Observers of the recently-conducted Louisiana state elections have wondered whether Gov. Bobby Jindal’s victory constitutes a mandate. Some have suggested, by pointing out that Jindal actually drew several thousand votes fewer than in his similar 2007 general election win, and that overall turnout for the contest was lower by about a fifth, that his win did not. Statistical analysis, however, proves otherwise.
One method that could reveal definitively that 2011’s results did not produce a mandate, or an overwhelming agreement to back Jindal’s agenda, is to look at the comparative change in turnout among all statewide elective single executive office contests. According to the logic of no mandate, increased nonvoting signals a lack of interest in the contest and, by implication, tepid endorsement by the entire public of the winner’s platform.
If that is the case, then the comparative loss to turnout in the governor’s race should be greater than for other executive offices.
Indeed, that should be especially noticeable in 2011 because, while like the gubernatorial most of these contests were noncompetitive, two, lieutenant governor and secretary of state, were barnburners and should have stoked much more interest than they did in 2007, compounded by the fact that then the lieutenant governor’s race at best was mildly competitive with its winner gaining a larger proportion of the vote than Jindal, and the secretary of state’s race was even less competitive.
However, the comparative results show the opposite. In 2011, turnout for governor was 78.83 percent of that in 2007, higher than for both of these offices, and in raw numbers also exceeded both in 2011 (also the case in 2007). Put another way, the non-competitive gubernatorial race held onto to proportionally more voters relative to a mildly competitive 2007 contest than did the very competitive other two contests in 2011 compared to much less competitive 2007 races. In an overall environment of less competition in 2011 – and why not, since Democrats did not field a single, major and quality candidate for any of the single statewide executive posts – more people proportionally were stimulated to vote in an governor’s race considered a foregone conclusion than they did in two highly competitive contests. This certainly does not imply a lack of an endorsement.
But another test is needed to ascertain the overall public mood, among voters and nonvoters alike, was widespread acceptance of Jindal. The former we know through the results and the latter may be discovered by looking at the relative change in total votes from 2007 to 2011 and how that relates to support for Jindal. This can indicate a rival hypothesis to the idea that nonvoting signals indifference: that nonvoting in fact shows satisfaction with one or all choices.
The act of voting incurs several costs to individuals, such as having to inform themselves, to find the time available to do it, to physically casting the vote, and to experience bureaucratic disincentives like waiting in line. Unless they feel their votes are of sufficient importance in the election, ultimately because they may decide the outcome, or because they feel compelled by civic duty to vote, the costs of doing it exceed the benefits, especially when alternative actions may produce greater perceived benefits (such as going to football games, hunting, relaxing at home, etc.).
Therefore, if somebody surveys the candidates available and draws one of two conclusions, that one is a sure winner or that they all seem acceptable in office (or most likely a hybrid, that the only ones likely to win all are acceptable), given the costs and benefits of voting compared to alternative activities, nonvoting makes sense. And in the case of this election, given that Jindal’s opponents received such small shares of the vote and he won over his nearest competitor by nearly 50 percentage points, likely the vast majority of nonvoters did not show up because they liked Jindal and/or saw him as a sure winner.
An empirical test can verify this. The change in Jindal’s share of the vote can be compared, through a statistical procedure popularly known as regression, to the change in overall turnout by parish. If there is a positive and significant relationship between the two – that is, in a parish as turnout in 2011 was higher relative to 2007 then Jindal received a greater share of the vote in 2011 than in 2007 – this implies that nonvoters staying home disproportionately would have voted for Jindal had they voted, as, if they had, his share would have increased in the lower-proportioned parishes to match that of those higher-proportioned.
That is the case. A coefficient can be calculated between -1 and 1 where -1 means a perfect inverse relationship, 0 means no relationship, and 1 means a perfect direct relationship. At 0.17, it shows a weak positive relationship; the higher turnout was relative to the last election, the greater the proportion of the electorate who voted for Jindal. Thus, we may conclude nonvoters stayed home not so much out of opposition to Jindal (and other candidates), but because a mildly disproportionate number of them were satisfied with what they saw as his imminent win without their input.
Posted by Jeff Sadow at 11:00