Google Analytics

Site Meter

Wednesday, May 17, 2006

The Effect of Endorsements…

The question is a big one, and one that people have been trying to measure for decades.  From the days of the unions to the farm workers, what is the impact of an endorsement on a candidate?  Many VP’s have been selected for this purpose. Pacs and groups have used their endorsement stickers as ways to negotiate with candidates and cut backroom deals for years. 


Well, as you get more and more local, the endorsements should have a bigger impact, as voters would be more in touch with each candidate and the group that is endorsing the candidate.  The most important endorsement, by far, is a yard sign in a voter’s yard.  This is the best endorsement possible, as it tells people that their peers and friends support a specific candidate.  If you look at your neighbors’ yard or friends’ yard and see a yard sign for a candidate it has a great effect on your opinion of the candidate. 


Endorsements by local groups also have a bigger impact versus national groups.  If you are endorsed by a local community group or civic association it has more pull than the endorsements by the national parties or groups.  So the overall effect of endorsements on candidates in a local race has always been interesting.   We just finished a local race here in Alexandria.  The voter turnout was low and the most tightly run race was for school board.  There were many endorsements this year, and though they mostly had a positive lift, the overall impact was not enough to make too many differences. 


To understand the impact of the endorsements we ran a regression analysis on the different candidates and precincts.  The model mapped all the candidates for each endorsement they received and the overall votes they received in each precinct.  We included some models to control for home precincts and current incumbent candidates. There were controls put in for party also, as there is an automatic lift for any Dem running in Alexandria.  To really understand the impact we needed to look at the share of the vote each candidate garnered within each precinct and also how this share correlated with other candidates. 


In order to ensure this measurement was as independent as possible, non-qualitative measures were used to identify the overall impact.  Measures included a wide range of independent measures, such as: win margin, share of total vote won, percent of possible vote, total votes, votes segmented by party, votes per voter estimated by candidate, and correlation of overall votes and percent of votes across School Board and City Council. 


Luckily, to reduce error, the races that we measured allowed for multiple votes per voter.  For example, for City Council each voter had six votes, out of eleven candidates.  For school board each voter had three to select from out of four to five.  Also, by segmenting by precinct, it was much easier to control for an anomaly that could have been caused by heaving up by candidates within a specific precinct, or the impact of a home precinct. 


So the findings are not too surprising overall.  The real story is that endorsements, in almost all cases, have a positive impact on the candidates that were endorsed by the group.  In most cases the endorsed candidates correlated with each other, and the non-endorsed candidates correlated.  The main point to realize is the measurement of endorsement is a measure of not only the overall impact in voters making decisions, but also in their decision to vote overall. 


The data from the recent 2006 city race shows that the biggest winner of the vote was a group that was called the 3%’ers.  It was a tax control group and had a big impact on the candidates that were endorsed by this group.  When controlling for party and for incumbency, the 3% group had a positive effect on the endorsed candidates by 3%.  This showed up in the total vote garnered by Andrew MacDonald and the high correlation he had with the other 3% candidates.  It was such a big impact he garnered the most votes for City Council.  The overall impact on Andrew alone could have had an even larger impact, but because he was endorsed by other groups it is difficult to extract any additional upside from the race.


Other endorsements did have a positive impact also.  The Washington Post did endorse the candidates running for School Board.  This endorsement was published on the Monday before the election and listed out the candidates for each precinct.  The correlation between these candidates was positive, and the lift appeared to be around 1% on these candidates.  This impact could have been much larger if the total number of people coming out to vote was larger.  It is expected that the Post endorsement did not bring out voters, it just informed them.


There were other endorsements and they all had some impact, though it was limited and hard to say for sure the estimate is reliable given the different groups.  The Sierra Club, the Parks groups, and ASG are a few of the groups that also did candidate endorsements.  There was a high correlation between these candidates, though this might be explained by the very targeted groups that probably already correlated with the candidates. 


The only group that had a negative impact was the school union.  This is the teacher endorsed sticker.  The candidates that were endorsed by the school union actually had a lower percent of the vote than the non-endorsed candidates as it relates to their share.  The margin of winning by the non-endorsed candidates versus the endorsed showed that the school union possibly had a negative impact to the candidates.  Now, this does not take into account other factors, and because the correlation is lower than indexing the other endorsements it appears to be more random than anything, which would suggest that not only is a negative impact, but it also could have absolutely no positive impact. 




So, what does this mean?  It means that having the ‘Teacher Endorsed’ stickers really did not make any positive difference in this election. There were many other factors, but the union was not one of them.  This was probably mostly caused by who turned out to vote, but it also shows that voters really place no value in the so called ‘Teacher Endorsed’ tag. 


The school union needs to look for new ways to offer their support and/or endorsement.  Maybe it is to look beyond the party and look at a better school board.  The 3% group included some Dems, the Washington Post included many different candidates, and the school union toed the party line.


It is hard to say what will happen in three years, but we do know that today, the union did not get the outcome they wanted.  I support the Teachers Union, it does great things, but we need to find a way to make a better impact.  Time to regroup. 

Disqus for Cold Kiwi