Lighting level and productivity: a field study in the electronics industry

Ergonomics
Vol. 50, No. 4, April 2007, 615–624
Lighting level and productivity: a field study in
the electronics industry
H. T. JUSLÉN*{, M. C. H. M. WOUTERS{ and A. D. TENNER{
{Philips Lighting Finland, Mattilantie 75, PL4, 04601, Mäntsa¨la¨, Finland and Helsinki
University of Technology, Lighting Laboratory, Finland
{Philips Lighting BV, PO Box 80020, 5600 JM Eindhoven, The Netherlands
The effect of illuminance on the speed and the quality (percentage of errors)
with which workers assemble electronic devices was studied in an electronics
factory in The Netherlands. For the study, the horizontal illuminance was
alternated per work shift between 800 and 1200 lux. The first test was done
during the summer and a second test during the winter. A significant effect of
illuminance has been found. With 1200 lux at the working plane, the speed of
production in the summer was 2.9% higher than with 800 lux. In the winter it
was 3.1% higher with the increased illuminance. There was no significant
effect of the illuminance on the percentage of errors.
Keywords: Lighting; Productivity; Speed; Industrial environment
1. Introduction
The influence of (artificial) illuminance on productivity has been the subject of several
investigations during the past 100 years. Recent findings of non-image-forming, psychobiological
effects via a photo-biological pathway in the brain (Brainard et al. 2001,
Berson et al. 2002, Hattar et al. 2002) and the increasing demand for higher productivity
have made the issue even more important. Improving lighting in the workplace could
have a positive effect on the performance of the workers. Lighting influences productivity
factors such as output, errors and accidents (Vo¨lker 1999, van Bommel et al. 2002).
A literature search yielded several field studies in which the effect of lighting on
productivity has been measured in an industrial environment (Ruffer 1925, 1927,
Goldstern and Putnoky 1931, Schneider 1938, Bitterli 1955, Stenzel 1962a,b, Crouch
1967, Lindner 1975, Carlton 1980, Buchanan et al. 1991). Most of these studies are old
but still interesting, although unfortunately many of them are not very well documented.
In most studies the lighting level has been increased and the effect of the increase on
productivity has been measured. The reported productivity increases were quite large, up
to 50%. In some of the studies not only the lighting has been changed, but also other
*Corresponding author. Email: henri.juslen@philips.com
Ergonomics
ISSN 0014-0139 print/ISSN 1366-5847 online ª 2007 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/00140130601155001

616 H. T. Jusle´n et al.
factors that could influence productivity. In addition to illuminance, the spectral and
spatial distributions of light are also important quality factors.
Perhaps the most famous field study in industry is the Hawthorne study, after which
the Hawthorne effect has been named. The Hawthorne effect (Mayo 1933,
Roethlisberger and Dickson 1939, Parsons 1974) is the effect that the study or
evaluation itself has on the individuals participating in that study; the feeling of being
observed and cared for can lead to improved performance. This effect was first revealed
in studies in the relay-assembly test room of the Hawthorne plants between 1927 and
1932 (Parsons 1974). The duration of the work and the rest pauses were manipulated in
these studies. A group of five women was separated and intensively observed. These
studies had nothing to do with lighting. However, before these studies, researchers had
conducted three lighting studies at the same place (Parsons 1974). These lighting studies
were not very well reported. They did not show a correlation between lighting level and
productivity (Snow 1927).
The most typical and relatively easy to measure productivity parameters in industry are
speed (or its inverse, production time) and the number of errors made. However, in many
cases these parameters are determined by the functioning of the machines involved or by
the quality of the components that have been manufactured somewhere else. To see
whether lighting has an influence on human performance, this study has been performed
in a factory where the speed of the manual assembly was not determined by other factors.
The factory already had a system for recording different kinds of errors that could be
used in the study. By changing the lighting in a fast, alternating way (almost every day),
the influence of other changing factors was minimized.
2. Method
2.1. Factory conditions
In the factory concerned, electronic control gear (viz. ballasts) for gas-discharge lamps
were produced. Machines installed some of the component parts before the semi-finished
products were transported to the manual assembly area. Figure 1 shows the workplaces
and the production lines. Here, three lines were operational, where the workers assembled
the remaining components on the circuit boards by hand (figure 1a). The number of
manually installed components depended on the product type. The tasks were performed
mainly in the horizontal plane.
An automated sunblind system at the glass facade prevented direct sunlight from
reaching the production line closest to the window (figure 1b). The lighting installation
(figure 1c) consisted of continuous lines of luminaires (each with two 49 watt T5 lamps
and a white reflector). This lighting had been installed at the end of 2002, a few months
before the study started. The installation provided a horizontal illuminance of
approximately 1200 lux at the work tables. Task-lighting luminaires were available in
the inspection areas, but not in the assembly area. Floor and walls were white throughout
the area.
There were approximately 17 workers simultaneously present on one production line.
Depending on the product type, four to five of them were working in the manual
assembly. Three production lines were operational during the morning and the evening
shift. The total number of people working at the production lines during the test period
was 119 (mean age 35 years). The participants (the manual assembly workers) were
mainly females. The working schedules stayed the same during the study period, and the

changing of the shift took place every week on Thursday. An example of the shift
schedules is shown in table 1.
There was no air-conditioning in the factory hall. During the winter period the indoor
temperature was constant at about 228C, but in the summer the indoor temperature rose
significantly, peaking at around 20.00 hours. Figure 2 gives an example of the indoor
temperature curve on a day in spring. During the summer, the shape of the curve did not
change much, only the amplitude varied.
2.2. Experimental set-up
Lighting level and productivity 617
Figure 1. Views of the factory hall: (a) an assembly table for the circuit boards; (b) the
manual assembly line closest to the window; (c) a view of the overhead lighting
installation.
To investigate the effect of illuminance on the speed of manual assembly work and on
the number of the errors manual assembly workers make, two lighting levels were
presented to the participants following an alternating scheme. The two levels were
achieved by switching on and off every third luminaire in a row. The corresponding
horizontal illuminance at the assembly surface was 1200 lux or 800 lux and the
corresponding vertical illuminance either 600 lux or 400 lux. There were two
measurement periods, one in summer (5 May 2003–20 June 2003) and one in winter
(5 January 2004–5 March 2004).

618 H. T. Jusle´n et al.
During the first study period, the lighting was changed randomly between the
shifts (see table 2 for part of the schedule). During the second study period (winter)
the lighting was changed in the morning just before 06.00 hours (see table 3 for part of the
schedule).
Productivity has been measured by using the system that was already running in the
factory. The data of the scanners that detected the products when entering the manual
assembly area were used to determine the production time. The data of the errors in the
manual assembly had to be extracted from error reports.
After the first study period, 34 persons working in the factory hall completed a
questionnaire. All light measurements were done after the second period.
3. Results
3.1. Questionnaires
Table 1. Shift schedules for the workers.
Schedule 1 (hours) Schedule 2 (hours)
Thursday 0600–1500 1500–2400
Friday 0600–1500 1500–2400
Monday 0600–1500 1500–2400
Tuesday 0600–1500 1500–2400
Wednesday 0600–1200 1200–1800
Groups change shift on Thursday morning.
Figure 2. Example of the variation in indoor temperature as a function of time in the
factory hall during a normal day in May 2003.
Figure 3 shows the results of the questionnaires. The lighting level was 1200 lux when
the questionnaires were completed. The participants had to state on a 5-point scale to
what degree they agreed or disagreed with the statements given. Apart from the
questionnaire, the workers were asked whether they knew that a lighting study had
been going on during the past 2 months. Even though the workers were informed in
advance, 37% of them answered that they did not know. From the answers to the
questions it can be seen that the employees thought that good lighting influences their
work (mean 4.2, SD 0.6). On average they said that they did not need more light for

their work (mean 2.3, SD 1.1) and that they would like more opportunities to control
the lighting in the working area (mean 3.4, SD 0.8). The answers to the other questions
were on average ‘no opinion’.
3.2. Production times
Lighting level and productivity 619
Table 2. Part of the lighting schedule during the first (summer) study period.
Monday Tuesday Wednesday Thursday Friday
Morning shift 26-May 27-May 28-May 29-May 30-May
Afternoon shift 26-May 27-May 28-May 29-May 30-May
Morning shift 2-Jun 3-Jun 4-Jun 5-Jun 6-Jun
Afternoon shift 2-Jun 3-Jun 4-Jun 5-Jun 6-Jun
Morning shift 9-Jun 10-Jun 11-Jun 12-Jun 13-Jun
Afternoon shift 9-Jun 10-Jun 11-Jun 12-Jun 13-Jun
Italic dates ¼ shifts with all lights on (1200 lux); bold dates ¼ shifts with one third of the lamps
switched off (800 lux).
Table 3. Part of the lighting schedule during the second (winter) study period.
Monday Tuesday Wednesday Thursday Friday
Week 4 19-Jan 20-Jan 21-Jan 22-Jan 23-Jan
Week 5 26-Jan 27-Jan 28-Jan 29-Jan 30-Jan
Italic dates ¼ shifts with all lights on (1200 lux); bold dates ¼ shifts with one third of the lamps
switched off (800 lux).
Only product types that have been produced under both lighting conditions during both
morning and afternoon shifts have been analysed. The mean production times and
standard deviations are presented in tables 4 and 5. The mean production time has been
calculated for all products for the different illuminances and for different times of the day
(morning and evening shift). The product types produced during winter and summer were
not identical, so the production times of summer and winter cannot be compared.
Different products with different production times were made on the same assembly
line. The mean production time per product type ranged from 36 to 63 s for the summer
period and from 44 to 60 s for the winter period. The difference is due to the fact that the
types and quantities of products were not equal in these two periods. To make it possible
to compare the results for the different products, the relative increase or decrease in production
time with regard to the 800 lux situation has been calculated (D ¼ (t 1200 7 t 800)/
t 800, where t ¼ production time). The relative difference has been calculated per product
and this dataset has been used for statistical testing. The mean relative production speed
change, where speed is the inverse of production time per product type, and the mean of
these average values, are shown in table 6 for different time spans. Three-factor ANOVA
(dependent variable: production time; factors: light; shift; and product type) and planned
comparison (for light and shift) were performed separately for both study periods. The
statistical study results are shown in the table 7.
Figures 4 and 5 show the mean relative production time change and its confidence
interval for different products during both shifts and under both lighting conditions for
both shifts.

620 H. T. Jusle´n et al.
Figure 3. Results of the questionnaire. There were 34 completed forms (average age 36.5
years, 31% male, 69% female). Scale: 1, Strongly disagree; 2, Disagree; 3, No opinion;
4, Agree; 5, Strongly agree.
Table 4. Production time (mean and SD) during the summer study.
Summer
mean (s) SD (s)
1200 lux, whole days 48.9 21.3
800 lux, whole days 51.5 20.9
Mornings only 50.7 21.9
Evenings only 50.0 20.4
1200 lux mornings 48.3 22.6
800 lux mornings 52.5 21.1
1200 lux evenings 49.4 20.1
800 lux evenings 50.5 20.6
Looking at the simple main effects, there was a significant effect for illuminance. At
1200 lux, the speed of manual assembly was higher than at 800 lux. The effect was a 2.9%
increase of production speed in the summer and 3.1% increase in the winter. During the
winter, the effect was significant for both shifts and during the summer for the evening
shift only. For the evening shift in the summer, the increase was as great as 6.3%. In the
winter study, the increase (with increase in illuminance) of the production speed was
4.6% during the morning and 1.6% during the evening.

Lighting level and productivity 621
Table 5. Production time (mean and standard deviation) during the winter study.
Winter
mean (s) SD (s)
1200 lux, whole days 49.7 14.3
800 lux, whole days 51.4 15.2
Mornings only 50.9 15.1
Evenings only 50.0 14.4
1200 lux mornings 49.8 14.4
800 lux mornings 52.1 15.7
1200 lux evenings 49.4 14.2
800 lux evenings 50.5 14.6
Table 6. Relative differences in production speed (means of product type means) due to change
in illuminance from 800 to 1200 lux.
Change in production speed (%)
Summer
Morning 70.3
Evening 6.3
Total 2.9
Winter
Morning 4.6
Evening 1.6
Total 3.1
Table 7. Results of the three-factor ANOVA and planned comparisons.
The three-factor ANOVA showed:
Significant main effect for the illuminance factor:
Summer: F(1, 18526) ¼ 6,9406, p 5 0.01
Winter: F(1, 39004) ¼ 63,943, p 5 0.01
Significant main effect for the product types factor:
Summer: F(14, 18526) ¼ 16,512, p 5 0.01
Winter: F(9, 39004) ¼ 8,3967, p 5 0.01
Significant main effect for the shift factor:
Summer: F(1, 18526) ¼ 8,6732, p 5 0.01
Winter: F(1, 39004) ¼ 14,397, p 5 0.01
Interaction between illuminance and shift was significant for both periods:
Summer: F(1, 18526) ¼ 8,7280, p 5 0.01
Winter: F(1, 39004) ¼ 15,267, p 5 0.01
Interaction between illuminance, shift and product was significant for both periods:
Summer: F(14, 18526) ¼ 11,682, p 5 0.01
Winter: F(9, 39004) ¼ 12,272, p 5 0.01
Univariate test for planned comparisons showed:
significant difference in production time for different illuminances for morning and evening shift during
winter and for the evening shift during summer, but not for the morning shift during summer.
Summer: Morning: F ¼ 0,04, p ¼ 0.83
Summer: Evening: F ¼ 20,9, p 5 0.01
Winter: Morning: F ¼ 87,5, p 5 0.01
Winter: Evening: F ¼ 7,0, p 5 0.01

622 H. T. Jusle´n et al.
Figure 4. Summer study. The mean relative production time differences (%) (1200 lux vs.
800 lux) for different products (indicated by A–O) and shifts (vertical bars denote
95% CI).
Figure 5. Winter study. The mean relative production time differences (%) (1200 lux vs.
800 lux) for different products (indicated by P–Y) and shifts (vertical bars denote
95% CI).
3.3. Errors
Table 8 shows the relative differences in error percentages for the 1200 lux situation
compared to the 800 lux situation. The error percentages of the manual assembly were
very low, typically less than 1% and often 0% per shift. The changes in error percentages
are not statistically significant.
4. Discussion
4.1. Methodology
The choice to alternate the illuminance in short periods was made to filter out potential
intervening effects. The most important intervening effects are experience, motivation and

Table 8. Relative differences in error percentages due to the change in illuminance from 800 to
1200 lux.
health of the manual assembly workers. It is not very likely that these effects alternate
synchronously with the lighting change scheme. The effect of being under study
(Hawthorne effect) is estimated to be very limited, because there was no measurement
equipment visible during the test periods. The questionnaire results also showed that the
illuminance changes were hardly noticed and certainly not actively discussed among
the workers. Due to this approach, no conclusions can be drawn about the durability of
the effects. Lighting might also affect quality of sleep, which then would affect the next
day’s performance. These longer-term effects cannot be studied with this approach.
4.2. Lighting and productivity
The results of this study show that the lighting level had an influence on the production
speed. The reason that there was no significant effect for the error percentage might be
caused partly by the fact that the error percentages were already very low. Another
reason might be that quality is a more important factor for this factory than speed. The
employees were encouraged to maintain the quality at all times. Differences in the
production speed for the different seasons indicate that visual performance was not
the only speed-influencing factor. There was no significant effect for the summer
mornings, but during the winter study the effect was stronger in the morning than in the
evening. One difference between these morning periods was that during the summer,
participants had already been exposed to bright daylight before the working day had
begun, whereas during the winter it was dark. The summer evenings were the periods
when the temperature in the hall was so high that there were some complaints. The higher
illuminance might even have increased the heat load in the hall, but nevertheless made
people work faster without compromising quality. It could even be assumed that part of
the effect was caused by the non-visual effects of light, such as direct brain stimulation,
but that remains speculation.
The results of the questionnaire show that participants believe that good lighting
influences their work, even though good lighting was not defined in the questionnaire. As
a group, they also answered that they did not need more light, but would prefer more
opportunities to control the lighting.
5. Conclusions
Change in error percentage (%)
Winter
Morning 736
Evening 21
Summer
Morning 715
Evening 2
Differences are not statistically significant.
Lighting level and productivity 623
Based on the results of this study, it can be said that illuminance is one way to influence
the speed of manual assembly under the conditions described in this paper. There was a
significant increase in speed of the manual assembly when 1200 lux horizontal lighting

624 H. T. Jusle´n et al.
level was provided instead of 800 lux. The effect was significant during both the morning
and the evening shift in the winter and during the evening shift in the summer. There was
no effect on the number of errors. The underlying cause of the increase in speed might be
sought in improved visual performance or psycho-biological effects.
Acknowledgements
The authors wish to thank the management of Philips Lighting Electronics Oss and
all factory personnel for the opportunity to perform this study. Special thanks go to
Rob Lamers, the initiator of the study, who also arranged all practical issues during the
research, to Michel Hovius for his help with the speed measurements, to Magdalena
Schymanska and Erik Heermans for their help with the quality data and to Henk
Slotman for his help with practical issues, especially during the winter period. The
assistance of Henk Verstegen, Chris Martens and Noe¨l Bonne´ of the Philips’ Lighting
Controls Department has been essential for the realization of the alternating
illuminance.
References
BERSON, D.M., DUNN, F. and TAKAO, M., 2002, Phototransduction by retinal ganglion cells that set the circadian
clock. Science, 295, 1070–1073.
BITTERLI, E., 1955, Licht und Arbeit. Bulletin Schweizerischer Elektronischer Verein, 46(12), 559–563.
BRAINARD, G.C., HANIFIN, J.P., BYRNE, B., GLICKAMAN, G., GERNER, E. and ROLLAG, M.D., 2001, Action
spectrum for melatonin regulation in humans: evidence for a novel circadian photoreceptors. The Journal on
Neuroscience, 21, 6405–6412.
BUCHANAN, T.L., BARKER, K.N., GIBSON, J.T., JIANG, B.C. and PEARSON, R.E., 1991, Illumination and errors in
dispensing. American Journal of Hospital Pharmacy, 48, 2137–2145.
CARLTON, B., 1980, Industrial lighting must be tailored to real-life working conditions. Electrical Times, issue
4562, 14 March, 10–11.
CROUCH, C.L., 1967, Working and living in a luminous environment. Transactions of the Illuminating Engineering
Society, 32, 41–53.
GOLDSTERN, N. and PUTNOKY, F., 1931, Die wirtschaftliche Beleuchtung von Webstuhlen; neue arbeitstechnische
Untersuchungen. Licht und Lampe, 22, 5–9, 25–28.
HATTAR, S., LIAO, H.-W., TAKAO, M., BERSON, D.M. and YAU, K.-W., 2002, Melanopsin-containing retinal
ganglion cells: architecture, projections and intrinsic photosensitivity. Science, 295, 1065–1070.
LINDNER, H., 1975, Beleuchtungssta¨rke und Arbeitsleistung – Systematik experimenteller Grundlagen. Zeitschrift
fu¨r die gesamte Hygiene und ihre Grenzgebiete, 21, 101–107.
MAYO, E., 1933, The Human Problems of an Industrial Civilization (New York: MacMillan).
PARSONS, H.M., 1974, What happened at Hawthorne? Science, 183, 922–932.
ROETHLISBERGER, F.J. and DICKSON, W.J., 1939, Management and the Worker (Cambridge, MA: Harvard
University Press).
RUFFER, W., 1925, Leistungssteigerung durch Versta¨rkung der Beleuchtung. Die Lichttechnik, 2, 53–58.
RUFFER, W., 1927, Licht und Leistung. Licht und Lampe, 4, 242–245.
SCHNEIDER, L., 1938, Fo¨rderung der menschlichen Arbeitsleistung durch richtige Beleuchtung. Das Licht, 8,
286–296.
SNOW, C.E., 1927, Research on industrial illumination. The Technical Engineering News, 8(6), 257, 272 – 274, 282.
STENZEL, A.G., 1962a, Erfahrungen mit 1000 lux in einer Lederwarenfabrik. Lichttechnik, 14, 16–18.
STENZEL, A.G., 1962b, Erfahrungen mit 1000 lux in einem Kamerawerk. Lichttechnik, 14, 351–353.
VAN BOMMEL, W., VAN DEN BELD, G.J. and VAN OOYEN, M., 2002, Industriebeleuchtung und Produktivität.
Tagungsband Licht, 15, 48–58.
VÖLKER, S., 1999, Eignung von Methoden zur Ermittlung eines notwendigen Beleuchtungsniveaus. PhD Thesis,
Technical University Ilmenau, Germany.