Discussion of most political matters involves numbers, but the debate over the health care reform bill has presented us with a huge discrepancy in numbers such as “how many people support repealing the bill?” and “how much will the bill cost or save?” as presented by different sides. Some numbers result from polling, others from government agencies, and still others from demographics and other statistics.
How is the average American to know which numbers are more accurate than the others? Which ones can be computed exactly and which ones can never be more than estimates? How are the statistics computed and what potential errors exist? Does everyone need courses in statistics, political polling, and accounting to truly understand?
Naturally, a person who has background in one or more of those areas has an advantage over someone who does not, but the rest of us can improve our interpretations of all these numbers with knowledge of a few basic concepts.
Numbers that result from polls, such as the percentage of the population that wants to repeal health care, can be determined in more ways than one. For example, a person could ask all of his/her friends what they think. Most Americans know that is not a valid way of polling, since the result is biased in favor of the individual’s own beliefs. Similarly, a poll that is restricted to a certain group will be affected by the self-selection of group membership (think about asking the health care repeal question at a rally held by the Tea Party). That is why the phone-in or text-in polls created by news networks do not really mean much. The sample consists only of people who watch a certain news show.
In order to give truly valid results, a poll must be conducted with a randomized sample. For instance, a computer program can be designed to select phone numbers at random from all the phone numbers in Texas, or even in the entire United States. Then pollsters call these numbers and ask the questions of interest. Sometimes even this method can result in misleading results, since respondents must agree to answer the questions, but it is one of the best methods available to those who seriously want to find the truth.
In the course of discussion about the health care bill, numbers regarding the financial effects of the bill were bandied about. Some said the bill would result in extra costs which would then be transferred to taxpayers, while others said the bill would actually save the government money. Most of the time numbers like this are meaningless since there is just no way of predicting them, considering the many factors that are involved. One might as well attempt to predict the value of the S&P six months from now. Any stock trader will tell you it’s impossible. An estimated value can be determined, but there are huge margins of error because we just don’t know what will happen between now and then.
The best prediction that can be made about a new government program is that it will cost more money – possibly just a little more, possibly a lot – because very few programs have been known to actually decrease costs.
When demographics are used only in a descriptive fashion, the information obtained can be very accurate. The Census conducted every decade produces an excellent resource for many scientific studies. In between, smaller studies are carried out with the same purpose of describing the characteristics of a certain population. Although demographics always fall behind reality (during the time in which the data are collected and interpreted, the population has already changed), they represent a valid way of presenting facts about a population as a whole.
Problems may occur, however, when data about smaller groups is extracted from the larger census or from a randomized polling study. At the highest level (the population or the random selections), the study may be sufficiently randomized, but lower levels such as “all blacks” or “all Christians” may not be randomized. It is a good idea to check the source of the numbers, regardless of whether they are demographics, polls, or financial predictions. This is particularly true in the health care debate, which still rages on despite having been signed into law. Legislators and their constituents should take a long look at the numbers associated with health care reform to find out where, when, and how they were obtained. With that information, one can make a determination of which side to support.