Why Remote Work Statistics Vary So Much (And What Leaders Should Know)

Remote work has become a defining economic story, but its measurement is surprisingly inconsistent. Depending on the survey, the percentage of U.S. workers who work remotely ranges from 20% to 54%. A new study published in the Review of Income and Wealth identifies four key methodological choices that explain this gap, with question wording being the biggest driver.

The Definitional Dilemma

The Remote Life Survey (RLS), conducted by Gallup in late 2020, shows how dramatically the headline number changes based on how you define "working from home." When asking how often respondents worked from home in the past month, the RLS found:

• 31.6% said they always worked from home.
• 41.2% when including those who worked from home 3-4 times a week.
• 46.9% when including those who worked from home about once a week.
• 53.5% when including those who worked from home once or twice in the past month.

This single survey produces a range of over 20 percentage points simply based on where you draw the line.

Four Sources of Divergence

The study identifies four main factors that cause estimates to vary:

1. Mode of Data Collection

Web-only surveys tend to overestimate remote work because people who are reachable and willing to respond online are more likely to have remote-friendly jobs. In the RLS, 31.4% of web-only respondents always worked from home, compared to just 6.3% of mail-only respondents. Relying solely on web responses would overstate the "always WFH" rate by about 1.6 percentage points.

2. Inclusion of Self-Employed Workers

Self-employed workers are much more likely to work from home. Excluding them reduces the pre-COVID work-from-home rate by roughly three percentage points. This matters because many pre-pandemic benchmarks, like the American Time Use Survey (ATUS), effectively exclude the self-employed, understating the baseline and inflating the perceived increase during the pandemic.

3. Occupational Composition

Surveys that overrepresent white-collar occupations will show higher remote work rates. Reweighting the RLS sample to match the occupational mix of the Current Population Survey (CPS) reduces the at-least-weekly WFH rate from 51% to 43%.

4. Question Wording (The Biggest Factor)

The most significant source of divergence is what you ask. During the pandemic, the CPS asked: "At any time in the last four weeks, did you telework or work at home for pay because of the coronavirus pandemic?" This effectively counts only new remote work due to COVID-19, not total remote work. By contrast, the RLS asked: "In the past month, about how often did you work from home as part of your job?" without referencing the pandemic.

When the RLS data is restricted to only those who were not already remote before February 2020, the headline rates drop sharply:

• "Always WFH" fell from 31.6% to 23.6%.
• "Sometimes or always WFH" fell from 53.5% to 28.2%.

Once this definitional alignment is made, the gap between CPS and RLS largely disappears.

Implications for Leaders

For leaders making data-driven decisions about remote work, the study offers four key takeaways:

1. Interrogate the concept, not just the number. Before acting on a statistic, ask: What exactly is being measured? Is it a level, intensity, or change attributable to a specific cause?
2. Treat survey design as an investment. The difference between a hastily assembled questionnaire and a rigorously designed instrument can be 10 to 20 percentage points.
3. Partner with experts who can quantify the consequences of design choices and align metrics with decisions.
4. Know your population and stay within its boundaries. A survey that excludes the self-employed or offline workers cannot characterize the full labor force.

Remote work is just one example. The same logic applies to measuring employee engagement, wellbeing, skills, and AI adoption. In an environment awash in data, the real advantage goes to those who insist on precision.