A view from an economist: Use of data
I received a phone call from a reader, questioning my use of statistics. I thank him for his call, because in an era of social media on the one hand, and accusations against the press on the other, it is critically important to carefully assess information we receive. In the case of statistics, we need to determine:
• whether the source is credible,
• the exact meaning of the reported data,
• any possible bias underlying the statistics,
• the depth and breadth of the topic.
Since it is difficult for me to simultaneously discuss policy issues and carefully reference my data (in 700 words or less!), I think it is important for me to comment on this issue now.
• First, when I provide information comparing several countries, I usually rely on the World
Bank World Development Indicators (www.wdi.worldbank.org). For information on the U.S. alone, I typically use U.S. government data sources, including the Census Bureau (www.census.gov) and the Bureau of Labor Statistics (www.bls.gov).
While these have been viewed as credible data sources, we are currently confronted with some disturbing events. Our broadest data source is the Census Bureau, which conducts a complete census every ten years and relies on sampling data each year in-between. Donald Trump has not yet appointed a director for this agency, and this, along with uncertain funding, is constraining the use of technology vital to accurate reporting.
According to Census Bureau statisticians, these will severely hamper an accurate 2020 census (which requires a heavy lead time in the development and testing of new technology). We should note that since the 2016 presidential election resulted in more red than blue states (and more red than blue districts within many states), it is in the current White House interest to avoid a complete 2020 count.
Equally disturbing is the fact that the government web site of the Environmental Protection Agency has been scrubbed of most studies and data pertaining to climate change. Similarly, scientists in this agency have been prevented from speaking at climate conferences to which they've been invited. I have serious worries about obtaining complete and accurate information in the current political environment.
2. Aside from credible data sources, it is important to understand how statistics are calculated.
For example, the unemployment rate is calculated as the number of people who are unemployed, as a share of people in the labor force. The labor force includes both employed and unemployed people, and one must be actively seeking employment to be called 'unemployed'. (This means that unemployment rates do not include people who are full time homemakers or students, or who simply do not want to work.)
3. Furthermore, as we know, statistics can and do lie. The best we can do is to try to understand the direction of any bias. For example, the unemployment rate understate the full problem of unemployment, since part time workers are counted as 'employed' (even though many of them prefer full time employment); and because people fall out of the unemployment statistics entirely when they become so discouraged in their job search that they give up searching. These may be the people hardest hit by the unemployment problem.
On the other hand, unemployment rates may overstate unemployment if people report that they are actively seeking a job when in fact they are not.
4. Finally, one statistic (like the unemployment rate), does not tell the full depth of employment issues, since low wages create additional hardship. Nor does it cover the breadth of the issue unless we also note that the teen unemployment rate is 2.5 times the national rate, and the black unemployment rate is twice that of the white rate.
If you do have any questions about information I report in my column, please do not hesitate to contact me via email@example.com. Type "To Jackie" in the subject line, and include your full name and email address. And, since I do want to leave you with information on at least one topic, let me point out that the 2016 unemployment rate was 4.9 percent, an all-time recent low (and quite a bit lower than when Trump said, "I heard it was about 43 percent."