Cornell University
Lighting the Computerized Office

Principal investigators:

Alan Hedge, Ph.D, Associate Professor
William R. Sims Jr., Ph.D, Professor and Chairman
Franklin D. Becker, Ph.D, Professor

Department of Design and Environmental Analysis
New York State College of Human Ecology
Cornell University, Ithaca, New York

Initial study report completed October 1989
Extracts and conclusions presented to the
Human Factors Society October 1989

Supplementary study report completed September 1990


Subject: Cornell University

In the last decade, during which personal computers have proliferated in the office, researchers have paid increased attention to lighting. Several studies have investigated the relationship of modern office lighting to the visual health, satisfaction and productivity of office workers.

  • The American Society of Interior Designers found that 68% of employees complain about the light in their offices.
  • A Silicon Valley study pointed out that 79% of VDT users want better lighting.
  • A 1989 Louis Harris survey, the Steelcase Office Environment index, revealed that workers think of eyestrain as the number one health hazard in the office—ahead of radiation, asbestos, even exposure to AIDS.

This degree of dissatisfaction is difficult to ignore. It confirms the need to identify the best methods of lighting the computerized office.

During the 1980’s, several studies attempted to quantify the relative merits of the various office lighting methods, including the two most commonly used: parabolic downlighting and indirect lighting. But in many of the studies, the computer workers did artificial tasks under controlled laboratory conditions. No matter how conclusive the findings, their applicability remained open to question.

Without a long-term, real-world field test, there were no definitive answers. However, such a test required a great amount of expense and preparation, plus the full cooperation of a company able to provide the right test facility and the right test subjects.

In the spring of 1988, Cornell University put all the elements in place. This study is the result.

The Structure of the Study

The study was designed to determine whether the selection of a parabolic downlighting system or a lensed indirect uplighting system would make a difference in the visual comfort, satisfaction, health or productivity of computer workers. To minimize the influence of differences in color schemes, furniture, work habits or tasks of individual employees, the researchers sought an installation where a suitably large group of professionals worked in similar offices on a variety of team based computer-oriented tasks, with the lighting system as the only significant variable.

The Lighting Methods Used

These were selected as the best quality commonly used for computer applications in mid 1988.

The Test Facility

The study took place in a Xerox Corporation office in Webster, NY which had already been scheduled for renovation prior to the study. It was occupied by 200 highly-skilled employees who regularly work with computers. At the beginning of the test, the building had several different types of lighting, including a large number of 2’ x 4’ downlights of a type seldom installed in computerized offices today. The building offered open office areas with partitioned cubicles, 10' x 15' enclosed offices with windows, and 10' x 15' and 10' x 10' windowless enclosed offices.

In mid-1988, the building was renovated. Color scheme, furniture, carpeting and fluorescent lamp type were standardized throughout. Approximately half of the workers received a ceiling-recessed parabolic downlighting system, the other half received the lensed indirect uplighting systems.

June 1989 – The First Study

A benchmark survey taken in late spring 1988, before the renovation, brought answers from 147 workers (92% return rate). The “after” survey taken a year later brought answers from 90 workers (61% return).

Work-related Health Complaints

Daily complaints of tired eyes and eye focusing problems twice as frequent among the parabolic group
The study isolated the frequency of health complaints, one clear indication of productivity.

In the small, windowless offices, symptoms were considerably more frequent within the parabolic group.


Lighting Modifications

Almost half the parabolics in the windowless offices had been altered.
In over three-quarters of the enclosed offices, at least one of the fixtures had been altered, either by disconnecting lamps within the fixture, disconnecting the whole fixture or putting up some form of paper baffle to shield the fixture. Eleven per cent of the fixtures in the open plan showed similar modifications. So far as the researchers could determine, no one attempted to modify the indirect system in any manner.


Parabolic group favors lensed indirect, indirect group preference is almost unanimous for lensed indirect.
The workers were asked to select, from all the lighting systems they’d ever worked under, which lighting system they preferred. Those who’d worked under the new parabolic system voted for the lensed indirect by a clear margin. Only one person who’d worked under the lensed indirect preferred the parabolic.

Effects of Lighting on Work Function

Subjects find parabolic more bothersome than lensed indirect
Because of the high skill level and high creative involvement of the test subjects, both Cornell and Xerox felt that the most important findings of the study would be those which isolated and quantified the workers’ satisfaction level. The subjects defined the effects of the two different types of lighting on work functions in general.

June 1990 – One Year Later

The researchers returned to verify their findings after the groups had each had a further year of acclimatization to the two systems. They wanted to earn whether satisfaction levels or the frequency of visual health problems had changed, and to see what actual effects the visual health problems created. The second survey brought answers from 121 workers (82% return).

Work-related Health Complaints

Problems under parabolic cut into worker productivity
In this study, researchers asked workers to report the amount of time they lost from various symptoms. Far more parabolic workers reported lost time, and a high percentage lost over 15 minutes a day. For instance 20% of the parabolic group reported losing more than 15 minutes a day due to eye focusing problems compared to only 2% of the lensed indirect group.

Lighting Modifications

Almost half the parabolics in the whole office area now modified or disconnected
Of the 164 office fixtures in the parabolic portion of the test building, 79, or 48% had been altered. Of that number, ten had been completely disconnected. In the windowless offices, over 75% had been modified (51 of 67). Only one worker in the indirect group modified their lighting.


Indirect group still strongly prefers lensed indirect, parabolic group now prefers indirect by equal margin
One year before, 48% of the parabolic group had preferred the indirect lighting. Now 75% preferred it.

Effects of Lighting on Work Function

Lighting problems now tied to lost work time
Unlike the previous study, subjects were asked to quantify the amount of productive work time they lost due specifically to lighting problems. High percentages of workers under the parabolics reported losing over 15 minutes a day.

Some Questions Raised by the Cornell Study

Any study which produces such important and potentially influential findings will immediately raise questions. The more concerned, interested and informed the questioner, the more specific the questions become. This is a sampling of some of the most commonly asked questions about the Cornell Study.

Were there any differences between the two groups that might have influenced the results?

The respondents in both the 1989 and the 1990 survey showed no significant differences in demographics or work profile between the parabolic and the lensed indirect group. All subjects spent a large portion of their working hours at the computer.

Generally, the subjects were highly-skilled workers whose tasks would better be defined as “creative” rather than “routine.” They were involved with graphics, language translation, technical documentation and software development. Many were working with two or three computers at the same station.

Their offices were similar and they were divided up into approximately the same balance between open-plan cubicles and enclosed offices. Almost everything other than the lighting was consistent between the two groups: office temperature, ventilation, noise level, chair comfort, desk space, storage space, privacy, furniture, wall and partition color and so forth.

Was the test structure biased in favor of either system?

The researchers began with certain hypotheses based on earlier laboratory studies. The study report states on page 11, “It was hypothesized a priori that the lensed indirect system would create a more favorable luminous environment, i.e., higher satisfaction, fewer glare problems, better visual health, than that produced by the parabolic system, and consequently one-tailed tests were used for all analyses of questions concerning the effects of the different lighting systems. Two-tailed tests were used wherever analyses were performed in the absence of a priori directional hypotheses.”

The survey questions were specifically structured to be neutral. In order to provide the most useful comparative data, the Cornell research team maintained contact with members of other research groups conducting work on office lighting. A number of the questions were identical to those used in survey questionnaires from other recent lighting studies.

One factor made the test something less than a strictly fair comparison. Since this was a real-world study, the workers were free to behave towards their lighting as they would have if no study had been conducted. Almost from the beginning, the workers under the parabolics requested that a high percentage of the fixtures be modified. BY 1990, 6% of the 164 parabolic office fixtures were disconnected entirely and an additional 42% had had one or two fluorescent lamps disconnected.

In the small offices, 76% of the parabolics were modified.

Does the cost differential between the two systems modify the impact of the findings?

From the beginning of the test, the researchers were well aware that the lensed indirect system cost more than the parabolic. Therefore, study results which favored the lensed indirect system over the parabolic wouldn’t be meaningful unless they also proved a superiority sufficient to offset the added cost. In this particular installation (a building with no unusual installation problems for either type of fixture), the cost differential for installed fixtures was between $100 and $200 per workstation, depending on the methods used to calculate the cost. The 1990 study revealed that the parabolic workers were losing substantial amounts of productive work time each day due to lighting-related health problems. The following three examples give an indication of the financial impact of that lost time, assuming a conservative installed cost differential of $200 per workstation and an average work loss differential of ten minutes per day.

Economic Impact: Three Scenarios

(Based on 25% salary burden and 8-hour work day)

Installed cost differential per workstation: $200 Installed cost differential per workstation: $200 Installed cost differential per workstation: $200
Annual salary per workstation: $15,000 Annual salary per workstation: $25,000 Annual salary per workstation: $35,000
Average additional time lost/day by parabolic workers: 10 minutes Average additional time lost/day by parabolic workers: 10 minutes Average additional time lost/day by parabolic workers: 10 minutes
Lensed Indirect pays for itself in 6.1 months Lensed Indirect pays for itself in 3.7 months Lensed Indirect pays for itself in 2.6 months

Did the “Hawthorne Effect” influence the results?

Since the findings of the study were based on subjective criteria, the researcher went to considerable length to avoid having the results jeopardized by this much-discussed phenomenon. Earlier studies suggest that workers respond positively to any change in their work environment and the involvement of a researcher.

In this study, the office lighting was replaced as part of a scheduled renovation, and the evaluation questionnaire covered many aspects of the office environment, not just lighting. The researchers were careful to disguise the fact that the study was primarily about lighting, and the 28-question, 169-item survey was presented as an “evaluative study of the three important factors in office design.”

Both halves of the building received lighting systems which claim to support computer use considerably better than the previous lighting. Therefore, any influence from the Hawthorne Effect should have been minimal and should have been equal for both study groups.

Why weren’t workers shifted during the test so they could experience working under both systems?

This was a real-world test, not a controlled laboratory test. A large-scale movement of employees from one system to the other would have added an unrealistic element to the test. It also would have made it apparent to the subjects that a comparison between the two lighting systems was the true focus of the test.

In fact, because this was a phased renovation, some parabolic workers moved in the indirect area before the parabolic area was completed, which may have had some influence on overall preference figures. Within each group, some changed offices and some didn’t. Whether workers moved offices within the group or not was examined statistically and shown to have no significant effect on their opinions about lighting.

Are the results specific to the particular luminaires used in the study?

In the case of the parabolic system, the answer is “no” with minor qualifications. Regardless of the sophistication of its design, direct parabolic downlighting provides a cone of light from an otherwise dark ceiling. Direct lighting creates direct shadows, contrast between the fixture and the ceiling and an environment that is more brightly lit within the cone of light than outside it. The answer is less clear-cut in the case of the indirect system. There is great variation in the ability of indirect lighting systems to distribute light across a wide area. An indirect system like the systems used in the study will provide a superior luminous environment if it can produce even luminance ratios across the entire ceiling. However, one with poor distribution will produce hot spots on the ceiling bright enough to cause contrast and shadow problems similar to those created by downlights. Substituting a lesser system for the one used in the study could have created considerably different results.

Why did the study use indirect luminaires with visible lenses rather than totally indirect fixtures?

The researchers felt it was important to compare the best indirect lighting then available with the best parabolic lighting then commonly available in order to provide as fair a comparison as possible. In the open office areas, the low-brightness visible lenses are claimed by the manufacturer to provide a higher satisfaction level because they provide a higher level of perceived illumination, a contention supported by earlier studies (Bernecker, Penn State University, 1984 and 1986). In the enclosed offices, the lensed indirect fixture provides a superior luminous environment to other commercially available indirect luminaires because the lens provides an extremely widespread distribution designed especially for enclosed areas.

Why were self-reported evaluations of productivity used instead of objective measurements?

At the beginning of the study, representatives from Xerox made it clear that, in their opinion, self-reported evaluation would be the most useful information these employees could provide. The creative nature of the high-salaried job functions created broad variety in the amount and quality of job output. Quality of product meant far more than the number of data bits entered. An attempt to quantify individual output would have been somewhat meaning less in any event, because much of the output in this installation was based on team effort rather than individual production.

Why are measures of preference included, since they seem to have little to do with productivity?

Preference is more important than might first appear. In the positive sense, the very fact that a group of workers prefers a specific type of lighting makes it superior lighting, regardless of whether an outside observer might insist that some other alternative is a “better” system. Preferred lighting contributes to the sense of “a nice place to work” and helps workers feel better about their environment and their job. It also aids in recruitment, since recruiters can point the lighting out as a job benefit. In the negative sense, any irritant in the workplace will contribute to employee dissatisfaction, and at a certain level, dissatisfaction turns to departure. Each personnel expert may place a different dollar value on any phase of the attrition-recruitment-rehiring-retaining cycle, but all agree that any reasonable technique to reduce employee dissatisfaction is a useful and practical tool.

Did differences in light level influence the test results?

Commonly recommended light levels for parabolic systems in computer installations are somewhat higher than those recommended for indirect systems. Normally, parabolic installations have an average illuminance of 50 to 70 fc and indirect installations have an illuminance of 30 to 50 fc. Both systems were installed according to standard practice and to provide average illuminances within the standard boundaries.

© Cornell University, 1991