By Jeff Bebee
Data analytics and visualization might sound like arcane subjects to association writers and business people. At AM&P’s Nov. 9 Lunch & Learn, however, Jonathan Hill dissected a number of graphs to show how organizations can convey misleading information. Hill, a data analytics expert for financial services companies, then fixed the inaccurate graphs to reveal what the underlying data actually say. It reminded us all of the need to fact-check and scrutinize the conclusions we make and publish.
Americans United for Life produced a graph suggesting that Planned Parenthood is ramping up its abortion activity and decreasing its life-saving cancer screenings and preventive health services. Until you notice that the abortion number at the top right (327,000) is higher on the y-axis than the screenings (936,000).
An accurate depiction of the data does suggest that Planned Parenthood’s preventive screenings have decreased dramatically since 2006, but the number of abortions has been relatively constant. The implied correlation between the two does not exist.
The US Department of Education used stacks of books to create a bar graph showing the increase in graduation rates during Obama’s administration. Hill found two issues with this graph:
- There is no context: What were graduation rates like under previous administrations?
- Scaling is not consistent: The number of books under the 82 percent stack is too high compared to the number of books at 75 percent.
The corrected graph shows that graduation rates have increased over the years and reached a high during the Obama administration. The inflection point occurs at the end of the Clinton administration and beginning of the Bush administration, suggesting that the increases in graduations could be due to policies set in motion before the Obama administration.
The National Review dubbed this “the only climate change chart you need to see.” It certainly looks like global temperatures are not increasing. Hill indicated that the scale of this chart is too large to notice any changes in either direction.
A good rule-of-thumb is to use a zoomed-in scale where all-time highs and all-time lows are at the top and bottom of the y-axis. The corrected graph does indeed show that global temperatures are increasing.
This pie graph seems to indicate that 57 percent of federal spending is for military, and that less than 1 percent is for food stamps. The conclusion is that military spending, not food stamps, is a bigger factor in the budget deficit. The misleading element is that only discretionary spending is shown in the pie graph, representing only one-third of federal spending. The remaining two-thirds is called mandatory spending, which includes Social Security, Medicare, Medicaid, and interest on the national debt.
When you add both discretionary and mandatory spending, you see that Social Security and health costs (about 50 percent) are a much larger portion of the budget than Defense and Homeland Security (16 percent). The ratio of defense to food and agriculture (wherein lies food stamps) is 4:1, not 57:1 as suggested in the pie graph.
You can contact Jonathan Hill at firstname.lastname@example.org.
Jeff Bebee was recently marketing director for the American Institute of Physics and serves on the AM&P Board of Directors.