The information age is in full bloom. Rapid developments in technology let us store, retrieve and analyze data — available, literally, at our fingertips — at an unprecedented rate. Academic journals in all disciplines, newspapers, websites and cable television bombard readers and viewers with statistics and statistical analysis. Opinion polls tell us how we think or feel and who will win an election before the actual voting takes place. As reliable or unreliable as the flood of statistical information may be, it inevitably forms the basis for countless important decisions.
Politicians the world over have become adept at manipulating numbers, either to pass important legislation or to avoid imposing politically unpopular policies such as tax increases (now often labeled as "revenue enhancers," falsely implying there is a difference).
People tend to put great faith in numbers. Numbers seem to represent indisputable fact. But as the saying attributed to Benjamin Disraeli goes, "There are three kinds of lies: lies, damn lies and statistics." Numbers themselves do not lie, but they can be manipulated, consciously or unconsciously, to distort or invent facts. Thus, when confronted with statistical data, it is important that people interpret the information from the proper perspective — aware of the risks inherent in accepting numbers as truths.
A popular method of presenting statistical information, particularly in the fast-paced world of cable television, is through graphical analysis. A picture or diagram may be worth a thousand words, but there is no guarantee the words are accurate or truthful. Diagrams, charts, and graphs are utilized to "help" the viewer visualize numbers. But a diagram designed to show an increase of 25 percent may be drawn with the new value four times the size of the original value, thus giving the reader the impression of a much greater change over time.
Another device is to show the axes of a diagram with incomplete labeling, so the reader is not able to determine the dimensions of the data. While a graph may appear to show large differences between two lines that represent spending, for example, the actual values may be very small.
To take an example: In the latest Republican "Pledge to America," a graph is presented purporting to show average government spending as a fraction of gross domestic product under recent administrations. What is immediately noticeable is the bar graph showing the size of the budget under President Barack Obama is twice as high as the two previous administrations. It also appears to indicate that spending under President George W. Bush was less than that under President Bill Clinton.
Closer inspection, however, reveals that the increase in government as a share of GDP looks so large because the Y-axis is only labeled with values ranging from 17 percent to 24 percent. While this represents the share of the budget, it is not obvious — and the small scale severely exaggerates the size of the increase.
The viewer's ability (or lack thereof) to comprehend and interpret data correctly may sway his or her opinion. The larger ramifications suggest that graphs such as these may be used to affect public opinion in a way dangerously close to downright deception.
The current crop of presidential candidates severely distort the concept of cause and effect, assuming that if two things happen around the same time, one must have caused the other. For example, Republicans often cite President Obama's policies to explain the economic downturn, despite the fact that unemployment began its upward trend in late 2007. Most economists and statisticians, including the nonpartisan Office of Management and Budget, correctly note the major contributing factor was the bursting of the bubble economy and corresponding financial crisis. The misleading interpretation of statistics, however, plays to an audience more interested in hearing what it wants, rather than the facts.
The Democrats, however, are not immune to this type of behavior. They have been known to present overblown estimates of budget savings and exaggerate the effectiveness of proposed polices by only presenting the results of highly unlikely best-case scenarios. One of the more blatant examples is the Democratic claim that the new health care law will reduce the deficit and "save taxpayers" more than $1 trillion over 20 years. No evidence is used to support this statistic. The Congressional Budget Office has pegged the savings at only $210 billion over 10 years and warned that estimates beyond a decade are "more and more uncertain." Furthermore, the law raises taxes to pay for much of its new spending, so saying it "saves taxpayers" anything is misleading.
My purpose here is not to cast doubt on all uses of statistics. Statistical information can prove extremely useful in simplifying complex phenomena, yet at the same time it can be misleading. An inaccurate reading of such information can unfairly alter a political election or dangerously affect government policies. It is vital, therefore, to approach statistics with caution and acumen. Unfortunately, until the public is made aware of the margin of interpretation inherent in statistics, such data will inevitably be accepted by many as unquestioned fact.
Dennis C. McCornac is a visiting affiliate assistant professor at Loyola University Maryland's Sellinger School of Business. His email is firstname.lastname@example.org.