[rayiner]

Here is a post-George Floyd poll: https://docs.cdn.yougov.com/86ijosd7cy/20200611_yahoo_race_p...

A plurality of African Americans (38-31) oppose cutting police budgets. African Americans are split 50-50 in whether we need more or fewer police on the streets. A supermajority (64-33) believe that the current police departments can be reformed.

A majority (51-17) support spending less on police and increasing funding for social programs, but try to reconcile this with the statistic above, where half want more police on the streets. (People might perceive this question as reduced budgets would hit management, etc., rather than beat cops).

[zpallin]

Nice. However, since yougov experiences sample bias due to their data collection method being only online participants, and the sample size of black people isn't anywhere near even 5% margin of error (meaning it could be completely wrong) this study is not usable to draw conclusions on its own.

[JamesBarney]

The sampling bias would have to be very large to throw off those numbers with 140+ African American participants.

Do you have a better poll that shows different numbers?

[zpallin]

I don't but that's not really my point -- the poll itself isn't a problem, it's the lack of multiple polls that cautions me to draw concrete conclusions from it as rayiner did. 140 people out of 30 million is about 10% margin of error. Add sample bias to that and this poll alone is nowhere near conclusive, although as I indicated it's still useful to reflect on the issue.