As promised, I have undertaken a little analysis of the 2012 General Election Results. First, I intend to do a little update to the discussion on the accuracy and potential bias of the exit poll conducted by student-volunteers and led by Dr. Ron McNinch and my former co-worker John Pineda (currently pursuing a Master's in Public Administration from UOG). Observe the following graph, which represents the corrected data (because I am using the Official GEC Election Count, not the Unofficial Result). This time, I represent it differently than before because I put the Official Result on the x-axis and the Exit Poll Result on the y-axis. This is really mostly about style, but also because people expect the "explanatory" variable to be represented on the x-axis. Since we expect the exit poll to be representative of the final result, the exit poll result is considered the dependent variable. But this doesn't matter so much. While I left out the R-squared and the equation of the line (and the line I represent is for demonstration purposes only, to show what it would look like if the exit poll were perfectly representative of the official result, in which case every point would fall on a line that runs parallel to the one represented).

There is still a demonstrable Democratic bias in the exit poll, based on the official results. In fact, just for full disclosure, I came up with a simple regression to test it. The following are the results of the regression analysis and the t-scores (put in brackets) of each coefficient (with the results reported in logs):

Exit Poll Results = -1.569 [-7.194] + 0.9325 [17.32] x Official Results + 0.06535 [4.059] x Democrat Dummy Variable

Each of these coefficients has a high degree of confidence (like 99.95% confidence interval or better), so this seems to be empirical confirmation of my hypothesis that there is a Democratic bias in the poll. I am not saying this was done intentionally and I am not sure whether it could be easily avoided or whether any precaution could have averted the bias. Also notice that the 0.9325 number is not the "1" that one would hope for (although 1 is within the 95% confidence interval). This may indicate that there is still a bias that I am not catching with my simple Democratic dummy variable. Maybe in the future I will look into whether there are variables I can throw in that would demonstrate any non-partisan bias (like age, sex or education of candidates).

What about my simple past predicting the future hypothesis? How well does that work with the most recent election (2010) and the 2012 General Election? Look at the graph below:

Notice that I put the 2010 Election Results along the x-axis (since we are trying to explain the 2012 Election) and the 2012 Election Results along the y-axis. The line represented has a slope of 1 and is "calibrated" so that it runs through the average of the candidates that have more or less kept their relative standing with each other. My initial guess about what would happen to candidates that ran unsuccessfully for the Guam Legislature in one of the previous two elections (both 2010, Joe San Agustin and William Sarmiento) has been verified. They both did better this time, relative to the stable candidates (which are all incumbents). I had a strong feeling that Benjamin J.F. Cruz' position was abnormally low for him in the 2010 election and that he'd "bounce back", which seems to have been vindicated. I had thought that some first-term senators would have gotten a noticeable "punishment", but that appears to be wrong. It looks like many of them have held their relative position since 2010, but Chris Duenas and Dennis Rodriguez, Jr., have considerably outperformed the last election. Obviously Tony and Tom Ada, respectively, have improved their standings, too, although I had not made any guesses about that.

I did not publicly make these predictions, so I suppose I don't get any credit, but I had worked on a very rudimentary model, which had these predictions worked in.

I congratulate all those who have made it in to the 32nd Guam Legislature. In a few days, I will probably return once again to more economics blogging (I have a piece I am trying to work on to address the question of what caused the 2008 financial crisis and the ensuing depression).

**Sort-of related posts:**

How well does the last election predict the next election on Guam?

Guam General Election Results & Exit Polls

Guam general election 2012

Memo about polling and Guam's election

A few thoughts on elections from the demand-side

Guam Legislature from a labor perspective and a rallying call

Old hands, new hands and flashbulbs

Comments on the Calvo administration's 'spending cuts' and the debt ceiling

Possible response from McNinch (different topic)

Lee Webber admits why he wants a part-time legislature

Functions of the Guam Legislature

A view of Guam's Primary election

## No comments:

## Post a Comment