Anatomy of a Political Poll

Politics in Virginia

Photo by Greg. (License: Creative Commons Attribution)

A new poll has hit the phone lines of Virginia. Run by a group calling itself "Survey 2006," it will show up on caller ID as "S2006 703-961-9120." Survey 2006's polls have been criticized as openly biased, badly written, and even push-polls. This one is no exception.

The computerized voice announces itself as "Survey 2006," though at the end of the poll is notifies the participant that the poll was conducted by "The Family Foundation of Virginia." As lengthy pauses by the participant are detected by the system and may prompt a disconnection, a verbatim copy of the survey will take the collaborative work of several individuals. What follows is a fairly accurate representation of the poll, though some minor differences in phrasing and language exist.

1. Are you registered to vote in Virginia?
2. On the issue of abortion, do you consider yourself pro-life?
3. Do you agree that marriage should be legally defined as between one man and one woman?
4. [Some background on how taxes may be collected on a persons estate.] Do you support the permanent repeal of the Death Tax?
5. Do you agree with US efforts in Iraq?
6. Do you support a tax credit to allow parents to choose where their children go to school?
7. Do you support the use of unborn fetuses in stem cell research?
8. Do you support legislation that would require a filter blocking pornography in all public libraries?
9. Do you think churches should be involved in political issues?
10. Should women have the right to request anesthesia for their unborn fetus if they elect to have an abortion in the second or third trimester?
11. Do you trust the Republican Party to spend your tax dollars?
12. Do you attend church, temple, or synagogue, at least once per week?
13. Are you male?
14. Have you ever contributed to a political candidate or political campaign?

While these questions are paraphrased, specific linguistic constructions like "Death Tax" and "Unborn Fetuses" have been preserved to maintain the general tone and framing of the survey. An analysis of the language used and the subtext of questions reveals a series of problems with this poll. Such an analysis follows:

1. This question exists to create a statistical base for the poll. In this case, registered voters in Virginia.
2. The decision to use the phrase "pro-life" to the exclusion of "pro-choice" represents a frame of the abortion issue. "Pro-Life" implicates its opposition as "Pro-Death" or "Anti-Life" rather than the more typical "Pro-Choice." This phrasing may introduce some bias into the results of this survey and certainly this question.
3. Given that Virginia has an amendment on the ballot this year which deals with this issue specifically, this question is unnecessarily vague. Linguistically, however, there is little bias injected into this statement.
4. The phrase "Death Tax" is a clear frame of the Estate Tax issue and will inject substantial bias into the poll. "Death Tax" implies a more universal and regressive tax than is the case and is typically associated with a strong right-wing agenda.
5. The phrase "US Efforts" is unnecessarily vague. This question's vagueness allows the question to be interpreted widely by the pollster and renders the question itself meaningless.
6. The use of the "choice" frame makes this question suspect in much the same manner as the "Death Tax" and "Pro-Life" questions above.
7. This question uses the phrase "unborn fetuses" making setting up the assumption of birth and therefore the assumption that the fetus is in fact alive. A more honest phrasing of this question would not make this implicit statement.
8. The "porn filter" question does not use any ideologically loaded frames, but fails to recognize the technical infeasibility of such a filter. Even so, the question, if not its subject matter, remains valid.
9. The "church involvement in politics" question is well phrased, though the term involvement is somewhat vague. Little to no ideological bias will be injected by this question and it will likely yield valid statistics.
10. Despite being a very strange question, this seems to harbor no unnecessary frames save for the "unborn fetus" phrase discussed above. This question may represent a push poll of sorts, espousing the position that a fetus can feel pain and may need anesthesia in the first place.
11. The "Republican Spending" question is well phrased and ideologically balanced. It may serve as a good predictor of party loyalty.
12. The "Religious Services" question is likely used to establish trends against other questions in the survey. It exhibits no obvious bias.
13. Gender balance is essential in a poll and allows some statistical correction for social norms..
14. Contribution to a political campaign can be used to gauge the degree of political activity an individual engages in. This is crucial to the overall statistical results.

It's just another poll is the vast field of American politics, a snapshot of the political terrain. Like all photos, however, it fixes the world as seen through the photographer's eye.

Back To Top