The Hawthorne effect describes possible changes in people’s behavior associated with knowing they are being observed or taking part in research. It does not mean that attention always makes employees more productive.
The term is used for several research-participation effects. Their direction, size and explanation depend on the setting, so an improvement recorded during a study needs more than the label to explain it.
Where the name came from
The name comes from Western Electric’s Hawthorne Works near Chicago. Harvard Business School’s historical collection separates illumination tests conducted from 1924 to 1927 from the relay-assembly experiments that began in 1927.
The relay studies involved questions about rest periods, working hours, incentives, group arrangements and attention. The archive records uncertainty about which factors explained higher output. Western Electric sought Elton Mayo’s guidance in 1928.
These were several studies, rather than one experiment proving that every workplace change increased productivity.
The familiar lighting story is disputed
Steven Levitt and John List reanalyzed the original illumination data. They rejected the familiar account that productivity rose after every lighting change, including dimmer lighting. Their analysis left room for subtler participation effects.
Challenging that historical story does not establish that observation never affects behavior. It limits what those experiments can demonstrate.
Why an increase alone is insufficient
Suppose two fictional groups know they are being studied. One receives new equipment; the comparison group does not. Both rise from 40 to 44 completed tasks per hour.
Each group improves by four tasks per hour, or 10%. The difference between their changes is zero. The numbers alone do not isolate a benefit from the equipment, and they do not prove that observation caused the common increase.
Practice, changes in the work, incentives or other shared conditions might contribute. The explanation depends on how the comparison was designed and what else changed.
What later research found
A 2014 systematic review by Jim McCambridge, John Witton and Diana Elbourne found evidence of research-participation effects, but substantial variation and possible bias limited conclusions about their mechanisms and magnitudes. Most included studies were in health sciences.
A later randomized online trial among university students found no differences in alcohol-consumption outcomes between its three research-information and assessment groups. That finding concerns its particular design and population, rather than every form of observation.
Participant reactivity concerns what the person being studied does. Observer bias concerns how a researcher measures or interprets the result. Either can complicate evidence, but they are different problems.
Last checked and updated on November 2, 2026 by the MBN editorial team.