots so far) are more Latino and less Republican.  The Republican share of new voters is 7% less than its share of all early voters, the Democratic As shown in Table 2, the new entrants to the electorate (those who did not vote in the Primary on 11/19/13 and make up about 16% of the ballots so far) are more Latino and less Republican.  The Republican share of new voters is 7% less than its share of all early voters, the Democratic percentage is roughly the same, and the Latino percentage is 4 points higher.

The relevant facts presented in these tables are even more striking given that early voters tend to be less diverse and more republican than the electorate as a whole.     

Table 1: Total Voted


Table 2: Total New Voters (did not vote in primary on 11/19/13)

No comments: