Content uploaded by David P Wyon
Author content
All content in this area was uploaded by David P Wyon on Feb 16, 2015
Content may be subject to copyright.
INDOOR ENVIRONMENTAL EFFECTS ON PRODUCTIVITY
David P. Wyon
Research Fellow, Johnson Controls Inc., Milwaukee, USA
Adjunct Professor, DTU, Copenhagen, Denmark
ABSTRACT
Thermal conditions within the thermal comfort zone can reduce key aspects of
individual human efficiency, such as reading, thinking logically and performing
arithmetic, by 5-15%. Poor air quality which gives rise to headache and fatigue
may have similar effects. Vertical temperature differences may cause air quality
problems at head height. The published evidence for such effects is reviewed.
Case studies of IEQ and productivity in the field can sometimes quantify
outcomes but seldom prove causation. A new approach to productivity research is
presented which can determine not only whether, but also how and how much
productivity is affected by specific aspects of IEQ. Claims that architectural or
engineering design features impact productivity would be tested by formulating
specific mechanisms in terms of chains of linked hypotheses. If any hypothesis in
the chain is untrue, the mechanism is invalid. Seven mechanisms by which the
office environment might affect productivity are defined. The 48 falsifiable
hypotheses involved can be tested in the field using existing outcome metrics, in
experiments which do not involve any long-term decrease in overall productivity.
INTRODUCTION
It is important to distinguish between individual performance and overall work
force productivity. Performance metrics such as speed or percentage errors are
unrelated to the monetary value of the work performed, while productivity metrics
are expressed in terms of value added per unit cost. Thus productivity is affected
by absenteeism and by health costs, and by the capital and running costs of a
building, none of which affect performance. Individual performance efficiency is
likely to have a major effect on productivity, as wage costs per unit floor area are
usually many times greater than the cost per unit floor area of providing and
running the building. There is good reason to believe that performance is affected
by such diverse intervening variables as motivation, wellbeing, subjective
comfort, Sick Building Syndrome (SBS), overcrowding, visual and acoustic
distraction, as well as by physical environmental variables. It is unwise to
consider the consequences of environmental engineering design for productivity
without also taking account of the effects of architectural design. The current
trend towards alternative office designs and towards increased occupant density
provides a good example: even if such changes do save floor space, and thus
reduce costs, if they reduce group or individual performance they may be having a
negative impact on overall productivity. This paper reviews published evidence
for the effects of thermal climate (cold and heat stress), humidity (per se and as a
factor in heat stress) and air quality (airborne particulates, VOCs) on human
performance. A number of published cost-benefit analyses are then reviewed.
They show that quite small changes in overall work force productivity would
justify major investment in the indoor environment. The approach recommended
by ASHRAE 700-RP is to compare one total solution with another in terms of
overall productivity in the real commercial world. This carries a high cost penalty
if inferior conditions must be established or maintained as controls for long
enough to prove beyond all doubt that they are inferior. It is worth noting that
nothing at all is learned from such demonstrations unless the results are
statistically significant, for at the end of this paper a complementary approach to
productivity research is presented, one that would offer new insights into exactly
how architectural or engineering improvements affect productivity. The
experiments would be carried out in the real commercial world, but because the
linked hypotheses which define each specific mechanism are studied separately,
overall productivity need hardly be affected.
SICK BUILDING SYNDROME (SBS)
Studies of large numbers of people working in Dutch office buildings show that
the total amount of sick leave due to SBS is likely to be 34% lower when
individual workers can control their own thermal environment (Preller et al.
1990). This prediction is based on the odds ratio or estimated population value, as
actual prevalence rates in the samples studied were not published. Similar studies
in UK offices by Raw et al. (1990) indicate that self-estimates of efficiency are
significantly higher when individuals can control their own thermal climate, or
their own ventilation, or the lighting levels where they work, in comparison with
similar offices where this is not possible. This study also showed that while 2 SBS
symptoms were quite normal, 6 SBS symptoms were associated with a 10%
decrease in self-estimated efficiency. In a recent experimental study in Canadian
offices (Nunes et al. 1993), workers reporting any symptoms of SBS were found
to be working 7.2% more slowly on one standardised computer task, designated
the Continuous Performance Task or CPT (P<0.001), and to be making 30% more
errors on another task designated the Symbol-digit Substitution Task or SST
(P<0.07). The CPT was a vigilance task in which subjects monitored a series of
displays appearing on a computer screen and had to respond to one that had been
designated as the target, while the SST was a complex coding task. 47 subjects
performed each task once a week for three weeks, without supervision, as part of
their daily routine. These three papers provide strong and quantitative evidence
for the link between SBS and productivity. Any factor which reduces SBS will
increase performance and reduce absenteeism, both of which affect productivity.
AIR QUALITY EFFECTS ON SBS
Macromolecular Organic Dust was found by Danish workers to be associated with
SBS (Gravesen et al. 1990). Fine dust particles account for a very small fraction
by weight of airborne dust, but they remain suspended for very long periods and
as they are respirable, i.e. not retained by the nasal passages, they may have a
disproportionately large effect on allergy and SBS. When charged, their mobility
in an electrostatic field is much greater than that of larger dust particles, and they
are therefore more easily retained on charged surfaces. This is the mechanism
proposed to explain the significant effect of ionisation on SBS in a problem
building (Wyon et al. 1991, Wyon 1992a). Additional filters and much finer filters
downstream of the original supply air filters would be necessary to eliminate such
particles. Air filtration techniques, their cost-effectiveness and their contribution
to productivity may well be a fruitful area for future research.
THERMAL EFFECTS ON SBS
The link between sensations of dryness and SBS was established in a passive
study of 2150 office workers in Finland (Jaakkola et al. 1989). Dryness and SBS
symptoms increased very markedly with air temperature in the range 20-24 C in
this study, and an intervention study of the SBS symptoms reported by 100
workers in a computerised office at various imposed temperatures in the same
range showed a very marked effect of thermal conditions on SBS (Krogstad et al.
1991)). Each thermal condition was maintained for a week. Virtually all SBS
symptoms increased with temperature from a minimum at 20-21 C, and the effect
was widespread rather than confined to a few sensitive individuals: the proportion
reporting headache and fatigue increased from 10% at 20-21 C to over 60% at
24.5 C, and other SBS symptoms, including skin problems, showed similar
effects. Thermal exposure is accurately known and experimentally manipulated in
intervention studies, while passive epidemiological studies must usually estimate
thermal exposure on the basis of infrequent measurements and must also accept
that other factors affecting SBS may well be confounded with temperature. These
disadvantages may explain the lack of an apparent effect of temperature on SBS in
some epidemiological studies.
SBS AND SENSATIONS OF DRYNESS
One of the most common complaints indoors in winter in cold countries is that the
air feels dry. This sensation is associated with SBS. Laboratory studies show that
people are ill equipped to detect humidity as such: they often report a decrease
when an increase has been imposed, and vice versa (Rasmussen 1971). It has also
been found that people are not distressed even by 78 hours at less than 10%
relative humidity (Andersen et al. 1974). The incorrect conclusion drawn at the
time from these findings was that low winter humidities indoors do not matter.
Humidification often causes health problems due to condensation on poorly
insulated outer walls or even inside building materials, resulting in mould growth.
An increase in sick leave due to SBS in humidified offices in Holland (Preller et
al. 1990, op. cit.) is believed to be due to these disadvantages.
A major experimental field study in large office buildings in Sweden (Andersson
et al. 1975) showed that a 2 C reduction in room air temperatures, from 23 to 21
C, dramatically reduced complaints of dry air, whereas humidification from 20%
to 40% RH, although it reduced complaints of dry air, caused a large increase in
complaints that the air was too humid: the proportion satisfied with the humidity
was unchanged. The air in the laboratory studies cited above was extremely clean,
and the quite different results obtained in offices suggest that the distress
associated with dry air is caused by the inability of dry mucus membranes in the
nose, throat and eyes to deal with airborne dust, at least for some individuals, and
that the drying associated with low winter humidity is greatly increased by room
air temperatures at the upper end of the conventional range for thermal comfort,
20-24 C. This may be one reason why a source of radiant heat is appreciated in
cold countries in winter - it allows air temperatures to be lower.
VERTICAL TEMPERATURE DIFFERENCES & SBS
Vertical thermal gradients are always positive, since hot air rises. In rooms where
the air flow must remove a high heat load, and particularly if this is to be done by
displacement rather than by complete or partial mixing, air temperature at head
height may be 2-3 C higher than at floor height. Cold feet and warm air to breathe
is the exact opposite of human requirements, and an individual who experiences
SBS or sensations of dryness will often be forced to lower the room temperature
to such an extent that it will be too cold for whole body heat balance as well as too
cold for the feet. Over 40% were found to experience local thermal discomfort,
even when conditions had been adjusted to provide preferred whole body heat
loss, when 72 subjects were exposed for one hour to two typical winter and two
typical summer conditions in an office module with displacement ventilation
(Wyon and Sandberg 1990). The experiment on which the recommendations in
ISO 7730 (1984) are based would have predicted only 5% dissatisfied under these
conditions (Olesen et al. 1979). In a later experiment, in which over 200 subjects
were exposed for one hour to vertical temperature differences of 0, 2 and 4
C/meter, with room temperatures resulting in the same three states of whole body
heat balance at each vertical temperature difference, local thermal discomfort was
found to be unaffected by vertical temperature difference, but highly sensitive to
whole body heat balance (Wyon and Sandberg 1996, Wyon 1994). This indicates
that thermal gradients are only a problem because they lead to an increase in air
temperature in the breathing zone. Even if the individual has a choice in the
matter, it is an uncomfortable one, between the risk of SBS and a room
temperature that is too low for comfort. Whichever is chosen, the end result is
likely to be that productivity is reduced by vertical temperature differences.
DIRECT EFFECTS OF AIR QUALITY ON PRODUCTIVITY
Carbon dioxide concentration is commonly used as an indicator of reduced air
quality. In occupied rooms, ppm CO2 will increase above recommended levels if
the outside air supply rate per person is inadequate, but this may increase or
decrease the concentration of respirable particles, depending on their source.
Odour from sources in the room will increase, as will air temperature and
humidity in non-airconditioned buildings. The practical difficulty of manipulating
any one aspect of air quality in isolation is perhaps the reason that so few
systematic studies of air quality on productivity appear to have been carried out.
No such studies are mentioned in what is otherwise a very comprehensive survey
of indoor climate effects on health, comfort and productivity carried out in 1980
( McIntyre 1980). Another reason may be that the New York State Commission
on Ventilation (1923), one of the earliest and still one of the most comprehensive
experimental studies of its kind, concluded that there was ". . no inferiority in the
quality of the product produced in stagnant air at 30 C, 80% RH, with 30 to 40
parts of CO2 per 10 000 (i.e. 3000-4000 ppm, author's note) when subject to this
condition for eight hours a day for four successive days . ." even under conditions
conducive to lowered standards of work. No evidence to the contrary appears to
have been published in the intervening years.
However, some specific air pollutants are known to affect both health and
performance. Danish researchers have shown that toluene levels of 100 ppm,
which were then permitted in workplaces in most countries, had a significantly
negative effect on manual dexterity, performance of a colour discrimination task,
and the accuracy of visual perception (B‘lum et al. 1985). Toluene at very low
levels is one of the most commonly occurring of all indoor air pollutants, but 100
ppm will only occur in workplaces where it is in use industrially, so that these
results do not constitute evidence that poor ventilation in normal buildings will
reduce productivity.
THERMAL EFFECTS ON PRODUCTIVITY
The present article is based on the author's review of the effects of healthy
buildings on human performance (Wyon 1993), in which the following detailed
conclusions were drawn:
1. Thermal conditions providing optimum comfort may not give rise to
maximum efficiency. In an experiment in which normally clothed young
American subjects performed mental work at different temperatures (Pepler and
Warner 1968), they were most thermally comfortable at 27 C, the temperature at
which they exerted least effort and performed least work. They performed most
work at 20 C, although most of them felt uncomfortably cold at this temperature.
Thermal comfort and measurement of the underlying state of heat balance
are usually assumed to be able to identify different combinations of conditions at
or close to thermal neutrality that will give rise to the same level of performance:
no difference in performance was found between two conditions of thermal
neutrality with very different clothing insulation and air temperature: 0.6 clo at
23.2 C and 1.15 clo at 18.7 C (Wyon et al. 1975). This was a climate chamber
experiment lasting only 2.5 hours. Subsequent research reviewed above shows a
powerful effect of temperature on SBS in field studies of longer duration, a
mechanism by which productivity might be substantially reduced in practice
under the warm air, thin clothing condition.
Another example of the failure of thermal indices based on human heat
balance and subjective thermal comfort to predict performance is to be found in a
series of experiments on heat-acclimatised European men living in Singapore
(Pepler 1958). At raised temperatures, performance of a tracking task was
consistently worse at 80% RH than at 20% RH, although temperatures had been
adjusted so that both levels of humidity produced identical levels of Effective
Temperature. Pepler did not claim that this difference was statistically significant,
but Poulton (1970) points out that a reliable performance decrement was found at
a lower Effective Temperature in the humid climate. Thus high humidity did
reduce performance under physiologically and subjectively equivalent conditions.
2. The effects of heat stress on human efficiency are not always linear. In the
experiment to which reference is made above (Pepler and Warner 1968), rate of
working showed a minimum at 27 C, although naturally it would have declined
below this value in extreme cold or heat. A similar reversal of the effects of
moderate heat stress on performance was demonstrated under conditions of
controlled hyperthermia (Wilkinson et al. 1964), and also in an intervention
experiment in Swedish school classrooms (Wyon 1970). Heat-acclimatised
factory workers resident in South Africa were shown to perform industrial tasks
significantly better at 32 C than at 26 C or at 38 C, during 8-hour exposures in
normal work clothing (Wyon et al. 1982, Kok et al. 1983). It has also been shown
that muscular strength increases to a maximum in moderate cold stress before
declining at more extreme levels (Clarke et al. 1958). These non-linear effects are
in marked contrast to the orderly decline of comfort below and above a maximum
at some neutral temperature.
However, the performance of a wide variety of simulated industrial tasks
involving finger strength and speed, manual dexterity, hand steadiness, etc. were
shown in the South African series of experiments to decline monotonically below
thermal neutrality (Meese et al. 1982). Critical air temperatures for unimpaired
performance varied from 18ø to 12 C, depending on the task. Finger speed and
finger-tip sensitivity were measurably impaired at the air temperature preferred
for thermal comfort (18 C), in comparison with the air temperature (24 C) at
which finger temperatures reached their maximum value.
3. The above conclusions together invalidate the usual assumption that
performance effects can always be deduced from studies of thermal comfort
alone.
4. The critical temperature for performance in temperate zones lies at about 30
C for normal humidity levels. This conclusion was reached by Pepler (1964) on
the basis of studies made in weaving sheds and coal mines by the Industrial
Fatigue Research Board in England.
5. Accident rates in temperate zones are lowest at 20 C, increasing by over
30% below 12 C and above 24 C. The studies on which this conclusion is based
were carried out in munitions factories over periods of 6, 9 and 12 months
(Vernon 1936). Accidents are assumed to increase in adverse working conditions
due to a decrease in human efficiency.
6. Moderate heat stress has an adverse effect upon efficiency for men, but a
much smaller effect on women. This conclusion is based on the above accident
data from munitions factories. The effect of cold on accident rates was closely
similar for men and women, suggesting that similar work was being performed by
both groups. A similar gender difference was found for the effect of intermediate
levels of heat stress on mental performance in laboratory studies of 18-year-old
Danish subjects (Wyon et al. 1979). The direction of this gender difference is the
opposite of that found for extreme levels of heat stress.
7. Moderate heat stress and fatigue interact to decrease efficiency. In studies
of 18 455 English coal miners (Vernon 1936, op. cit.), accidents always increased
with the number of hours worked in the first six hours of a shift, but did so more
rapidly at 25 C than at 18 C, and more rapidly at 28 C than at 25 C. After 6 hours
accidents, and presumably workrates as well, were observed to decline at the two
higher temperatures, as the risks involved in maintaining workrates became more
obvious to the miners themselves. The adverse effects of moderate heat stress on
the performance of work by Swedish schoolchildren was found to be greater in the
afternoon, when they were tired, than in the morning (25).
8. Moderate heat stress increases the dependence of accident frequency on
age. Vernon (1936 op. cit.) found that age was barely a factor in accident
causation at temperatures below 21 C, but that the relative accident frequency
increased with temperature by up to 40% for older men in the range 22-30 C.
9. Performance of simulated industrial work is worse at 10 C than at 17 C, and
worse at 24 C than at 20 C. These conclusions were drawn respectively by Pepler
(1964), from the Industrial Fatigue Research Board experiments, and by Wyon
(1974), from the report of the New York State Commission on Ventilation. Note
the excellent correspondence of these performance data with the field accident
data. Both experiments were carried out under realistic working conditions,
subjects working a full 8-hour day for several weeks. The South African
experiments discussed under Point 2 above indicate corresponding effects on the
many industrial tasks which were studied both above and below thermal
neutrality, although the temperature for optimum performance was found to lie as
much as 10 C higher for these heat-acclimatised factory workers.
Just as the effects of air pollutants may be expected to be much greater for
individuals with pre-existing allergy or hypersensitivity to specific pollutants, the
effects of heat stress on performance may be much greater for individuals with
pre-existing medical conditions that make them sensitive to heat. In what must
necessarily be a rare type of experiment, Danish patients with ischemic heart
disease were exposed to moderate heat stress (Andersen et al. 1976). Their
performance and their thermal comfort sensations were the same as those of
healthy control subjects at thermal neutrality, but were significantly lower at
raised levels of temperature. There were some differences in cardiopulmonary
function, but no differences in thermoregulation. The difference was interpreted as
being due to anxiety, a mechanism which would be even more pronounced under
uncontrolled field conditions of heat stress at work.
10. Laboratory tests of rapid skilled arm movement indicate performance
decrement at 13ø and 29 C below the level achieved at 21 C. A further and more
marked decrement takes place between 29ø and 38 C (Teichner and Wehrkamp
1954). The correspondence with field accident data is again good. A deterioration
of the ability to perform rapid skilled arm movements is logically an important
factor in accident causation and will also affect performance in some kinds of
manual work.
VEHICLE CLIMATE EFFECTS ON DRIVER PERFORMANCE
Vehicles represent a very common type of enclosed space in which people live
and work. Research in this field is therefore relevant to the present concern. It has
been shown experimentally that driver performance is reduced by the extreme
heat stress experienced in summertime Arizona in vehicles with no air
conditioning (Mackie et al. 1974). There was a significant increase in the number
of "moving violations" of the highway code, speed was more variable and drivers
looked less often in the rear view mirror. As bad driving causes accidents which
cost money, this is a decrease in productivity. In experiments in Sweden, the
present author has recently shown that driver vigilance was significantly lower at
27 C than at 21 C (Wyon et al. 1996). Subjects missed 50% more of the signals
introduced to the driving task via the instruments and rear-view mirrors of a
specially prepared test vehicle, and their responses to the signals they saw or
heard were 22% slower. 83 subjects drove the vehicle for one hour each on a
predetermined route. It was possible to show that the decrement in vigilance
performance was significantly greater in the second half-hour, and significantly
greater in town than on the open road. Taking into account the additional heat-
stress caused by unavoidable direct sunshine on vehicle occupants, 27 C
represents a level of heat stress that is exceeded even in Sweden for a large part of
the year. Air temperatures approaching 27 C may even be necessary in winter to
counteract local cold discomfort due to cold window surfaces. The conclusion
drawn from this experiment (Norin and Wyon 1992) was that "effective climate
control in vehicles and good design of the thermal environment is not a luxury,
and should not be an optional extra".
The same test vehicle was driven by a further 100 subjects for one hour each in
city traffic, with and without an air ioniser in operation (Wyon et al. 1995). The
subjects were not even aware of the presence of the ioniser, which had the
measurable physical effect of reducing the number of respirable particles in the
vehicle air, but only a negligible effect on the total dust content of the air.
Subjective eye distress as reported on visual-analogue scales was significantly less
for those 50 subjects who had been randomly assigned to the ioniser-on condition,
and there was an accompanying significant increase in driver vigilance as
measured in exactly the same way as in the previous experiment. Thus existing
thermal and air quality conditions in vehicles have measurably negative effects on
the productivity of drivers.
INDIVIDUAL CONTROL AND PRODUCTIVITY
Individual control of the indoor environment was shown above to lead to a
reduction in sick leave due to SBS and to an increase in self-estimates of
productivity. These positive effects on measures related to productivity indicate
how important it is for individuals with differing requirements of the indoor
environment to be able to make their own compromises, instead of having to
endure the compromises that are optimal for the group to which they belong.
Individuals whose eyes are particularly sensitive to air movement will trade off
this factor against air quality, either by closing windows or by reducing ventilation
rates, others with allergy or hypersensitivity will choose to accept draughts if they
are an inevitable consequence of improved air quality. The ergonomics (human
factors) of healthy buildings were discussed in some detail by Wyon (1991b). A
system that can be fitted under existing work surfaces such as office desks,
computer work stations, telephone exchanges, etc. to enable the individual to
control his or her own heat balance to an extent corresponding to ñ3 C room
temperature, and providing a personal fresh air supply to the breathing zone via an
air filter that can be changed by the individual as required, was developed by
Wyon et al. (1991) for the Swedish Board of Public Buildings. Many such devices
are now available in the US.. Wyon (1996) has estimated in some detail the likely
effects on the performance of four different tasks of providing various degrees of
individual control of the thermal environment. The calculation assumes that
individuals whose neutral temperature differs from that of the group will use
individual control to approach neutrality unless this would reduce their
performance. The predicted advantage for average group performance of
providing individual control equivalent to ñ3 C about the group optimum
temperature was 2.7% for logical thinking, 7.0% for general office work, 3.4%
for very skilled manual work, and 8.6% for very rapid manual work. The figure
for logical thinking was closely similar to what was actually observed in a field
experiment which used the rate at which claims were processed in an insurance
company as the outcome metric for task performance (Kroner et al. 1992), when
individual control units were disabled. This result is of course suggestive rather
than conclusive, as once individual control has been provided, deprivation may be
sufficiently frustrating to have a negative effect on performance. However, if one
assumes that the observed effect was mediated by a loss of individual thermal
control rather than by frustration, the agreement of observation with theory is
excellent. The positive effect of individual control on performance is predicted to
be somewhat greater if room temperatures must be maintained above or below the
group optimum temperature for comfort, for example for reasons of energy
economy.
FIELD STUDIES OF PRODUCTIVITY
Another major study of productivity in an insurance company, using daily
increases in computer file size as the dependent measure (Zyla-Wisendale and
Stolwijk 1990), found that productivity as so measured was 16% lower in winter
than in summer. No objective measures of indoor or outdoor climate were found
to have a significant effect on this measure, and there were large differences from
day to day for the same person. The significance and size of the seasonal
decrement demonstrates that large and systematic changes in productivity can take
place almost unnoticed. The study by Kroner et al. (1992) cited above compared
objective measures of productivity with self-reports of perceived productivity
during a move to new premises: actual productivity fell by 30% immediately
following the move, which is hardly surprising, before reaching new and higher
levels than before the move. It is worth noting, from a methodological point of
view, that perceived productivity did not fall at all during the move. Another
North American study of 94 government offices (EPA 1989) found that 30% of
the occupants complained of headaches, 44% of fatigue or drowsiness, 37% of
eye strain, and 69% of "reduced productivity due to poor indoor air quality". It is
against this background that it has been estimated that office productivity in the
USA could be improved by 20% by improving the indoor climate (Woods 1989).
COST-BENEFIT ANALYSES
There have been many attempts to quantify more closely the possible benefits of
improvements in the indoor environment, in relation to the cost of the
improvements. One such analysis (Holcomb and Pedelty 1992) takes as its
starting point the only known study (48) in which measurements were made both
of ventilation rates and of the incidence of Upper Respiratory-tract Infections
(URI): the relative risk of URI was 1.51 in buildings with 1 l/s/p fresh air, in
comparison with buildings with 18 l/s/p. This corresponds to a 3.6% increase in
URI frequency for each 1 l/s/p decrease in fresh air supply. National surveys
(Collins 1989) show that 9.5 days per person over 18 are lost per year due to ill
health in the USA, corresponding to 2.6% of the total, and that 50% of all acute
health conditions are caused by respiratory conditions. The derivation of these
relationships is somewhat questionable (W. Fisk, personal communication):
infiltration rates are usually much higher than the 1 l/s/p ventilation rate assumed;
only two buildings were studied, so a linear relationship is simply assumed; the
buildings were barracks, where more time per week may have been spent than in
offices; URI is less likely than most other illnesses to result in time off work; only
about 240 of 365 days per year are workdays; and so on. They were nevertheless
applied by the authors to an actual building in New York with 667 employees, and
a fresh air supply of 2.5 l/s/p. The first cost of upgrading the fresh air supply to 10
l/s/p, as recommended by ASHRAE, was calculated as $28500, with increased
running costs of $5575 per year. The above relationships lead to an expected
decrease in absenteeism whose wage value would be $70235 per year. Net savings
would be $35977 in the first year, and $64477 in each subsequent year.
The so-called "leverage" of the costs of environmental improvements on
productivity benefits has recently been quantified by Lorsch and Abdou (1994),
using published estimates of the increased construction costs, running costs and
energy costs associated with increasing ventilation capacity (Eto and Meyer
1988). Assuming that annual energy costs are $22/m2, that capital costs are
amortised at 10% over a 15-year life-cycle, and that operating costs will increase
by 15%, improvements with a capital cost of $54/m2 would be paid for by an
offsetting gain in productivity of only 0.5%, while a capital cost of $269/m2
would be paid for by a gain of 1.8%. A 50% increase in energy costs would
require a gain of only 0.25-0.5% in productivity.
Dorgan and Dorgan (1994) classified office buildings in the USA in 5 categories:
Healthy (20%), Generally Healthy (40%), Unhealthy, Cause Unknown (20%),
Unhealthy, Cause Known (10%), and Actually Causing SBS or BRI (Building
Related Illness) (10%). The total number of commercial buildings in the USA is
stated to be 4.5 106, the total floor space, not including warehouses or parking 4.5
109 m2, and the number of occupants 68.3 106. It is assumed that productivity in
healthy buildings would increase by 1%, and in unhealthy buildings by 6%, if all
were upgraded to meet current ASHRAE ventilation standards. The initial cost of
doing so is estimated to be $88 billion ($88 109), and the resulting annual
increased cost to be $4.8 109, or $1.3 per m2. The annual benefit in terms of
improved productivity is estimated to be $55 109, or $12.4 per m2. The initial
average economic payback time is estimated to be 1.6 years.
Demonstration experiments
Any proposal to demonstrate directly and rigorously in real commercial
operations that office layouts or environmental conditions have an effect on
profitability is subject to a Catch 22 restriction: although employers would like
very much to know the outcome, they are most unlikely to permit interventions
that are expected to lead to a reduction of profitability to persist for long enough
to eliminate alternative explanations. Even in experiments involving interventions
expected to increase profitability there will always be a control group, so both
designs involve a deliberate and intentional reduction of potential profitability,
with no guarantee that statistically significant results will be obtained even if the
conditions studied do have a substantial effect on individual productivity. This is
because there are innumerable other factors which affect profitability, such as
market conditions, the cost and availability of labour and raw materials, the
actions of competitors and the demands of customers, none of which is under
experimental control. Although the effects of extraneous factors will affect the
average profitability of both groups similarly, the risk is that they will so increase
the variance of the dependent measures that the expected differences of a few
percent - all that is necessary to justify quite large investments in the indoor
environment, according to the cost-benefit analyses summarised above - would
remain unproven.
Demonstration experiments of the kind recommended by Woods (1996) in the
ASHRAE 700-RP Final Report do not advance our understanding of how
productivity is affected by different aspects of the indoor environment. If they are
carried out on pre-existing groups, such as school classes or groups of employees
accustomed to working together, as they usually must be, the time course of
events is extremely vulnerable to the influence of exogenous factors acting on the
members of one group but not on another. These may be as impossible to quantify
as the the existence of a role-model in the group. Passive comparisons between
pre-existing groups are of course vulnerable to confounding: the significant
negative correlation between CO2 levels measured in Norwegian classrooms and
the health indices and test performance scores obtained from the children
occupying them, as reported by Myhrvold et al. (1996), may simply reflect
systematic and unsuspected differences between the collection areas of the
schools - parents who have ensured that there is good ventilation in their
childrens' classrooms are likely to have interested themselves in their health, diet,
attitude to school work and much else besides, and vice versa. This study is
continuing with interventions to upgrade the classrooms which were found to have
defective ventilation, as recommended by Woods (1996 op cit.). However, even if
health and performance indicators can be shown to improve in these classrooms
after the upgrades, the demonstration will fail if a similar improvement takes place
for some other reason in the well-ventilated classrooms, or if the alternative
explanation of a "Hawthorne effect" (in this case a positive behavioural response
to the perceived concern shown by the school authorities in approving the
classroom upgrades) is not rigorously ruled out by ensuring that even the children
in the well-ventilated classrooms are led to believe that they too have had their
classroom ventilation upgraded.
A new approach to productivity research
A new approach would be to formulate specific mechanisms to explain exactly
and in detail just how a given change in office layout or environmental conditions
might be expected to affect individual performance. Each mechanism should be
defined by postulating a chain of falsifiable hypotheses, each of which must be
true for the mechanism to be valid. A key concept in this approach is that of the
intervening variables, the outcome metrics for one hypothesis which become the
independent or driving variables for the next link in the chain. It is then possible
to devise experiments to test each link in the chain separately, usually by very
different means: the intervening variables may relate to the group or to an
individual, and may be indices of health, mood, motivation, comfort, behaviour or
even performance - individual performance is an important intervening variable
between IEQ and the productivity of the group. Part of the economy of this
approach lies in the fact that not all of the hypotheses in a chain may need to be
tested, as the chain is invalidated if any one of the hypotheses proves to be untrue.
However, as in the following example, there may be several alternative routes
between cause and effect, involving different chains of hypotheses.
Example: Office design and dust exposure
Claim: Clean offices affect productivity
Mechanism (1): By decreasing occupants' exposure to dust
Mechanism (1) involves the following ten relevant hypotheses, all of which are
falsifiable:
In offices, the amount of dust present is affected by -
A - the shelf factor - area of shelves per unit floor area (link to E & F)
B - the textile factor - area of textile material per unit floor area (link to E, F)
C - the storage arrangements - closed or open (link to E & F)
D - the occupant density (link to E & F)
E Dust increases sub-clinical symptom intensity (link to G)
F Dust decreases occupants' feelings of wellbeing (link to H)
G Sub-clinical symptoms - SBS - decrease individual performance (link to J)
H Feelings of wellbeing increase individual performance (link to J)
J * Work force productivity is affected by task performance (link to K)
K * Profitability is affected by work force productivity (almost by definition)
* It will often be possible to assume the truth of these hypotheses
Mechanism(1) is valid if one or more of the following eight sets of linked
hypotheses are true:
A-E-G-J-K or A-F-H-J-K
B-E-G-J-K or B-F-H-J-K
C-E-G-J-K or C-F-H-J-K
D-E-G-J-K or D-F-H-J-K
Similar sets of hypotheses may be constructed for the postulated effects of other
factors:
Claim: Occupant density affects productivity
Mechanism(2): High occupant density decreases productivity by degrading
working conditions
Links via visual & acoustic distraction, air quality, temperature, crowding
Claim: Alternative office layouts affect productivity
Mechanism(3): Open offices degrade physical working conditions
Links via visual & acoustic distraction
Mechanism(4): Open offices encourage and support team interaction
Links via team interaction, group behaviour, individual performance
Mechanism(5): Open offices decrease privacy
Links via managerial supervision, group behaviour, individual performance
Claim: Good office design affects productivity
Mechanism(6): Good office design increases subjective feelings of wellbeing
Links: privacy, view-out, visual & ergonomic design affect wellbeing,
affecting motivation, absenteeism, time-on-task & individual performance
Mechanism(7): Good office design ensures that fresh air reaches every occupant
Links: occupant density, partitions & natural ventilation affect local
ventilation, wellbeing, sub-clinical symptoms, motivation, cross-infection
& absenteeism
It can be shown that Mechanisms (1-7) involve at least 48 different falsifiable
hypotheses. The formulation of appropriate chains of hypotheses is critical to the
approach. An appropriate level of detail must also be selected: for example, as it
has already been shown by Gyntelberg et al. (1994) that shelf factor affects SBS,
the truth of hypotheses A & E could perhaps have been assumed without testing
the logically necessary additional hypothesis that office workers' dust exposures
are increased by shelf factor. This was in fact demonstrated in their studies.
OUTCOME METRICS
Measures of individual performance: Occupant contribution to productivity can
be assessed in an experiment in one or more of the following ways (with the
examples cited in this paper):
1. Simulated work - subject performs a realistic but artificial task
(Wyon 1974, Wyon et al. 1982)
2. Diagnostic tests - subject performs a test procedure unlike any real task
(Nunes et al. 1993, Myhrvold et al. 1996)
3. Embedded tasks - outcome metric derived from part of an existing task
(Wyon 1970, Wyon et al. 1995, 1996)
4. Existing measures - existing outcome metrics are made available
(Zyla-Wisendale and Stolwijk 1990, Kroner et al. 1992)
5. Absenteeism - new or existing records of sick leave are used
(Preller et al. 1990)
6. Self-estimates - subjects report their own perceived level of efficiency
(Raw et al. 1990, Kroner et al. 1992)
Measures of visual and acoustic distraction: The ability of a subject to
concentrate on a very demanding central task can be measured under different
conditions of distraction. Subjective estimates of distraction reveal only
occupants' own mental models of what is distracting, and these may be mistaken.
Measures of motivation: Motivation can be inferred from observed behaviour, or
systematically obtained by peer or managerial evaluation. Subjective estimates of
motivation made by occupants themselves reveal only their own mental models of
what is motivating.
Measures of wellbeing and of sub-clinical symptom intensity: The SIFT approach
(Wyon 1994b) is adequate for the purpose. Estimates of general and specific sub-
clinical symptom intensity are obtained by asking subjects to mark visual-
analogue scales during control and intervention conditions established without
their knowledge.
Measures of awareness of other people: Awareness of other people can be
assessed by determining what aspects of other people can be accurately stated
when the subject has just been unexpectedly deprived of contact with them, for
example by being asked to step into another room to participate in an unprepared
interview. The aspects polled might include the presence of other people, details
of their clothing, their arrivals and departures, the identity and nature of their
visitors, and their work activities, including their telephone conversations. The
necessary corroborative information should be covertly acquired by various means
prior to the interview.
Measures of privacy and of view out: Privacy can be evaluated objectively in
terms of the number of people who can see or hear each occupant. View out can
be evaluated in terms of what categories of object or activity can be seen through
each window, e.g. walls, other windows, whole buildings, trees, grass, parks, sky,
traffic, birds, animals, various human activities. These can be rank-ordered in
terms of preference by a sample of occupants, enabling a weighted assessment of
the quality of available view out to be derived.
Measures of environmental variables: Objective measures of temperature, noise,
lighting, local age of air, surface dust contamination, airborne dust and personal
dust exposure are available.
Experimental design: Measures are available for all of the variables involved in
the linked chains of hypotheses. It is therefore possible in every case to devise an
experiment in which the independent variable is altered by means of an
intervention, and the resulting effect on the dependent variable is observed and
measured. Control groups in which the dependent variable is observed and
measured under similar conditions, but with no intervention, will usually be an
advantage. If it is not possible to arrange for control groups, and for random
assignment of subjects to treatment groups, "multiple crossover design" may be
appropriate, in which the intervention is reversed and then repeated a number of
times, until the probability that the observed results could be caused by an
uncontrolled extraneous variable is small enough to be rejected. Subjects then
serve as their own controls.
Summary of the recommended approach
Specific mechanisms by which alternative office designs may reasonably be
claimed to affect productivity are expressed as sets of linked hypotheses.
Hypotheses can then be tested one by one. The necessary experiments involve a
wide variety of outcome metrics, but the crucial point is that the hypotheses are
falsifiable - they are either true or untrue, and experiments can be devised to
determine which is the case. If any one of a particular chain of hypotheses is
shown to be untrue, the mechanism described by the chain is not valid. By
disassembling the usually vague and general claims about architectural design into
the constituent parts of the mechanisms by which they are claimed to act on
productivity, the claim can be either validated or disproved. Different aspects of
office design can be quantitatively compared with each other in terms of their
relative effects on the intervening variables. This makes it possible to optimise the
design before going to the trouble and expense of demonstrating the size of the
overall effect on bottom-line productivity and profit.
CONCLUSIONS
Published experimental data indicate that conventionally acceptable indoor
working environments may be affecting human performance by various
mechanisms by as much as 5-15%. Cost-benefit analyses which assume an impact
on overall productivity of as little as 0.5% show that the payback time for a
general upgrading of currently unhealthy office buildings, defined so as to include
about 40% of the building stock, would be as low as 1.6 years. A new approach
to productivity research is suggested, by which it would be possible to compare
the impact of architectural design features on productivity with that of physical
environmental factors and to test the validity of any postulated causative
mechanism with full scientific rigour.
REFERENCES
Andersen, I., Jensen, P.L., Junker, P., Thomsen, A., and Wyon, D.P. 1976. The
effects of moderate heat stress on patients with ischemic heart disease.
Scandinavian J. Work, Environment & Health, 4, 256-272
Andersen, I., Lundqvist, G.R., Jensen, P-L and Proctor, D.F. 1974. Human
response to 78-hour exposure to dry air. Arch. of Environmental Health, 29,
319-324
Andersson, L.O., Frisk, P., Löfstedt, B. and Wyon, D.P. 1975. Human responses
to dry, humidified and intermittently humidified air in large office buildings.
Document Series, 11, 69p. Stockholm: Swedish Building Research
Brundage, J.F., Scott, R.McN., Lednar, W.M., Smith, D.W. and Miller, R.N.
1988. Building-associated risk of febrile acute respiratory diseases in army
trainees. J. Amer. Medical Ass., 259, 2108-2112
Bælum, J., Andersen, I., Lundqvist, G.R., Mílhave, L., Pedersen, O.F., V‘th, M.
and Wyon, D.P. 1985. Response of solvent exposed printers and unexposed
controls to six-hour toluene exposure. Scandinavian J. Work, Environment
& Health, 11, 271-280
Clarke, R.S.J., Hellon, R.F. and Lind, A.R. 1958. The duration of sustained
contractions of the human forearm at different muscle temperatures. J. Appl.
Physiol., 143, 454-473
Collins, J.G. 1989. Health characteristics by occupation and industry: United
States 1983-1985. Vital Health Statistics., Vol.10, No.170. Hyattsville,
Maryland: National Center for Health Statistics
Dorgan, C.E. and Dorgan, C.B. 1994. Link between the indoor environment and
productivity and health issues. Report from Dorgan Associates, Madison,
Wisconsin.
EPA. 1989. Report to Congress on indoor air quality. Vol.II: Assessment and
control of indoor air pollution. Office of Air & Radiation (ANR-445),
Environmental Protection Agency of the USA
Eto, J.H., and Meyer, C. 1988. The HVAC costs of fresh air ventilation. ASHRAE
Journal, 30(9), 31-35
Gravesen, S., Skov, P., Valbjørn, O. and Løwenstein, H. 1990. The role of
potential immunogenic components of dust (MOD) in the Sick-Building
Syndrome. Indoor Air '90, Toronto, Proceedings Vol. 1, 9-13
Gyntelberg, F., Suadicani, P., Nielsen, J.W., Skov, P.S., Valbjørn, O., Nielsen,
P.A., Schneider, T., Jírgensen, Wolkoff, P., Wilkins, C.K., Gravesen and S.,
Norn, S. 1994. Dust and the Sick Building Syndrome. Indoor Air, 4, 223-238
Holcomb, L.C., and Pedelty, J.F. 1992. Comparison of employee upper
respiratory absenteeism costs with costs associated with improved
ventilation. Report from Holcomb Environmental Services, Olivet,
Massachusets
ISO 7730. 1984. Moderate thermal environments - determination of the PMV and
PPD indices and specification of the conditions for thermal comfort. Geneva:
International Organization for Standardization
Jaakkola, J.J.K., Heinonen, O.P. and Sepp„nen, O. 1989. Sick Building
Syndrome, sensation of dryness and thermal comfort in relation to room
temperature in an office building: need for individual control of air
temperature. Environment International, 15, 163-168
Kok, R., Lewis, M.I., Meese, G.B. and Wyon, D.P. 1983. The effects of moderate
cold and heat stress on factory workers in Southern Africa. 4: Skill and
performance in the heat. South African J. Science, 78, 306-314
Krogstad, A-L., Swanbeck, G., Barregård, L., Hagberg, S., Rynell, K.B., Ran, A.
and Jorulf, L. 1991. Besvär vid kontorsarbete med olika temperaturer i
arbetslokalen - en prospektiv undersökning (A prospective study of indoor
climate problems at different temperatures in offices, in Swedish). Göteborg,
Sweden: Volvo Truck Corporation
Kroner, W., Stark-Martin, J.A. and Willemain, T. 1992. Using advanced office
technology to increase productivity: the impact of environmentally
responsive workstations (ERWs) on productivity and worker attitude. Troy,
New York: Rensselaer Polytech Inst, Center for Architectural Research
Lorsch, H.G. and Abdou, O.A. 1994. The impact of the indoor environment on
occupant productivity, Part 1: Recent studies, measures and costs. Report
from Lorsch Associates
Mackie, R.R., O'Hanlon, J.F. and McCauley, M. 1974. A study of heat, noise and
vibration in relation to driver performance and physiological strain. Report
DOT HS-801-315, Human Factors Research Inc., NTIS Catalog Nr. PB 238
829
McIntyre, D.A. 1980. Indoor Climate. London: Applied Science
Meese, G.B., Kok, R., Lewis, M.I. and Wyon, D.P. 1982. Effects of moderate cold
and heat stress on factory workers in Southern Africa, 2: skill and
performance in the cold. South African J. Science, 78, 189-197
Myhrvold, A.N., Olsen, E. and Lauridsen, 0. 1996. Indoor environment in schools
- pupil health and performance in regard to CO2 concentrations. Indoor Air
'96, Nagoya, Proceedings Vol. 4, 369-374
New York State Commission on Ventilation. 1923. Report of the New York State
Commission on Ventilation. New York: Dutton
Norin, F. and Wyon, D.P. 1992. Driver vigilance: the effects of compartment
temperature. In: Design & Performance of Control Systems, SAE SP-916, 7-
11. Detroit: Society of Automotive Engineers
Nunes, F., Menzies, R., Tamblyn, R.M., Boehm, E. and Letz, R. 1993. The effect
of varying level of outdoor air supply on neurobehavioural performance
function during a study of Sick Building Syndrome (SBS). Indoor Air '93,
Helsinki, Proceedings Vol. 1, 53-58
Olesen, B.W., Schíler, M. and Fanger, P.O. 1979. Discomfort caused by vertical
air temperature differences. In: Fanger, P.O., and Valbjírn, O., eds., Indoor
Climate. Copenhagen: Danish Building Research Institute, 561-579
Pepler, R.D. 1958. Warmth and performance: an investigation in the tropics.
Ergonomics, 2, 63-88
Pepler, R.D. 1964. Psychological effects of heat. In: Leithead, C.S and Lind, A.R.,
eds., Heat stress and heat disorders., 237-271. London: Cassell
Pepler, R.D. and Warner, R.E. 1968. Temperature and learning: an experimental
study. ASHRAE Transactions, 74, 211-219
Poulton, E.C. 1970. Environment and human efficiency. Springfield, USA:
Thomas
Preller, L., Zweers, T., Brunekreef, B. and Bolej, J.S.M. 1990. Sick leave due to
work-related health complaints among office workers in the Netherlands.
Indoor Air '90, Toronto, Proceedings Vol. 1, 227-230
Rasmussen, O.B. 1971. Man's subjective perception of air humidity. 5th
International HVAC Conference, May 1971, Copenhagen, Proceedings Vol.
1, 79-86
Raw, G.J., Roys, M.S. and Leaman, A. 1990. Further findings from the office
environment survey: productivity. Indoor Air '90, Toronto, Proceedings Vol.
1, 231-236
Teichner, W.H. and Wehrkamp, R.F. 1954. Visual-motor performance as a
function of short-duration ambient temperature. J. Exp. Psychology, 47, 447-
450
Vernon, H.M. 1936. Accidents and their prevention. Cambridge, England:
University Press
Wilkinson, R.T., Fox, R.H., Goldsmith, R., Hampton, I.F.G. and Lewis, H.E.
1964. Psychological and physiological responses to raised body temperature.
J. Appl. Physiol., 19, 287-291
Woods, J.E. 1989. Cost avoidance and productivity in owning and operating
buildings. Occupational Medicine: State of the Art Reviews, 4, 753-770
Woods, J.E. 1996. Research protocols for evaluating the relationship between the
indoor environment and productivity. ASHRAE Research Project 700-RP.
Atlanta: ASHRAE TC2.1
Wyon, D.P. 1970. Studies of children under imposed noise and heat stress.
Ergonomics, 13, 598-612
Wyon, D.P. 1974. The effects of moderate heat stress on typewriting performance.
Ergonomics, 17, 309-318
Wyon, D.P. 1992a. Sick buildings and the experimental approach. Environmental
Technology, 13, 313-322
Wyon, D.P. 1992b. Barriers to productivity. Building Services, The CIBSE
Journal, 14, 29-31
Wyon, D.P. 1993. Healthy buildings and their impact on productivity. Indoor Air
'93, Helsinki, Proceedings Vol.6, 3-13
Wyon, D.P. 1994a. Thermal gradients, individual differences and air quality.
Healthy Buildings '94, Budapest, Proceedings Vol. 2, 765-770
Wyon, D.P. 1994b. Symptom Intensity Feedback Testing (SIFT): behavioural
science may be able to provide the key to curing sick buildings. Healthy
Buildings '94, Budapest, Proceedings Vol. 3, 42-47
Wyon, D.P. 1996. Individual microclimate control: required range, probable
benefits and current feasibility. Indoor Air '96, Nagoya, Proceedings Vol. 1,
1067-1072
Wyon, D.P., Andersen, I.B., and Lundqvist, G.R. 1979. The effects of moderate
heat stress on mental performance. Scandinavian J. Work, Environment &
Health, 5, 352-361
Wyon, D.P., Andersson, B. and Söderling, M. 1991. Field trials of technical
measures to reduce Sick Building Syndrome (In Swedish, with English
Summary). Malm”, Sweden: Malm” General Hospital, 80p
Wyon, D.P., Fanger, P.O., Olesen, B.W., and Pedersen, C.J.K. 1975. The mental
performance of subjects clothed for comfort at two different air temperatures.
Ergonomics, 18, 359-374
Wyon, D.P., Kok, R., Lewis, M.I., and Meese, G.B. 1982. Effects of moderate
cold and heat stress on the performance of factory workers in Southern
Africa, 1: Introduction to a series of full-scale simulation studies. South
African J. Science, 78, 184-189
Wyon, D.P. and Sandberg, M. 1990. Thermal manikin prediction of discomfort
due to displacement ventilation. ASHRAE Transactions 96(1), Paper 3307
Wyon, D.P., Sandberg, M. 1996. Discomfort due to vertical temperature
differences. Indoor Air, 6, 48-54
Wyon, D.P., Thews, J. and Larsson, S. 1991. Individual micro-climate control for
office workers: a case study. International Benchmark Network Conference,
Stockholm: National Swedish Board of Public Building (KBS)
Wyon, D.P., Wyon, I. and Norin, F. 1995. The effects of negative ionisation on
subjective symptom intensity and driver vigilance in a moving vehicle.
Indoor Air, 5, 179-188
Wyon, D.P., Wyon, I., and Norin, F. 1996. The effects of moderate heat stress on
driver vigilance in a moving vehicle. Ergonomics, 39, 61-75
Zyla-Wisendale, N.H., and Stolwijk, J.A.J. 1990. Indoor air quality as a
determinant of office worker productivity. Indoor Air '90, Toronto,
Proceedings Vol. 1, 249-254
_____________________________________________________________
Keynote Address to "IAQ '96", Baltimore, 6-8 October 1996