Civil liberties vs. greater security-not a zero-sum game. Difficulties in polling the issue.
Last week CNN observed that Americans may soon be asked to choose between giving up some of their civil liberties for greater security. This is what is known as a zero-sum question. Either get more security while giving up civil liberties or hold on to one’s civil liberties and get less security. Either/or.
Now it should be apparent that the real world situation isn’t quite so zero-sum. It is possible to give up one’s liberties and still not get security or conversely not give up those civil liberties and still get the security. And there are a whole slew of gray steps in between, such as giving up just a few civil liberties and getting the security needed, or holding on to one’s civil liberties and getting only slightly less security. Obviously there are many intermediate shades of gray possible. But the issue is often, as in this example, posed in terms of a zero-sum game. Which, of course, is deceptive and can obscure the very existence of all the shades of gray that are possible choices.
If this musing by CNN were the basis for a public opinion poll, it is easy to see how the very wording of the question could and would likely affect the results. Security is so basic a need that giving up an abstract concept, as civil liberties to most Americans, might be a relatively easy choice.
Now imagine that the same question were differently worded, say, for example, if the respondents were given the choice of having the possibility of greater security against yielding the protections against encroachments by government granted by the our founding fathers in the Bill of Rights. A different result could be expected. Even the most conservative Tea Party type respondents, might cringe at the thought of government restricting their right to carry weapons or to limit their right to free expression of their views. Adding the sacrosanct “founding fathers” to the equation, albeit doing no violence to the truth of the matter, would likely further prejudice the outcome in a significant manner.
This all points to the importance of question construction in polling. Modern polling organizations have many tools at their disposal to accomplish good question construction, assuming that is their goal. At the simplest level, they could divide their test samples into two groups, similar in all relevant characteristics save for which wording of the question they received. One group would get the question being tested for, which is called the experimental group, the other, a neutral wording of the same question. This is called the control group. The results could then be compared to determine the effect of the question’s wording. There are many other techniques to check question validity and reliability. I gave only the one simple example to make the point.
It should be noted that polls that are not in the service of truth abound. For example Republicans might commission a poll for public release to emphasize the unpopularity of Obamacare, while Democrats might do the same to emphasize its favorable rating. But the major polling organizations, such as Gallup and Rasmussen, in their non-commissioned polls, seek the truth. Albeit even with those organizations, expediency in terms of cost and time often produce problematic polls. Next week, barring some international or domestic crisis, I’ll examine the major such polling problem, and it is a big one
For now, at the very least, be aware of the effects of how the questions are phrased and whether they are ”loaded” in such way as to slant the results. An overly simplistic zero-sum question that overlooks obvious shades of gray should pop-up a mental warning sign in your heads.