And the answer is… we think… more or less…
As the presidential election moves closer and closer, voters have questions that need to be answered in a clear, concise and unambiguous way. For example, how many jobs are being created? Or, getting down to the election itself, which candidate is likely to win, President Obama or Governor Romney?
And it’s not just at election time that questions come up. Businesses are interested in such issues throughout the year as they are trying to determine what production should be or how many people need to be hired.
So, with regard to the question above as to how many jobs were created in September, the answer is 873,000. No, no; wait, that’s not right. The answer is 114,000.
Perhaps that one is too complicated. So, how about the presidential race here in Ohio? Who is ahead in the polls at this time and seems likely to win our state’s electoral votes? According to the Quinnipiac/CBS News poll, the president is nicely ahead — it would seem — with a lead of 50 percent to 45 percent. No, no; it’s closer than that according to the Fox poll, with Obama at 46 percent and Romney at 43 percent.
So, what gives here? Numbers seem to be all over the place, no matter if we are examining the strength of the job market or the voting habits of Ohioans. How can there be such a wide divergence of numbers that should produce a definitive answer? Is somebody simply lousy at doing their job and we get a bunch of flakey figures?
No, actually the differences are a function of the means by which data are gathered in this country (and all others) given limited financial resources to collect useable information.
Consider the issue of how many people is this country have jobs and how many new jobs are created in a particular month. Obviously, there is a definitive number for each of the above questions. But since the U.S. has (as best we can determine) more than 155 million people who want a job, it is impossible to contact each one of them every month to inquire about their employment situation. As a result, each month (as discussed in previous columns) the government conducts two separate random samples, one of approximately 60,000 households and the other of approximately 400,000 individual business establishments. From those two survey-based research efforts, they infer from the specific sample(s) to the general population. While there certainly are definitive answers as to how many people have jobs (that figure being called a “population parameter”), the survey-based research produces a “sample statistic” which will, hopefully, estimate well the unknown population parameter.
Of course, we don’t and won’t know what the precise figure is (the population parameter), so each sample statistic represents a methodologically reasonable best-guess. But since all random samples of the same size from the same population cannot be expected to produce identical estimates, there is a level of confidence and a margin of error that must be considered due to variability which may exist from one sample to another. As such, it is always important to note the confidence level utilized in the sampling work (typically set at 95 percent) and the margin of error for any particular sample size (for example, plus or minus 3 percentage points).
So, in the presidential contest in Ohio, if a particular poll says Obama leads Romney (or vice versa, as I don’t want to show partiality) by 50 percent to 45 percent, with a margin of error of 3 percentage points, President Obama’s support should range between 47 and 53 percent (that is, 50 plus/minus 3 percentage points) in 95 cases out of 100 (producing a 95 percent confidence level). Similarly, the level of support for Governor Romney would range between 42 and 48 percent. Since the two sets of support figures overlap within the “margin of error,” the race is considered a statistical tie.
Virtually all data, whether measuring support for candidates or the state of the economy, are gathered in this fashion and represent a best estimate of some population parameter. The process does not involve a fudging of the numbers (such as some “Chicago guys” manipulating job figures for political advantage), but rather an honest attempt to estimate some important variable. It may not be precise, but the process tends to produce useable results… results that are certainly better than nothing when helping someone in a decision-making process.
Dr. James Newton serves as chief economic advisor to Commerce National Bank and is an auxiliary faculty member in economics and statistics at OSU-Marion and OSU-Newark. Dr. Newton’s views do not necessarily reflect those of Commerce National Bank or OSU-Marion/Newark.