Sunday, December 14, 2014

"Why the Democrats Lost: Part II" : Analysis of 2012,2013,2014 MatchBacks by Voting Age

The data from this analysis follows from Part I. The 2012 vote for Obama and Cantwell contained a much higher percentage of youth voters than the 2014 vote for the 42nd legislative seats.  Whether these youth votes were suppressed, discouraged, or uninterested in voting for  the Democrat candidates in 2014 is unknown. I will hazard this question:  Satpal Sidhu did the best of the Democrat  legislative candidates. He had the narrowest margin of loss.  Was it because he was perceived as more pro business or more moderate? Or was it because he was a person color who attracted more young Hispanic and Native American votes? Or both?  R Code is here.

The table below shows six categories of votes by age for the Whatcom County 2014 General Election. Table entries 1 : 4 are only from ballots marked as "good".  Undeliverables and Challenged ballots are part of categories 5 and 6.

MB2014RNPCT All 2014 GE Whatcom County Voters By Age
MB2014RNPCT42 All 2014 GE 42nd LD Voter By Age
MB121314PCT All Whatcom County Voters By Age Who Voted in all three of 2012,2013,2014 GE
MB121314_42PCT All 42nd LD Voters By Age Who Voted in all three of 2012,2013,2014 GE
MB12PCTin14PCT 2014 Registered Whatcom Voters By Age Who Voted in 2012 GE only
MB12_42PCTin14PCT 2014 Registered 42nd LD Voters By Age Who Voted in 2012 GE only

Note how the descriptive statistics for categories (1,2) of all of Whatcom County and the 42nd LD are equivalent and also  for categories (3,4) which consist of good ballots from those who voted in each of the last three general elections. Both those populations are older by mean and median than categories (5,6) and their statistical equivalence as such is remarkable. Those populations that voted only in 2012 (categories 5,6) but were still registered as of the 2014 GE have means and medians that show a much younger population.  The last two graphs in this post confirm the youth of those non voting (or not counted!) populations in a visual manner that this table of descriptive statistics can not.

Group MB2014RNPCT MB2014RNPCT42 MB121314PCT MB121314_42PCT MB12PCTin14PCT MB12_42PCTin14PCT
SampleSize 75,994 52,782 54,423 36,630 17,649 12,298
Mean 55 55 58 58 42 44
Stdev 17 17 16 16 17 17
Median 57 57 60 60 39 41
MAD 18 18 16 16 19 18
RangeDiff 91 91 89 89 81 81
Min.Age 18 18 20 20 20 20
Max.Age 109 109 109 109 101 101

The following two graphs show the age curves for all ballots marked as good for the entire county and the LD 42nd in the 2014 General Election. The graphs are nearly identical in shape and form and show the demographic dominance of a combined boomer/prewar (48 - 72 ?) "hump".

If we look at 2014 "good" ballots for just those populations that voted in each of the three of previous elections, we see the same demographic dominance of a combined boomer/prewar (48 - 72 ?) "hump".  These populations are even older in mean and median than the  all county, all 42nd 2014 votes.

Witness below the stunningly (youth) 'left skewed'  graph of those currently registered  that voted only in 2012 and did not vote in 2014. ('Left skewed' is a statistical description of a graph deviation from a normal distribution.)

No comments: