Five common mistakes business leaders make in statistical analysis

H.G. Wells was ahead of his time in 1903 when he wrote, “Statistical thinking will one day be as necessary for efficient citizenship as the ability to read and write!” And he was right. Today, some basic statistical knowledge is essential to running a staffing business.

With a deluge of data from government and other organizations, staffing firms are bombarded with market research and industry statistics. But these statistics can be ambiguous and require explanation. Often, staffing firms are stumped by the numbers and make errors. Here are some common slip-ups made while reading statistics.

1. Confusing correlation and causation. This is the most common mistake. When a report says two things are correlated, it doesn’t mean one causes the other; it merely means an association exists between the two. For example, when a report says unemployment reduction and economic growth increase are correlated, it means that there is an association between them, but one does not necessarily lead to the other.

A few years ago, Bloomberg ran an article warning of this same mistake, using several correlations to illustrate how easy it is to conclude causation. One was whether Facebook caused the Greek debt crisis, with a graph depicting an increase in Facebook accounts at the same time the debt crisis was growing in Greece.

2. Reference category. Misunderstanding the reference category is another big mistake, especially while reading the trend patterns and growth rates — year over year or month over month.

For example, the GDP growth reported by bureau of economic analysis for the first quarter 2015 is -0.7%, which may lead to the conclusion the US economy is shrinking. But a closer look at the definition of GDP says quarterly estimates are expressed as seasonally adjusted annualized rates, and it’s estimated as a quarter-to-quarter change. So it’s relative to the prior quarter GDP, which in this example grew 2.2%; also, the change is annualized. But the information on the baseline or the reference category is buried in the body of the news release.

The lesson here: Do not just read the title of an article and make conclusions, as titles give only high-level assumptions.

3. Know your metrics. Lack of understanding about metrics can also lead to false conclusions. For example, the leading economics index (LEI) and lagging economic index are different, as the former is a composite average derived from a set of leading indicators while the latter is derived from a set of lagging indicators and they each measure different economic events. So an understanding of what the metrics involve is very important before using them to make decisions.

This is applicable when interpreting survey results. So try to find the exact questions asked by the agency issuing the indicator and check for possible ambiguity in wording.

4. Mean vs. median. Another mistake is in interpreting the summary statistic, mean/average. An average is meaningful when the parameter in the population is normally distributed. In Staffing Industry Analysts’ Pulse survey, the percentage year-over-year growth reported by five companies might be 7%, 8%, 9%, 10% and 11%, for a mean of 9%. But with a skewed sample — 7%, 8%, 9%, 36% and 90% — the mean of 30% is not a good measure of the center. In this case, the median is the best measure, which is the middle number after you order them. But in cases where the data is skewed to the left, the median may deflate measure.

So when reading reports, look for both statistical measures, mean (also the center but derived after adding numbers involved) and median (the middle), to get a good grasp of the situation.

5. Overinterpretation. Some people are tempted to extrapolate a trend beyond the range of data. A monthly prediction based on the past several months’ data is a good statistic, but predictions made too far in the future may be inaccurate. For example, SIA’s US staffing industry forecast for 2015 provides useful guidance for planning revenue growth and budgeting in the current year, whereas our forecast for 2016 should be used as a directional tool, as long-term trends are subject to change.

So having a basic understanding will help staffing firm owners and managers to steer their businesses in the right direction.