[see also the more detailed re-analysis et al 2011; meta-analytic failure to replicate] The “Hawthorne effect” draws its name from a landmark set of studies conducted at the Hawthorne plant in the 1920s. The data from the first and most influential of these studies, the “Illumination Experiment”, were never formally analyzed and were thought to have been destroyed.
Our research has uncovered these data.
Existing descriptions of supposedly remarkable data patterns prove to be entirely fictional. We do find more subtle manifestations of possible Hawthorne effects.
We also propose a new means of testing for Hawthorne effects based on excess responsiveness to experimenter-induced variations relative to naturally occurring variation.
…In this paper, we provide the first statistical analysis of these nearly 90 year old data. We were able to locate microfilm relating to the illumination experiments in collections at the University of Wisconsin-Milwaukee and Baker Library at Harvard Business School. Our search for the original data was triggered by a sentence in 1978’s appendix that mentioned the University of Wisconsin-Milwaukee data archive and an index created by the reference librarians (Stanley Mallach & Steven 1977). Indeed, going through that archive we found a number of documents relating to the illumination studies, including a graphical depiction of one room’s data from the original illumination experiments, which appear to represent the means by which the data were originally recorded by the researchers. Further searching led us to the Baker Library at Harvard University, which contained graphical records for two additional rooms. These 3 data series are the basis of our empirical analysis. To the best of our knowledge, these illumination data have never previously been coded and statistically examined.
Our analysis of the newly found data reveals little evidence in favor of a Hawthorne effect as commonly described, ie. productivity rising whenever light is manipulated. A naïve reading of the raw data does produce such a pattern, however. This is only because all lighting changes occurred on Sundays, the only off day for workers. The empirical fact is that productivity is higher on Mondays than on Fridays and Saturdays. Output on Mondays is equally high, however, whether or not a lighting change occurs on that particular Monday. Thus, researchers seemingly misinterpreted the day of week effect with the Hawthorne effect.
…Despite the prominence of the original Hawthorne experiments, when scholars have later re-analyzed the data, the results have not been as clear-cut as the original researchers claimed. 1978 were the first to carefully analyze data from what is known as the “first relay” experiment at Hawthorne, concluding that most of the variation in production rates could be explained by differences in variables such as managerial discipline and the amount of employee rest. Consequently, there was relatively little scope for “unmeasured changes in the human relations of workers.” 1992, again focusing on data from the first relay experiment, attempts to more directly measure the magnitude of any Hawthorne effects. He finds no evidence, either in the raw data or after controlling for other factors, to support the traditional interpretation of the Hawthorne data.