The US contingent workforce appears to have grown less than first estimated, according to a report this month that follows up a 2016 study by the same authors — Harvard University professor Lawrence Katz and Princeton University professor Alan Kruger.

It found the percent of workers in alternative work arrangements likely only rose by one or two percentage points between 2005 and 2015 instead of the five percentage-point increase in the initial study.

Alternative work arrangements include independent contractors, on-call workers, temporary help agency workers and contract workers.

Katz and Kruger took a new look at the data after a report released by the US Bureau of Labor Statistics last June found 10.1% of US workers were in alternative work arrangements based on survey information gathered in May 2017.

Their first study found the percent of workers in alternative jobs had risen to 15.8% in 2015 from 10.7% in 2005.

Revisiting the data, Katz and Kruger wrote in the new report that differences between their first report and the BLS numbers were caused by cyclical conditions, differences in survey methods and sampling issues.

If a tighter job market increases the fraction of workers in traditional employment, this factor would have impacted survey results, they noted. The jobless rate was 1.1 percentage point lower in 2017 than in 2015 when their survey was conducted.

In addition, the BLS used different survey methods, contacting respondents by phone or in person. The survey for Katz and Kruger’s initial report was done online.

They also reported the BLS allowed proxy responses with one member of a household responding to the survey on behalf of others. Proxy respondents may be less knowledgeable about other household members’ work and may be less willing to report alternative work arrangements.

Katz and Kruger noted the difficulty in determining the number of those in alternative work arrangements, according to their conclusion.

“Estimating the percentage of workers in alternative work is a difficult task in household surveys. Since only a relatively small proportion of workers are currently working in alternative employment in any given week, and often for relatively modest amounts of income or short periods of time, respondent (or coder) misclassifications and other nonsampling errors are likely to exert a sizable impact on estimates.”