What bothered voters about Hillary Clinton?

Reading Time:

YouGov asked Americans in mid-2016 whether the following Clinton-related controversies bothered them:

  • "Her handling of the attack on the American consulate in Benghazi"
  • "Her refusal to release transcripts of her paid speeches"
  • "Using personal e-mail instead of government e-mail when she was Secretary of State"

Set aside whether these transgressions are really serious, or whether the media had something to do with emphasizing them.

The chart below summarizes what people told pollsters - I'm just showing the proportions of people who said an issue bothered them a lot:

Data: YouGov poll, May 26 - 31, 2016

The chart shows the share of people who answered "Bothers me a lot" (Other possible answers were: "Bothers me a little"; "Doesn't bother me at all"; "Not sure")

On any issue, about a fifth of Democrats was quite disturbed ("very bothered") by Clinton's behavior. In fact, 20% of Democrats agreed with Republicans that Clinton's handling of the Benghazi attack was a big deal. It is a fairly large percentage, and I'm not aware of anyone having investigated these Democrats' reasons for being so critical of Clinton.

Most Republicans looked at Clinton's past actions very unfavorably but they were relatively forgiving when it came to paid speeches, at least compared to the other controversies.

Overall, 63% of Americans disapproved of Hillary Clinton's email server setup. Given that 11% of respondents had no opinion, the percentage of people who didn't look at Clinton's emails skeptically was extremely low.

Personal email use: crosstabs

Responses by gender, age, party ID, and family income specifically for the question related to the use of a personal email server are below:

How America voted in 2016: A generational divide?

Half of young Americans are either conservative or moderate. But many ideological moderates voted for Clinton in 2016....

Out of sample predictions from OLS regressions: a K-folds tutorial in R

Evaluating predictions out of sample, OOS. Splitting datasets into training and test, holdout data using R....