Spin Doctors

By Patrick Maxwell

We have been saturated these past months with gloomy statistics relevant to our economic crisis. When I think of mathematicians, my mind’s eye, however prejudiced, conjures sturdy men in permanent press slacks and sensible women in knee-length skirts, their movements and manners measured and deliberate. They labor as actuaries at insurance companies, calculate launch windows for satellites at the Jet Propulsion Lab and govern the fate of a fair number of farms across America.

Govern the fate of what?

Statistics is the science of the systematic collection, organization and mathematical analysis of information in order to present and/or interpret it in significant and diverse ways. To be fair, a statistician must master substantially more than mathematics, but, when all is said and done, it’s always the numbers we remember, not the process. There’s another component of statistics, a method of presenting only selected slices of data from the loaf prepared by the statistician or presenting the intact loaf on either a gilded platter or a moldy peel to change minds or empower agendas. This is called spin. Spin has reached epidemic prop ortions in our world culture, embellishing and modifying our perception of nearly everything from warfare and economics to sports and relationships. The word is used as a noun and a verb, and a person who spins particularly well is called a spin doctor, the best of whom invariably end up as White House press secretaries or spokespersons for oil or tobacco conglomerates.

For example: Let’s use unemployment figures from 1933 and 2008. In 1933, 13.5 million workers, nearly 25 percent of the work force, were unemployed. On a particular day in 2008, 11.6 million workers—about 7.6 percent of the total—were out of a job. If I’m the secretary of labor and the administration tells me no more money will be allocated for extending unemployment benefits, I might spin these statistics in a positive way by saying Labor Department statistics out today show that the percentage of unemployed workers is less than a third of what it was in the depths of the Great Depression. On the other hand, if I’m a columnist for The New York Times and I want to underscore the bleak state of the economy, I could write that the swelling number of unemployed workers in the United States is fast approaching, and could soon surpass, the 13.5 million in 1933. The confusion is caused by failing to emphasize that the work force being analyzed had nearly tripled during the 75 years between the calculations. Still, both statements are based on the same statistics and are factually correct; yet—excluding an analysis such as this—it’s unlikely you’d ever see them in the same paragraph.

Naturally, agriculture is not immune to this process of spin that spawns seemingly conflicting statements from the same bed of statistics. The release in February of the 2007 Census of Agriculture presented the media with some quick facts to aid in any rushes to judgment that might be needed. Tables under the economics tab on the census home page revealed market value of farm products up 28 percent, government payments up 22, farm-related income up 79 percent and net cash income up a whopping 84 percent since 2002. No news anchorperson would resist spouting those tender morsels, and most urban consumers would eat them right up without validating the ingredients. And, those seemingly innocuous numbers will probably end up playing a major role in shaping farm policy in the new Obama administration, their arrival coinciding with huge projected budget deficits due to legacy debt and freshly passed bills for economic stimuli. While Obama had already pledged to target farm subsidies and direct payments for big cuts long before his actual election, these figures will certainly increase the President’s clout as he prepares to battle with Congress. Predictably, accounts in every type of media related to wealthy farmers and nonfarmers alike feeding at the USDA pig trough have increased exponentially.

Meanwhile, a significant number of farmers who are struggling with their finances are trying to reconcile those same statistics with what’s really going on in their own barns and fields. They could download the entire census and use the searchable database to tease out statistics that reflect their atypical private misery, but, apparently the USDA itself is not insensitive to the fact that aggregate statistical analysis of farms can be misleading or, at the very least, imprecise. While reading a periodic report prepared by the Economic Research Service (ERS) of the USDA titled “Structural and Financial Characteristics of U.S. Farms: 2004 Family Farm Report,” I found chapter two, “Farm Household Income, Farm Structure and Off-Farm Work” by Erik O’Donoghue and Robert A. Hoppe, who are frequent contributors to this report. Using data collected for 1997 through 2001, the authors presented a buffet of charts and tables that revealed financial trends in farm household income infinitely more specific than a 20-second sound bite the nightly news could ever hope to offer. Just as important as the tables and charts was the text of the report, which explained how some figures in a particular table might not have accurately reflected current trends due to gross mathematical averaging, and then the authors were likely to recast the table using more specific parameters to show how profoundly the results can vary when you don’t stir everything into the same pot.

Just to be sure, though, sometimes the best way to get a point across is to exaggerate conditions to an absurd level. Suppose we consider the yearly net cash incomes of five fictitious farmers. Two of the farmers had a net loss of $20 each in farm income, one broke even at zero, one came ahead $40 and the fifth netted $800,000. Statistical analysis declares the mean (or average) net income to be $160,000 per farmer in this case. As you might imagine, farmers one, two, three and four would likely reject that notion, and, who knows? Number five might be insulted as well, having his success so misrepresented. So, let’s compute what’s said to be the more precise statistical measure: the median income of the five farmers. The median calculates the middle value in a distribution of numbers, above and below, which there exists an equal number of values. The median net income for the five farmers is zero dollars: two of the farmers make less than zero dollars and two make more than zero. While this calculation is more accurate for more of the farmers sampled, it hardly reflects the true situation on the ground.

The report by O’Donoghue and Hoppe, which is far narrower and easier to digest to make my case than the 2007 census, certainly didn’t reveal gross misrepresentations like my example, but it is a convenient showcase for how aggregate or individual statistics plucked out of context and spun by journalists or politicians could have a thoroughly altered effect. A crucial aspect of this report was the division of farms into seven family farm typology groups. The first five groups are categorized as small family farms and the first three of those five concede that farming is not their principle occupation. These are limited resource farms, retirement farms and residential/lifestyle farms. The remaining two groups of small family farms are listed as farm occupation/low sales and farm occupation /high sales. The sixth group is large family farms and the seventh is very large family farms. (For the record, the 2007 census often uses these same seven groups plus an eighth: nonfamily farms. The 2007 census is the first farm census to try to extensively count the first three groups of small farms, farms on which farming is not the principle occupation. As a result, 2007 census farm counts are not directly comparable to earlier years.) To show how small changes in what a statistical sample considers can modify the results, i used a chart from the report that tracked how the change in mean off-farm income of selected commodity groups ranged from slight to drastic just by adding or subtracting the residential/lifestyle farm group. At an average of $86,947 for all farm commodities in 2001, the residential/ lifestyle group had by far the highest average off-farm income in the report. However, o’donoghue and hoppe showed with their graph that that group, because they were part-time farmers, engaged disproportionately in different types of farming according to how much time was required. Thus, if you consider raising cattle, which is one of the most convenient part-time farm types where residential/lifestyles are 48 percent of the total, mean off-farm income rises from about $32,000 if they are left out of the equation to just over $60,000 if you factor them in. Talk about the tail wagging the dog. On the other hand, if you consider dairy farming, with its huge time and labor commitment where the residential/ lifestyles make up only 7 percent of the total, their off-farm riches only bump the mean from about $18,000 when they’re not factored in to $20,000 when they are.

I can only hope that these examples will encourage anyone who reads or hears a statistic that just doesn’t seem logical to be more skeptical, to analyze how the statistic was calculated and by whom, and to especially reflect on who delivered it and in what arena . And, if I’m really lucky, perhaps I’ve helped some farmers realize that they’re not really just average after all.

The author, a new contributor to Farming, is a lifelong farmer and occasional writer from Newport, N.Y.