Brownback winning by large Margin in Kansas (59-32) SUSA

[Hash]

http://www.surveyusa.com/client/PollReport.aspx?g=d6163984-1a7f-4902-b4d3-f003d08c7276

It's refreshing to see a SUSA poll where Black voters actually seem reasonably in line with their actual voting record.  The partisan breakdown might be a little much however, with 57% of Likely voters being Republican.

It seems that you didn't listen when we told you to stop looking at crosstabs.

[Hash]

http://www.surveyusa.com/client/PollReport.aspx?g=d6163984-1a7f-4902-b4d3-f003d08c7276

It's refreshing to see a SUSA poll where Black voters actually seem reasonably in line with their actual voting record.  The partisan breakdown might be a little much however, with 57% of Likely voters being Republican.

It seems that you didn't listen when we told you to stop looking at crosstabs.

Why?? I think looking crosstabs is positive for people who want to know everything about the poll. Imagine, if Kirk is leading gianoulias by 5% in a poll, but is carrying 40%of the black vote, should I think kirk is leading??

Because categories like "black voters", "18-29" which are a tiny part of the sample have a gigantic margin of error and provide no serious statistical information. Only subsamples that involve the whole sample or something or a good chunk of it are useful.

[Hash]

http://www.surveyusa.com/client/PollReport.aspx?g=d6163984-1a7f-4902-b4d3-f003d08c7276

It's refreshing to see a SUSA poll where Black voters actually seem reasonably in line with their actual voting record.  The partisan breakdown might be a little much however, with 57% of Likely voters being Republican.

It seems that you didn't listen when we told you to stop looking at crosstabs.

Why?? I think looking crosstabs is positive for people who want to know everything about the poll. Imagine, if Kirk is leading gianoulias by 5% in a poll, but is carrying 40%of the black vote, should I think kirk is leading??

Because categories like "black voters", "18-29" which are a tiny part of the sample have a gigantic margin of error and provide no serious statistical information. Only subsamples that involve the whole sample or something or a good chunk of it are useful.

I agree. But, SUSA polls always show republicans crushing democrats with young people and picking up 20-30% of afroamericans.

Thanks for giving me an example of a reason why we should ignore these types of subsamples.