Thursday, October 6, 2022

Is polling and related media coverage artificially inflating Gov. Abbott’s ratings in the Texas gubernatorial race?


The polling industry and the public recognizes that polls are imperfect measures of political events and outcomes, and sometimes can be dead wrong.  The 2020 presidential election is a good recent example of the extent to which Donald Trump’s ratings were significantly under-estimated, although no consensus has emerged about the exact reasons for these polling inaccuracies.  The unwillingness of Republican voters to respond to polling was one factor that seemed to gain more acceptance in the polling industry.

Interestingly, polls surrounding statewide and municipal elections do not appear to capture the same level of scrutiny as national polls even though they are also susceptible to various sources of bias.  The race for governor in Texas is a good example of how media and polling activities can combine to artificially elevate the standing of a favored candidate -- i.e., Gov. Gregg Abbott.  Let’s first discuss some two potential sources of polling bias, then follow with a discussion of potential media bias surrounding this campaign.

Potential Polling Bias

The most recent poll was conducted in September 6-13, 2022 by The University of Texas at Tyler Political Science Department and co-sponsored by The Dallas Morning News. [i]  Methodological information about the poll is presented on this web site which shows that 1,268 interviews were completed with registered voters using a sample derived from two different sources: a panel used for on-line surveys and a telephone listing used for telephone interviews.  Both English and Spanish-language options were reportedly provided to the survey respondents.  In reviewing a survey tabulation provided on their web site, the following two findings were observed:

·         One question that asked respondents about their employment status during the last week prior to the survey showed that the proportion of retired White respondents (42%) was twice as high as Black (21%) and Latino (15%) respondents. Compared to other national polls like the General Social Survey, this proportion of White respondents is unusually high, and may artificially inflate the poll’s rating for Gov. Abbott whose support comes primarily from White voters.  Whether due to actual interviewing or weighting adjustments, why were retired White respondents given so much weight or presence in this poll?

·         Potential Sampling Bias: The selection of poll respondents from two differences sampling sources or frames points to a potential sampling bias. Despite the detailed methodological steps that described the manner that these two sources were combined into one total sample with a margin of error of plus or minus 2.9 percent,   one needs to recall that panel respondents are not usually selected at random and are usually paid incentives to complete surveys on a periodic basis, while telephone survey respondents are usually selected at random and not paid an incentive to participate in a survey.  Also, online panel members are typically younger, more educated and have higher incomes than the general population, and often exclude people of color and immigrants – a profile that could also bias the poll ratings in favor of Gov. Abbott. [ii]  Demographically speaking, these two groups of respondents are apples and oranges. Given this information, it would be helpful to know the proportion of Black and Latino poll respondents who were selected from a panel or the telephone listing to verify this potential sampling bias and its influence on the poll ratings. 

Potential Media Bias

·         A second tab on the UT-Tyler web site asks respondents to identify their main source of news, and shows that Spanish-language news – like Telemundo and Univision -- were used by an extremely small proportion (2%) of the poll respondents.  In Texas, this is highly unusual given the number of eligible immigrant voters: 52 percent are Latino immigrants [iii]  that past research confirms frequently use Spanish-language news sources.   Even though the UT-Tyler poll methodology  explains that both English and Spanish-language options were provided in the poll, it would be helpful to know the actual unweighted number of Latino surveys that were completed in English and Spanish.  Why? Because pollsters often state that language options are provided in a survey, but that does not tell us the extent to which these options were actually chosen by the survey respondents. If Texas immigrant voter participation in the poll was inadvertently minimized due to poorly implemented language options, one would expect Beto O’Rourke’s ratings to be systematically lower since Texas immigrants tend to support Democratic candidates. Again, a potential unearned benefit for Gov. Abbott.  

·         The recent televised debate between Gov. Abbott and Beto O’Rourke also revealed both logistical and media bias that favored Gov. Gregg Abbott.  To begin, Gov. Abbott’s campaign deliberately defined the debate logistics in his favor by scheduling the debate on a Friday evening that is known to attract few audiences of potential voters and further required that no live audience was present in the auditorium during the debate.  Despite these limitations, an informative debate still took place and provided viewers an opportunity to understand these two candidates in more detail. It was clear, for example, that Beto O’Rourke performed well  by focusing on the repeated failures of Gov. Abbott over the past eight years of his administration – facts that were verifiable by external sources. By contrast, Gov. Abbott focused on Beto’s “flip-flopping” on issues like gun control and alleged statements by Beto (e.g. “Beto supports open borders,” “Beto supports de-funding the police”) that were not supported by any facts presented by Gov. Abbott. Interestingly, it was Gov. Abbott who actually took millions in funding from law enforcement agencies in Texas to support his failed border wall project.[iv]  The main outcome of the debate was Gov. Abbott’s evasiveness and unwillingness to explain (a) his lack of accountability for the Uvalde massacre; (b) why he did not taken decisive action over the past eight years to avoid the recent energy crisis that cost the lives of 700 Texas residents; [c] his lack of support for gun legislation that has resulted in numerous mass killings during his administration; and (d) his support for restrictive abortion laws. Thus, the debate was a significant learning opportunity for its viewers.

Media bias, however, was evident in the follow-up report in The Dallas Morning News by Gromer Jeffers [v] who was one of the panel members that presented questions during the debate. Jeffers concluded based on insights from other experts that the debate outcome was fairly inconsequential and not likely to change the minds of Texas voters unless Beto was able to muster strong voter turnout – which would have more significance on Election Day than “throwing jabs at a debate.”  While one cannot argue about the value of voter turnaround, it seems unwarranted, and perhaps a sort of media bias, to describe the recent debate as “jabs” thrown at a debate.  These “jabs” were very informative in distinguishing the two candidates:  Gov. Abbott as the incumbent who blames others for his failures over the past eight years vs. Beto O’Rourke who voices strong support for policies that will improve the quality of life for Texans.   

In conclusion, the Texas Poll results suggest that Beto O’Rourke continues to face an uphill battle in beating Gov. Abbott in this highly competitive race.  The media’s consistent coverage regarding the polling advantage that Gov. Abbott currently enjoys is clearly benefiting his campaign but also likely to discourage voter turnout by supporters of Beto O’Rourke. But history tells us that polling results are not infallible and sometimes undeserving of the confidence accorded them by media sources. While there is some reason to believe that the Texas Poll ratings for Gov. Abbott may be inflated, there does not appear to be much concern about obtaining an objective, independent critique of the methodology used.  Beto O’Rourke’s campaign has been consistently on-target in documenting the failures of the Abbott administration over the past eight years and is aggressively reaching out to various segments of potential Texas voters.  By contrast, Gov. Abbott’s well-funded campaign is driven by advertising messages that are false and designed to portray Beto O'Rourke as a governor who will wreck the Texas energy industry, worsen border security problems, and promote a socialist agenda.  If the recent Texas Poll ratings have been artificially inflating Gov. Abbott’s standing and Beto continues his successful strategy of relying on the facts and focusing on voter turnout, Texans may witness an unexpected surprise this November.   

It should be noted that these concerns for potential methodological biases in the Sept. 2022 Texas Poll were emailed to the poll director on Oct. 3, 2022 using the email posted on the UT-Tyler web site. No response has been received at this writing.

 

Reference Notes


[i]  UT Tyler Political Science Department, Center for Opinion Research.  Accessed at: https://www.uttyler.edu/politicalscience/pollingcenter/ 

[ii] Rincon, E.T. (2020). The Culture of Research, Publisher: The Writer’s Marq.

[iii] Noe-Bustamente, L. and Budiman, A. (2020 March). Most of the 23 million immigrants eligible to vote in 2020 election live in just five states. Pew Research Center, Accessed at:  https://www.pewresearch.org/fact-tank/2020/03/03/most-of-the-23-million-immigrants-eligible-to-vote-in-2020-election-live-in-just-five-states/

[iv] Kriel, L. and Trevizo, P.  Texas has spent billions of dollars on border security. But what taxpayers got in return is a mystery.  Texas Tribune, Accessed at:  https://www.texastribune.org/2022/04/18/texas-border-security-spending/

[v] Jeffers, Jr. G.  (2022 October).   Abbott, O’Rourke near finish line. Accessed at:  https://www.pressreader.com/usa/the-dallas-morning-news