Donald J. Trump (DJT) received roughly 63 million votes in 2016. Geroge W. Bush (GWB) received 62 million votes in 2004. At the same time the US Population has grown from 293 to 322 million people (not necessary the population eligible to vote). On a map the coalitions look similar, also the demographic composition looks similar (see (ropercenter.cornell) and here (Pew)).

Also, I'd like to exclude everyone who is younger than 30 in 2016 (the voting age is 18, everyone who is younger than 30 couldn't vote in 2004). People who are dead in 2016 but voted for GWB, are excluded by the scope of this question, because I'm only interested in one direction: DJT's voter who voted for GWB; not GWB's voter who voted for DJT.

So the question is: given a DJT's voter in 2016, what is the probability that they voted for GWB in 2004 (based on polling, statistics, voting registers, ...)?

A starting point is this analysis between Barack Obama and DJT (Rasmussen). The bottom line is, that Barack Obama received a lot of votes and roughly 10% voted for DJT, but "it would not be surprising if Obama’s percentage were somewhat inflated and Romney’s a bit diminished". Is there a similar analysis between DJT and GWB?

  • 2
    This seems to be a Speculative question. It could be answerable if there are statistics to back it up, but not regarding the analysis of the possibility since we won't exactly know.
    – Panda
    Commented Jul 24, 2017 at 11:37
  • 1
    Feel free to edit the question to improve it; but why is it speculative to ask if the voting behaviour has changed between the last and the current republican presidents? in fact, this seems like to be an important question.
    – Guest1232
    Commented Jul 24, 2017 at 11:41
  • The issue is the wording of the question. If you ask for polls that explored that relationship it is ok, if you ask us to opinate about how likely that relationship is then it is off-topic. As you express it, it seems the former.
    – SJuan76
    Commented Jul 24, 2017 at 11:47
  • 2
    I don't have an answer to this, but I suspect that FiveThirtyEight has the answer somewhere.
    – Bobson
    Commented Jul 24, 2017 at 13:00
  • 1
    Trende and Byler of RealClearPolitics write about how the county level results have shifted. The challenge with individuals is that that information is confidential. So at best we'd have polling data that might be incorrect (people remember their votes differently than they cast them; or just lie to pollsters).
    – Brythan
    Commented Oct 5, 2017 at 20:08

1 Answer 1


Since the actual (non-provisional) ballots cast are anonymous, it's not possible to answer the question with anything other than a guess based on polls or surveys. And, since the pre-election polls did not look much like the election results, we can all surmise that the accuracy regarding the shared/overlap of voters would be incorrect as well.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .