Trends in Voting by Cohort

[Annatar]

I have recently been looking into whether cohort's change their voting patterns over time and whether the claim that as people grow older, they vote more conservative is true or false.
I looked at the 2008 CNN and 2016 CNN Exit polls to see whether over a period of 8 years, the same group of voters shifted more Democratic or Republican relative to the nation as they aged. The 3 generation's I looked at were: Baby Boomer's, Generation X, and Generation Y, also known as millennial's. The age brackets I used were as follows:

Baby Boomer's: 1946-1964
Generation X: 1964-1982
Generation Y: 1982-1990, I did not go past 1990 as the youngest voters in 2008 were born no later then 1990.

I calculated the 2 party vote for each generation by looking at the age breakdown, when generation's overlapped the age exit polls, I used averages and then checked back to make sure my data was congruent with the actual election results, the shift is calculated relative to the national shift.

My results were as follows:


2008: 2 party vote, 46R, 54D, -8R

Baby Boomer's: 50R, 50D, +0R, +8R relative to nation,
Generation X: 43R, 57D, -14R, -6R relative to nation
Generation Y(1982-1990): 33R, 67D, -34R, -26R relative to nation

2016: 2 party vote, 49R, 51D, -2R

Baby Boomer's: 54R, 46D, +8R, +10R relative to nation, +2R shift
Generation X: 50R, 50D, +0R, +2R relative to nation, +8R shift
Generation Y(1982-1990): 42R, 57D, -15R, -13R relative to nation, +13R shift

As can be seen, all the cohorts shifted republican compared to the nation as a whole, it should be noted though that due to migration and turnout differences, not all those who were part of one generation and voted in 2008 also voted in 2016. But broadly, the vast majority of those belonging to one generation who voted in 2008 also voted in 2016.


It is interesting to note that there has been virtually no shift among baby boomer's, they went from leaning 8 points more republican then the nation as whole in 2008 to 10 points by 2016, virtually unchanged. The major movement seems to have been in Generation X which shifted from being 6 points more democratic then the nation as a whole to 2 points more republican.

It seems ageing has had little impact on how baby boomer's have voted but had a much bigger impact upon Generation X, the shift in even bigger for those born 1982-1990 but the huge Democratic margins in 2008 among those voters were likely unsustainable and the shift therefore in all likelihood does not mean as much.


The question I have therefore is what caused Generation X to shift so much over the 2008-2016 period, from a generation that leaned moderately Democratic to one that leans slightly republican relative to the nation?

[Annatar]

As for ER, I think he's being a bit too eager to jump to conclusions, which is only possible if he's willing to ignore past results. Putting aside the voting habits of seniors under Clinton and other past presidents, in 2016, those ages 30-39 voted only marginally more Republican than they did when they were in the 18 - 29 group in 2008. In fact, if you factor in what was definitely a "backlash bonus" for Obama, there has been little shift to the GOP. The real damage was them going 3rd party instead of going for Clinton, which was a notable issue among 18-29 year olds as well.

If we exclude 3rd parties, in 2008, 32% of 18-29 voters went for Mccain, whereas 46% of the nation did so, a 14 point gap, in 2016, 39% of 30-39 voters went for Trump whereas 46% of the country did, a 7 point gap. No doubt there was a "backlash bonus" for Obama, but even with it, I do think 30-39 voters were significantly more Republican in 2016 then Democratic voters were in 2008. I agree with you though, if you exclude 3rd parties, the movement towards the GOP becomes less.

In any case, I do not think 18-29 voters in 2008 can be compared to 30-39 voters in 2016, firstly there is not perfect age overlap, secondly as voter turnout is higher for 30-39 voters, there must have been millions of young voters who stayed home in 2008 who proceeded to vote in 2016 when they were in the 30-39 age range.


This by the way is the primary problem with my analysis, even though 22-30 voters 8 years later will be 30-38 voters, there are vast number's of 22-30 voters who will have stayed home but will vote 8 years later, hence some of the shift towards the GOP over time may be due to the fact that voters belonging to that cohort which stayed home when they were younger came out to vote when they were older.

[Annatar]

I am familiar with Pew Research data, the issue I have always had with it is that is always significantly diverges from the exit polls and election outcomes in a way that cannot be explained by differences in turnout among different party supporter's.

For example, if 18-29 voters were R+5 in 1994, how did Clinton win 18-29 voters by 19 points in 1996, a 24 point gap.
Among 30-49 voters, which were R+2 in 1994, Clinton win by 9 points, an 11 point gap.
Furthermore, Dole did best among older voters, despite Pew Research having them as being the most democratic cohorts. The differences in my opinion cannot be simply explained by R leaning voters going more heavily for Perot.

Pew Research data in 1994 would have us believe the younger cohorts were more Republican and vice versa, yet in the election of 1996, the reverse occurred, younger voters overwhelmingly broke for Clinton over Dole whereas older voters split somewhat more evenly.

If we use the Pew 2014 data, we run into the same issue, if GenX were D+11, how did they split 50/50, an 11 point gap, and if Baby Boomer's were D+6, how did they go R+8, a 14 point gap.

I am not saying the Pew Research Data is wrong, just that its political affiliation data diverges substantially from election outcomes, whether in 1996 or 2016.

I am of the personal view that election exit polls are a more reliable way to gauge cohort political leaning then political affiliation surveys since the political affiliation surveys do not correlate at all with actual voting outcomes and there are always large gaps.