I guess it depends on who you read

SurveyUSA polls monthly on the gubernatorial approval ratings for all 50 states. In Texas, these polls are paid for by KEYE (Austin) and WOAI (San Antonio).

Here's KEYE, since I couldn't find the story on WOAI's site:

In a new, exclusive CBS 42 News poll by Survey USA, Texas Governor Rick Perry's job approval rating has slipped slightly since mid-June, but his numbers are still looking good compared to where he was a few months ago.

In the poll, taken July 20, Perry registered a 47 percent job approval rating while 48 percent said they disapproved. That's down from a 51-45 spread in mid-June.

Perry is still well ahead of the 40 percent approval ratings seen in polls for most of 2006.

Perry's strength does not bode well for his three opponents' chances heading into the full-blown campaign season.

Here's RG Ratcliffe in the Houston Chronicle:
A new poll conducted for a pair of Texas television stations has found Gov. Rick Perry's job approval rating declining and nowhere more than in Harris County.

The county has become a particular weak spot in the governor's re-election campaign, largely because of conservative anger over his inability to pass property tax appraisal caps and for adopting a business tax to pay for property tax cuts.


That's a drop from the June poll when Perry's approval rating was 51 percent positive and 45 percent negative.

But that June poll also was done shortly after Perry had saturated the state with television advertising touting that the average Texas homeowner will receive a $2,000 property tax cut under legislation passed in a special session he had called.

The June poll found 55 percent of those surveyed in Harris County approved of Perry's job as governor. The July poll found that number had declined to 35 percent approving and 65 percent disapproving.

Those are certainly two divergent views, aren't they?

The "poll" is by Survey USA, done 7/14-7/16 of 600 adults (not registered or likely voters), +/- 4.1%. Many pollsters don't like SUSA, as they are automated surveys, not polls. As I've written before, I think these concerns are largely unfounded. SUSA has a good track record.

I have two big problems with Ratcliffe's story:

1. His story focuses on the fact that Perry's Harris County support has dropped precipitously. The problem is that this was a subsample, and thus not very reliable. Harris County contains about 16% of Texas population. Thus, if 16% (or even 20%) of SUSA's poll was from Harris county, then only about 100-120 were surveyed.

You know where I'm going with this right? The margin of error on a subsample of 100-120 is huge. The margin of error is so big that

However, it's even more complicated than that. Because SUSA is a survey, not a poll, it's possible that the survey size could have fluctuated. Given what I know about their surveys, I don't think they ensure that they get a proportional subsample. So really, the subsample probably could be anywhere from 50-200.

How reliable is the Harris County subsample? Frankly, we really have no clue. And yet Ratcliffe didn't even mention this in his article.

2. Here's my other problem: even if you assume that the Harris County numbers are statistically valid and significant, it is pure conjecture on Ratcliffe's part that Dan Patrick, the business tax, and appraisal caps are the reason. Now, I think he might be right, but he has no evidence to back that up. He hasn't polled it. He's just taking an educated guess.

It's a nice storyline that Patrick has so much influence that he's stirred up a hornets nest. It's probably true, but since Ratcliffe has no proof, shouldn't he at least mention the possibility that (if there is a drop) it might be for other reasons?

To Ratcliffe's credit, he did add a short sentence mentioning that the survey was simply of adults, and not likely voters. He didn't explain that this is likely to understate Perry's electoral support.

In short, did Perry's approval go down in Harris County between mid-June and mid-July? Probably. But we don't know how much, because this subsample is so small and unreliable that it's not worth writing a story on. Unfortunately, Ratcliffe didn't even mention that.

Posted by Evan @ 07/22/06 12:38 PM


Previous Entry | Home | Next Entry


Can you explain what you mean when you say SurveyUSA is conducting "just surveys, not polls?" It's not a statement I've encountered before.

Posted by SurveyUSA @ 07/24/06 11:52 AM

Sure. Polls are weighted. Surveys are always reported just as the data comes.

Posted by evan @ 07/24/06 12:13 PM

Thanks, Evan. SurveyUSA polls are, in fact, weighted, and never reported "raw."

SurveyUSA’s complete statement of methodology can be seen <a href="http://surveyusa.com/method...">here</a> on our website.

Posted by SurveyUSA @ 07/24/06 12:51 PM

what crap surveyuas has 49% of texas voters in some region labeled east texasm not including Harris, and not a region labeled south texas. plus their demos are not at all on the mark. Bell will be in second place.

Posted by hamiltonfan @ 07/28/06 07:24 PM

Add Comments

No flames or impolite behavior. HTML will be stripped. URLs will be transformed into hyperlinks.