More On the Hawthorne Effect

Hawthorne factory floorA few days ago, I posted an item about an article that Steve Levitt and John List recently published in which they said they found “lost” data on whichthe “Hawthorne Effect” was supposed to have been partly based. Levitt and List claimed that the data showed no such effect, but only an effect on productivity resulting from changes in the seasons.

Yesterday, however, I received the following interesting “comment” from Charles Wrege, who has been studying the Hawthorne experiments periodically for the last half-century. Because of its length and importance, I have decided (with Mr. Wrege’s permission) to move it out of the comments section and incorporate it into a “primary” posting here.

I look forward to other comments by those familiar with the story of the Hawthorne studies.


Christopher Green:
The long lost data on the Hawthorne illumination tests was uncovered by my wife (“B” Wrege) and myself 52 years ago in l957. the material Steve Levitt and John List have “discovered” are the few output records on the microfilms that Dr. Richard Franke wisely had made in l977. The original reports of all the tests, the charts, photographs and personal LOG of Charles E. Snow, the NRC investigator, are in collection 5167, Kheel Center for Labor-management Documentation and Archives, Cornell University.

This material reveals that there MIGHT have been a Hawthorne Effect in the interaction of Mr. Snow and the 16 operators in the test area as Snow had to collect output data 5 times a day to alter the mixed artificial and natural illumination to check on effects of fatigue and the effect on output. The necessary calculations and determination of the hourly incentive earnings resulted in close interaction. In French’s ORIGINAL coining of the term “Hawthorne effect” he said it was when the operators gained social status from the way they were treated by a person of higher social status. Snow’s close association with the operators gave them the name “Charlie’s Girls”.

The real reason for the increases in output were the manufacturing changes at Western Electric a year BEFORE the illumination tests, the effect of the 1923-1924 recession which resulted in less orders for relays, layoffs and with the remaining operators (including the 16 in the “E” type relay test group) only able to assemble a few different types of relays  rather than 8-10 DIFFERENT relay types each day (Snow LOG, January 29, 1925).  There was an average production rate of 55-58 relays per hour due to a piece rate of $1.00/100 relays, on two days (January 7 and March 11, l925) the output was 60 relays per hour due, perhaps, to the assembly of relays paying $1.10/100 relays.

Steve Levitt and John List have said the output was low on Saturday and Monday and attributed the Monday one (partly) to the introduction of a change in illumination intensity. This demonstrates the fallacy of relying only on output charts and not understanding REALITY. Mondays were the day the manufacturing requisitions had to be distributed to the operators and when the individual piece rate slips were prepared for the operators. there were 200 operators (including 16 “E” type relay operators and 7 “A” type relay operators). All this procedure made work start later on Monday, thus lower output. Output was lower on Saturday because on that day, unfamiliar relays, in small orders were assembled.

Steve Levitt and john List have also  suggested that while Snow could change the artificial illumination intensities, he could not control the natural lighting and the operators knew this and that natural illumination may have been more important than artificial. This is a fallacy. Hawthorne, like all factories, schools, homes, streetcars (trams in England) had shades. In his LOG notes for March 4, l925 he records how he adjusted the window shades (as the day before) to reduce the sunlight to make the test area illumination more like the 6.5 foot-candles he was attempting to secure.

The authors apparently rely on the average illumination intensities given by Snow, but the actual intensities at the work bench level were the critical ones. Individual records of this are in Snow’s LOG  of March 11, 1925 (the day ‘E” output was 60 relays per hour) the output the previous day was 55 relays per hour, but the illumination intensities which were supposed to be 33 foot-candles in department 6329 were only 19.1 foot-candles at a line voltage of 116 volts. On January 13, 1925, the  output was 58 relays an hour, and while the illumination was supposed to be 25.5 foot candles it was actually 25.7 foot-candles at a line voltage of 117 volts. In short, you need more accurate data to gauge the lighting effect.

I am afraid we need more definitive data from Steve Levitt and John list before we can regard their work as a true account of what happened at Hawthorne.

Consulting production charts and submitting the figures found there will not reveal what actually happened. the Hawthorne Studies have been handled in a very “unscientific” way for too long. The very fact that the account of the illumination tests found in Roethlisberger & Dickson’s book Management and the Worker was apparently copied by them from a report written by Max Horwath of Hawthorne in 1929 (and preserved by Richard Franke’s efforts) and yet NO ONE ever thought to talk to Mr. Snow , the original investigator, reveals that scholars are too prone to believe a published account rather than hunt for the ORIGINAL DOCUMENTS. it is easier to use a published source (or the documents that someone else has preserved) than to hunt for the originals.

Chuck Wrege, June 11, 2009


About Christopher Green

Professor of Psychology at York University (Toronto). Former editor of the Journal of the History of the Behavioral Sciences. Creator of the "Classics in the History of Psychology" website and of the "This Week in the History of Psychology" podcast series.

4 thoughts on “More On the Hawthorne Effect

  1. [Mr. Wrege sent me more information about the Hawthorne studies, which I thought might be of interest to AHP’s readers. The substantive content of three e-mails is below.]

    [First e-mail]
    There is a facet of the Relay Assembly Test Room that could had a psychological effect ton the operators which was mentioned to Elton Mayo once ink930 has been ignored: the HIGH LEVEL OF NOISE. The automatic recording device (called a “Productivity Device “ by 1929) was a rebuilt Morkrum Prininting Telegraph Reperforator that originally was built to Western Electrics specifications by the small Morkrum Company of Chicago o, because Western Electric could not handle small runs (except for Relays) MILLIONS was their forte and in 1927 they did manufacture 6 BILLION piece parts and proud of it,

    But while the 5 channel tape pf the Reperforator limited the number of operators in the Test Room to 5 operators and did not identify the relays by type of relay which actually determined their piece rate earnings and in a small room (which was only 2’ x 24’ to originally hold 30 small hand coil winding machines), the device was VERY NOISY. On February 2, 1928, a noise level of .85 millimeters was recorded in the Test Room and translated to decibels this is 80 to 85 decibels if measured in water. The Test Room was a crude place, merely a wooden wall enclosure in the southeast corner of Building 37-5 with no insultated walls, with 6 inches open at the bottom and 10 inches on the top for ventilation so noise was everywhere…

    Coil Winding Department 6325 was in a room to the west and north and had 300 both hand winding and automatic winding coil winding machines, the operation of these machines raised the overall noise level to approximately 88 decibels. Through the cooperation of Albert Fisher, Curator of the Royal Air Force Signals Museum, RAF Henlow, Bedfordshire , England , this noise has been recreated by the use of a Creed Printing Telegraph. I will send you Alf Fisher’s e-mail with 5 minutes of this noise. You may find it interesting There have been a few studies by Melmad et al,, that show a benefit in the workplace when performing simple repetitive of ambient noise of 77-95 decibels which might induce an “Arousal Effect.

    [Second e-mail]
    MIGHT I SUGGEST ADDING TWO MORE IMPORTANT ITEMS ? (very short) The FIRST is an indication of the pressure from higher authorities on the operators to work and a result was more effort on their part because of layoffs….despite any lighting level as the fear of layoff was greater than the physiological or visual response to the light level (THUS FRUSTRATION by Snow.) The comments also indicate that the fewer changes in types of relays, or the fact that the operator like the relay she had to assemble was important. The SECOND item deals with the REALITY of coil winding.

    #1 Quotes from Snow LOG on fear of layoff and few relay types and influence on output.
    February 28, 1925 LOG: 66643-5 (Jack Shell and Jack Spring Inspection) “No. 861 worked very steadily on Friday. Possibly she heard a discussion about a layoff –sits right in front of gang chief: “No. 899 went up very high for her. The previous day the gang chief had to talk to her about her low percentage efficiency….” 6329 (relay assembly) “No.136 high on Friday – felt exceptionally well, only two changes of relay type, and liked the relays she worked on. “ observations such as this almost negate any influence of levels of illumination.

    #2 Reality in hand coil winding: Breaking wire and poor wire and Defective spools.

    Hand coil winding was difficult and required considerable training to wind layers of fine wire PERFECTLY in two DIFFERENT layers with a machine that could run at 3500 HP

    Added to this was the fact that if the wire mill produced a bad batch of wire that would break easily then all these breaks have to be repaired and insulating “onion skin” paper inserted. Some operators would come inn early to inspect reels of wire to see if there were weak spots and select those and leave the bad wire for others. Since OUTPUT was the goal as well as HIGH earnings the operators did everything to have an “edge”,

    A second problem could be DEFFECTIVE SPOOLS. An operator might wind everything correctly , but subsequent testing might reveal a defect in the spool used. The wiring would have to be done over and the operator received no credit for the first winding since it was assumed she should have checked the spool first before winding the wire. Life was never EASY at Hawthorne.

    [Third e-mail]
    When I said that the hand coil winding operators would come in early in the morning (probably at 6:30 A.M. as Hawthorne started work at 7:30 at the sound of the plant whistle), they did not start at 7:29 or 7:31, but ALWAYS at 7:30 SHARP.

    By coming in early the operators could sort through the order numbers of the wire and if the reel was numbered 7654 and the 7654 batch was known to have the tendency to break they could look for a reel of wire known to be good, for instance 7700, known to be O.K. and leave the reel of wire that broke easily for some other operator. It was everyone for himself at HAWTHORNE. It was NOT a nice place.

    Unfortunately, as an OUTSIDER Snow did not know about this practice.

    It was kept a secret and only became evident in the 1929 early Hawthorne interviews which were discontinued as they were too revealing about Hawthorne’s bad features. They were made by Richard Frazier of MIT who was a “ghost writer” for George A. Pennock, Hawthorne’s Technical Superintendent. Frazier was an “outsider” and just recorded what he heard. A true “HAWTHORNE PERSON” might have ignored such data, but Frazier wrote it as he heard it.

  2. Chris Green:
    In my haste to forward you information I neglected to add that the illumination at the worklevel was essential and Mr. Snow in his LOG of March 25, l925 recorded two examples of illumination at the level of “Working Material” for the different operations in the different departments:

    “SURVEY OF JANUARY 13, 1925: 200-WATT LAMPS:”(LOG,March 25, 1925)

    Hand Coil Winding: Avg. Illumination.intensity in Foot-candles: 21.0 , at “Working Material”: .8, Relay Assembly: Avg. illum.intensity in Foot-candles: 25.7, at “Working Material: 4-9 F.C., Jack Shell & jack Spring inspection: Avg. illumination intensity in Foot-candles:14.5, at “Working Material 3-6 f. Ii will not list more, but similar conditions existed at other surveys. (which I can provide to ANYONE interested) so while the overall illumination might be at one level, it was the illumination at the actual work level that was the crucial one.

    Induction or receiver coil output.

    I In my research i have been aided by Mr. walter Hall, a retired electroic engineer who in his lifetime has operated hand coil winding machines and much of my knowledge of coil winding machines has been result of hours of conversation on the subject. In the 1920’s Induction Coils were mounted in wooden boxes separate from the desk set used by the subscriber. In this situation an induction coil was not subject to rough use (and the same applied later on) this meant that if a hand coil winder had a wire break during winding, the operator could repair the break by twisting the wires together and then insulate ithe break with a piece of “onion skin” paper. to provide insulation from the other wires. In this way she would achieve a completed coil which would pass inspection so she would be credited with a completed coil, but there was always the possibity it would fail in the future. Since the phones were owned by Bell System, not the subscriber, it would be the Associated Telephone Companies responsibility to replace to the defective coil. there was no easy way to trace the original operator, so she had a completed coil to her credit. A similar situation existed with the two receiver coils in the telephone receiver hanging on the telephone hook of the “candlestick” phone of the 1920’s.

    So when Snow recorded a completed coil on the part of the operator he had no way of knowing if it really was a good one, but one that merely passed inspection, but that would fail in the future. In this sense the record of completed coils may be questionable. They are only completed coils that passed inspection. Joeseph Juran in l997 (one of the founders of Statistical Quality Control) who worked at Hawthorne starting in l924, said that the foremen at Hawthorne had two objectives: to meet the schedules from the Associated Telephone Companies in regard to QUANTITY, and to keep piece rates high to PREVENTunionization. Becauseof these two objectives, QUALITY was secondary so even coils (or other products) might not meet the standards of quality (despie 5,000 inspectors) but because AT&T owned 99.8 % of Western Electric and if they wanted a certain quantity of items they would get them, by :”Hook or Crook:” As in the old musical show, “Damn Yankees,” “Whatever Lola wants Lola Gets”, Whatever AT&T wanted AT&T would get, even questionable induction and receiver coils..

    You should recognize that the fine copper wire the hand coil winders worked with was manufactuired at Hawthorne in their own wire mill. There was no way, in the 1920’s, for the mill to determine the impurities that existed in copper used in the fine wire used in hand coil winding. Impurities could cause wire breaks and the operator had little control over this situation, but this situation certainly caused “Cognitive Anxiety” on the part of the operator so the task of hand coil winding was a demanding one. The operator knew she was paid for a completed coil that passed inspection and that was her goal. As illumination intensities were decreased in the 1926-1927 tests from 11.0 to 4.1 foot candles (the last for only one day) the operators could have maintained their output but probably at a physiological and psychological cost. But the “Bogy” of 16 induction coils was still a threat hanging over the operator’s heads, but the possible earnings served as an incentive also,

    Chuck Wrege June 12, 2009

Comments are closed.