Printer Friendly

The effects of moderately raised classroom temperatures and classroom ventilation rate on the performance of schoolwork by children (RP-1257).

Two independent field intervention experiments were carried out in school classrooms in late summer (in 2004 and 2005). The air temperature was manipulated by either operating or idling split cooling units installed for the purpose. In one of these experiments, the outdoor air supply rate was also manipulated. The conditions were established for one week at a time in a blind crossover design with repeated measures on two classes of 10- to 12-year-old children. Six to eight exercises exemplifying different aspects of schoolwork (numerical and language-based) were performed as part of normal lessons. Pupils indicated their environmental perceptions and the intensity of any symptoms on visual analogue scales. Their thermal sensation changed from slightly too warm to neutral, and the performance of two numerical and two language-based tests was significantly improved when the temperature was reduced from 25[degrees]C to 20[degrees]C (77[degrees]F to 68[degrees]F). When the outdoor air supply rate was increased from 5.2 to 9.6 L/s (11.0 to 20.3 cfm) per person, their performance of four numerical exercises improved significantly, confirming the results of previously reported experiments in the same series. The above improvements were mainly in terms of the speed at which tasks were performed, with negligible effects on error rate. Most school classrooms worldwide experience raised air temperatures during increased thermal loads, e.g., in warm weather; these results show that providing some means of avoiding elevated temperatures would improve educational attainment.

INTRODUCTION

Unsuitably high temperatures are common in classrooms, even those in cold countries. For example, a survey of temperatures in a large number of schools in Sweden showed that classroom temperatures were generally 23[degrees]C-25[degrees]C (73.4[degrees]F-77[degrees]F) in the shoulder seasons (April to September), 3[degrees]C-6[degrees]C (5[degrees]F-10[degrees]F) above what teachers and pupils preferred. Some classroom temperatures were as high as 30[degrees]C (86[degrees]F), which is quite remarkable for such a cold country (Eriksson et al. 1967). The most common reason for such high temperatures is that classroom ventilation rates are too low to remove the excess heat load caused by sunshine entering the windows, which until relatively recently were traditionally designed to provide as much daylight as possible, with large glazed areas that faced the sun. This is especially the case in the many schools that have only natural ventilation, as windows must often remain closed to exclude external noise and prevent draft, but it may also be the case in schools with mechanical ventilation and no cooling.

Very few data are available on thermal effects on the performance of schoolwork by children. A recent wide-ranging and authoritative review of research by Mendell and Heath (2005) of the factors that might influence student performance found only one peer-reviewed study of how the air temperature in classrooms affects schoolchildren's performance (Schoer and Shaffran 1973). These authors reported three experiments in which 10- to 12-year-old pupils in matched pairs were assigned either to a classroom without cooling (where the temperature was about 26[degrees]C [78.8[degrees]F]) or to an adjacent air-conditioned classroom (where the temperature was about 22.5[degrees]C [72.5[degrees]F]). The classrooms were especially built for the purpose. Each group then worked in the same classroom every school day for six to eight weeks. Nineteen different tests were applied, ranging from very simple and repetitive tests (such as crossing out certain letters in a text) to school exercises stated to be current at the time (such as coding numbers onto machine-readable punched cards), and the students' performance was significantly better in the classroom that was always cool, on average by 5.7%. However, the subjects knew they were taking part in an experiment (because they were taken by bus each day to the experimental classrooms and instructed by experimenters who were not their normal class teachers) and knew when they were being tested (because each test was performed under maximum effort conditions and timed with a stop-watch), and by talking to each other over six to eight weeks, they must have known that there was a difference in temperature between the two classrooms. This means that the observed difference in performance could have been due to a gradual process of discouragement and growing resentment between two groups of pupils. This interpretation is supported by the original authors' own analysis showing that the difference in performance between the groups increased over time, while the parallel processes of acclimatization, familiarization, and learning would all be expected to reduce over time the negative effects of temperature on performance.

Mendell and Heath (2005) did not review the comprehensive set of experiments on the effects of classroom temperature on the performance of schoolwork that was carried out in the 1960s and 1970s in Sweden, probably because the only report of them in an archival peer-reviewed journal (Wyon 1970) was a brief summary of their findings. In these experiments, three parallel classes of 9- to 10-year-old children were exposed for two hours to each of three classroom temperatures-20[degrees]C, 27[degrees]C, and 30[degrees]C (68[degrees]F, 80.6[degrees]F, and 86[degrees]F), encountered in balanced order, and four classes of 11- to 12-year-old children were similarly exposed to 20[degrees]C and 30[degrees]C (68[degrees]F and 86[degrees]F) in the morning and the afternoon in a 2 x 2 design, again in balanced order of presentation of conditions (Holmberg and Wyon 1969). The temperatures were artificially raised in these experiments, while in the study by Schoer and Shaffran (1973) they were artificially reduced. The children performed a number of school exercises, including numerical tasks (addition, multiplication, number-checking) and language-based tasks (reading and comprehension, supplying synonyms and antonyms) so that their rate of working and the number of errors they made could be quantified. The children's performance of both types of task was significantly lower at 27[degrees]C and 30[degrees]C (80.6[degrees]F and 86[degrees]F), in comparison to 20[degrees]C (68[degrees]F). In the numerical tasks, the effect was on rate of working, but reading comprehension as well as reading speed were reduced by raised temperatures. Performance tended to be lower, though not significantly lower, at 27[degrees]C (80.6[degrees]F) than at 30[degrees]C (86[degrees]), and the negative effects of raised classroom temperatures were significant in the afternoon, when the children were fatigued, but not in the morning. The magnitude of the negative effect of temperature on performance was much larger in this study than was found in the study by Schoer and Shaffran (1973), often as great as 30%. The appearance and behavior of the children were systematically observed in these studies from behind one-way glass, and both were significantly affected by raised classroom temperature (Holmberg and Wyon 1972): the children became visibly hot but were very slow to adjust their clothing; girls became restless but continued to work, while boys began to behave in an undisciplined way and could be seen to concentrate less well. In another experiment in the same series, carried out in a language laboratory rather than a classroom, significant and negative effects of artificially raising the temperature from 20[degrees]C to 27[degrees]C (68[degrees]F to 80.6[degrees]F) could be shown when the children had to listen and speak a word, though not when they were listening and writing (Ryd and Wyon 1970). In a fourth experiment, performed this time in a climate chamber in England, in which groups of four 12-year-old boys were exposed to 20[degrees]C, 23.5[degrees]C, and 27[degrees]C (68[degrees]F, 74.3[degrees]F, and 80.6[degrees]F) in balanced order, no effects of the intermediate temperature could be shown (Wyon 1969), while the highest temperature caused children to perform schoolwork more slowly and to complete a diagnostic test of cue-utilization (the Tsai-Partington test) more rapidly, indicating that raised temperatures reduce arousal or alertness.

The results of the studies summarized above suggest that increased classroom temperatures can have negative effects on the performance of schoolwork by children. However, they were all obtained nearly four decades ago, and the results differ in terms of the magnitude of the effects and yield little information on how far below 27[degrees]C (80.6[degrees]F) it is possible to extrapolate the findings. Mendell and Heath (2005) concluded that no other studies on this issue have been carried out since then, probably because the main focus of indoor environmental research has been on thermal effects on the performance of office work by adults. This research was recently reviewed by Wyon and Wargocki (2006a), who concluded that thermal discomfort distracts attention and generates complaints, while warmth lowers arousal, exacerbates sick building syndrome (SBS) symptoms, and has a negative effect on mental work. The same effects may also be expected to occur for children and their performance of schoolwork, and children may be more affected by environmental effects even though they typically complain less about them. The present experiments were therefore designed to determine whether avoiding elevated temperatures in classrooms can improve the performance of schoolwork by children, and if so, by how much. In addition, the present experiments investigated the effects of increased outdoor air supply rate on the performance of schoolwork by children as a continuation of two other experiments in the same series, reported in a separate paper by Wargocki and Wyon (2007).

METHODS

Experimental Design

This study was designed as a series of field experiments in existing classrooms occupied by children performing their normal schoolwork. This was more natural for children than transporting them to a laboratory where they might behave abnormally, e.g., exert extra effort to perform well. Two experiments in which the air temperature in classrooms was manipulated were performed in the present series, all of them in the same school in Denmark, which is situated in the cool temperate area of Northern Europe; two experiments in which the outdoor air supply rate to classrooms and filter condition were manipulated are reported in another paper (Wargocki and Wyon 2007). The two experiments reported here were both crossover experiments in pairs of classrooms, in which two air temperatures were imposed in the same week, one in each adjacent classroom. The temperature conditions were switched between the classrooms the following week (crossover design). One experiment (Experiment 1T) was a 2 x 2 design in which each air temperature was re-imposed but with a different outdoor air supply rate. In Experiment 2T, the air temperature was changed but the outdoor air supply rate remained constant. Both experiments were performed in late summer in two successive school years, and the supply air filters were always new. Both experiments were performed as repeated-measures designs, i.e., the comparisons between conditions were always within-subject comparisons, to eliminate any bias due to individual differences in the ability to perform schoolwork. The sequence of exposures is shown in Table 1. During the experiments, the teachers and pupils were allowed to open the windows and doors as usual, and no changes in the schedule of normal school activities were made, so as to maintain the teaching environment and routines as normal as possible. The interventions were all improvements to existing conditions and were approved by parents, teachers, the School Board, the responsible local authority, and the Danish Ethics Review Board once this had been satisfactorily explained. Children were not asked for their consent so that they would remain unaware that they were taking part in an experiment in which classroom temperature and ventilation rate were being manipulated.

School, Classrooms, and Ventilation

The school is located in Denmark. It is an elementary public school for children aged 6 to 16 years and is run by the local authority. The school was selected for the experiment partly because it had six identical classrooms, partly also because energy conservation had led to fan speeds being reduced to well below their design level, and partly because large fenestration facing south led to large solar heat gains that considerably increase classroom temperatures (Figure 1). The school buildings were constructed in the 1950s and are made of bricks and concrete with large glazed areas; smoking is not allowed. Mechanical ventilation was installed in 1997. The two classrooms used in the present experiments were part of a row of six identical wings opening off the same straight north-south corridor, all with cathedral-height ceilings and large glazed south-facing facades with five openable windows. Each classroom had a floor area of 65 [m.sup.2] (699.7 [ft.sup.2]) and a volume of 187.5 [m.sup.3] (6621.5 [ft.sup.3]). The classrooms have typical school furniture and floors covered with linoleum; outdoor clothing is left on hooks in the corridor, just outside the classrooms. No cooling was available in the classrooms. They are supplied with 100% outdoor air, filtered (F7 class bag filters) and pre-heated from a central air-handling unit (AHU) situated in the basement; there is a cross-flow plate heat exchanger in the AHU for heat recovery. The intermittent operation of the AHU (the system was on nine hours per day and off during weekends and school holidays) is controlled by a computer. Each pair of classrooms in a given experiment was supplied with outdoor air from the same AHU. Pre-existing vertical brick shafts are used to transport the air from the basement, where the AHU is situated, to the classrooms. Supply air entered each classroom through supply grilles located in the wall above the openable windows and left through exhaust grilles close to the floor in the west wall adjacent to the corridor (Figure 1). Further details of the school, classrooms, and ventilation are given by Wargocki and Wyon (2007).

Interventions

To reduce the classroom temperature, wall-mounted split-unit air conditioning was installed in each classroom, consisting of an outdoor unit, situated on the roof, connected to two low-noise indoor units installed on the walls perpendicular to the south facade, above the height of the ventilation inlet grilles. Two indoor units were installed to keep the noise level as low as possible. The maximum capacity of the cooling system was 6 kW, but the units were always operated at low speed to reduce noise (about 25-30 dB(A) per unit according to their specifications), thus the actual capacity was only 5 kW. The cooling capacity required was estimated by calculating the heat loads from occupants and the sun for the period of the year for which the experiments were scheduled (August-September). The capacity installed was estimated to be sufficient to keep classroom temperature at 20[degrees]C (68[degrees]F) with outdoor temperatures up to 30[degrees]C (86[degrees]F) and windows closed. The fans of the indoor units of the split air-conditioners were operated continuously, independently of whether the cooling was on or off, to create a placebo condition. The cooling units were operated on the weeks when the temperature was to be reduced according to the study plan. Otherwise the temperature in the classrooms was dependent on the weather conditions. No heating was installed to increase indoor temperatures, to maintain the realism of each exposure. However, this meant that in this condition the temperature increased during the school day, reaching its maximum in the afternoon. The temperature was fairly constant in the condition with reduced temperature in the classrooms. Using continuous measurements of temperatures in the classrooms, daily average temperatures were estimated for the periods when children were present in the classrooms, according to their class schedule (short breaks between lessons were excluded). These were then averaged to produce weekly average indoor air temperatures for each classroom, while the standard deviation of the weekly average was calculated from the temperatures measured during a week in the periods when the children were present in the classroom; they are reported in Table 2 in the "Results" section of this paper.

[FIGURE 1 OMITTED]

The AHU was modified for the experiments by fitting a larger fan motor with a frequency controller and by fitting manually controlled dampers that made it possible to more easily rebalance the distribution of supply air between the classrooms; IRIS dampers were fitted to make it possible to estimate the airflow with an accuracy of [+ or -]7% by measuring the pressure drop with a digital micro-manometer that had an accuracy of [+ or -]1% (Wargocki and Wyon 2007).

The outdoor air supply rate was double-checked using a venturi hood with a fan that compensated for pressure drop and a damper (a method with an accuracy [+ or -]5% of the reading). Prior to the interventions, the outdoor air supply rate to each classroom was determined to be 180 [m.sup.3]/h (105.9 cfm), although the ventilation system was designed for an outdoor air supply rate of 600 [m.sup.3]/h (353.1 cfm) to each classroom to meet the requirements of the Danish Building Regulations (DHBA 1995) for a classroom occupied by 30 persons. The much lower outdoor air supply rate that was found could have been due to the energy conservation measures that had been implemented some years previously or to small defects in the AHU. The outdoor air supply rate was changed during the experiment by increasing the fan speed and rebalancing the system. A maximum rate of 800 [m.sup.3]/h (470.9 cfm) (9.7 L/s per person [20.6 cfm/person]) could be achieved. The limiting factors were the small cross-sectional area of the brick shaft through which the supply air was brought from the basement, the bends in the ducting, and the resulting increase in the airflow noise level. The reference outdoor air supply rate was maintained at 180 [m.sup.3]/h (105.9 cfm) (2.2 L/s per person [4.7 cfm/person]). The air supply rates were double-checked each week after rebalancing, using both the hood method and IRIS dampers with a micro-manometer.

The actual effective ventilation rates in the classrooms were estimated with a general mass balance equation (McIntyre 1980) using the measurements of C[O.sub.2] made when pupils were in the classrooms. As there was no restriction on window opening, this method was selected to include the contribution of infiltration. The error of this method was estimated to be [+ or -]10% (Taylor 1997). Theoretical buildup of C[O.sub.2] concentration was fitted to the measured buildup by adjusting the assumed air change rate and the assumed production rate of C[O.sub.2] per person, which was allowed to vary between 15 and 20 L/h (0.009-0.012 cfm) as there are few data on the production of C[O.sub.2] by children of this age and the available data suggest that C[O.sub.2] produced by children is similar to what is produced by adults, most probably due to the higher activity level of children (Pejtersen et al. 1991; ECA 1992). The number of children in the classroom was obtained from the records kept by the teachers. Minimizing the square-root errors describing the difference between the theoretical and measured buildup of C[O.sub.2] was the criterion for a good fit.

As many estimations as possible were derived for each day, depending on the available C[O.sub.2] data. They were averaged to obtain daily effective ventilation rates. These were then averaged to produce weekly effective ventilation rates for each classroom and used to calculate the standard deviation of the weekly ventilation rate, which is reported in Table 2 in the "Results" section. The standard deviation is stated as an estimate of the uncertainty of the quoted average values because in addition to the instrumental error it includes all the chance factors that will undoubtedly have introduced unexpected variation, such as the door opening, sudden gusts of wind, etc. Such sources of variation would not be included in a conventional uncertainty estimate that was based only on instrumental accuracy. From the measurements of ventilation effectiveness in classrooms with children absent and from continuous measurements of the C[O.sub.2] concentration at two locations in the classrooms with children present (Figure 1), the air in the classrooms was judged to be well mixed.

Physical Measurements

Silicon-based nondispersive infrared sensors for the measurement of gaseous C[O.sub.2] (accuracy [+ or -][30 ppm + 2% of the reading]) were connected to miniature battery-powered data loggers and used to monitor C[O.sub.2] levels every 1-5 minutes in each classroom at a (childproof) height of 2.2 m (7.2 ft) (accuracy of recording the signal from the C[O.sub.2] sensors was [+ or -]1% of full scale). Three similar data loggers were used to monitor temperature (accuracy [+ or -]0.7[degrees]C at 21[degrees]C [[+ or -]1.27[degrees]F at 70[degrees]F]) and relative humidity (accuracy [+ or -]5%RH) continuously in each classroom at a height of 2.2 m (7.2 ft). Similar data loggers were placed in the supply and exhaust ducts of each classroom. State loggers were used to record when any of the windows or the entrance door was open. Those mounted on the doors proved unreliable due to the frequent heavy shock of closing doors; thus, no usable record of door opening is available. The rate of dust sedimentation onto horizontal surfaces was measured each week by placing clean glass plates on existing rails or on small brackets on the walls perpendicular to the windows at a height of 2.2 m (7.2 ft). A surface dust meter was used to assess the percentage of the surface covered by dust at the end of the week. This is accomplished by using forensic gelatin tape to lift the dust and then inserting the tape into an instrument that measures the amount of light scattered from a laser beam. Spot measurements of airborne particle density were made for 20 minutes at the end of each week, after the children had left the classroom, using a dust monitor; the size ranges that could be assessed were > 0.75, > 1, > 2, > 3.5, > 5, > 7.5, > 10, and > 15 [micro]m (sensitivity 1 particle/L and reproducibility [+ or -]2%). An ultrafine particle counter, measuring in the size range of 0.02-1 [micro]m, was also used. At the same time, spot measurements of air velocity (accuracy [+ or -][0.05 + 0.05 of the reading] m/s), noise (accuracy [+ or -]2 dB(A)), and operative temperatures (accuracy [+ or -]0.3[degrees]C [[+ or -]0.5[degrees]F]) were made in the classrooms. The instruments were calibrated before use. Weather data for the whole period were registered.

Measurements of Performance

Each week, in appropriate lessons, the children's usual teachers administered parallel versions of language-based and numerical performance tasks representing different aspects of schoolwork, from reading to mathematics. The presentation of tasks was distributed fairly evenly over the whole week, and the teachers were asked to apply the same task always on the same weekday. No more than one task was performed during one lesson, and generally no more than two to three tasks were performed per day. The tasks were selected so that they could be a natural part of an ordinary school day. They included: (1) addition--the pupils added two four-digit numbers; (2) multiplication--the pupils multiplied two-digit numbers by three-digit numbers; (3) subtraction--the pupils subtracted two four-digit numbers; (4) number comparison--the pupils checked columns of two seven-digit numbers against each other, the numbers being made similar or different by rotating three of the digits in the middle; (5) logical thinking (i.e., grammatical reasoning)--the pupils categorized statements describing the order of the letter pairs AB and BA as True or False (Baddeley 1968); (6) acoustic proofreading--while listening to a recorded voice reading a text aloud, the pupils read the text, marking the inserted errors (10 errors were inserted per page of the transcript in such a way that they could not be found without listening and reading simultaneously); (7) reading and comprehension--the pupils read text with choice points inserted (to determine whether the children understood the text, they had to mark one of three different words at each choice-point; all three words were correct in the immediate context of the phrase into which they had been inserted, but only one was correct in the context of the whole text); and (8) proofreading--the pupils read a prepared text in which four different kinds of errors had been inserted: spelling errors, two kinds of grammatical errors (one obvious in the context of the phrase in which it occurs and one correct in this context but incorrect in the wider context of the preceding text), and logical errors. Following complaints by some parents that the children were being "tested unusually often," Tasks 2, 4, and 8 were omitted in the second experiment.

The tasks were specially developed so that their difficulty was appropriate to the age of the children, in consultation with the class teachers. In developing these tasks, the aim was that they should resemble standard teaching material as closely as possible. The mathematical calculations were familiar to the children, but the form of the other tasks was new to them. The teachers taught the children how to perform the tasks by working through examples with the class to make sure that the children understood each task. The duration of the tasks was short enough to ensure that children could not complete them in the time available. Up to 10 minutes were allocated for each task, except in the case of acoustic proofreading, where sometimes up to 16 minutes were used. If any of the pupils completed the tasks before the allocated time, all other pupils were immediately told by the teacher to stop working, and the actual time that had been available for performing the task was noted. Different versions of each test were prepared and versions were confounded with occasions (i.e., first to fourth week). Performance was measured in terms of speed (how quickly each pupil worked per unit time) and errors (the percentage of errors that were committed); in the case of proofreading, false positives were also recorded. In the case of acoustic proofreading, only the errors and false positives were recorded because the speed of performance was imposed by the rate at which the text was dictated. The children's performance was first analyzed using a complete design analysis, i.e., analyzing for each exercise only the results obtained from those pupils who had taken that test in all conditions and then repeating the analysis using all available data, i.e., including the performance of pupils who had not performed the exercise in all conditions (incomplete design). If performance differed significantly between occasions, disregarding the interventions--perhaps due to learning, increased familiarity with the exercise, fatigue, or differences between test versions--the analyses were repeated after adjusting the results for this effect. The adjustment was made by multiplying the individual performance of each pupil on a given task in that week by a coefficient calculated as the ratio of the average performance of that pupil's class on that task in the first week this task was introduced to the class by the average performance of the pupil's class on that task in that week.

Measurements of Perceptions and Symptoms

The children marked visual analogue (VA) scales each week during the last lesson each Friday to indicate the intensity of various SBS symptoms and their perceptions of the environment. Each scale was a 100 mm (3.9 in.) horizontal line with end-labels describing the perception or symptom intensity (Figure 2). The distance of the mark from one end of the scale was recorded in millimeters (0-100). The items on the VA scales included the following: the perception of classroom temperature, air movement, air dryness, air freshness, illuminance and noise, and the symptoms of nose congestion, throat, lip, and skin dryness, eyes hurting, hunger, fatigue, sleepiness, and headache. The students also indicated whether they had slept badly or too little the preceding night and whether they felt like working on the day the VA scale was marked. The VA scales were administered by the teachers, who, by going through examples, also taught the children how to use them. This was done to make sure that the children understood how to use the scales. As in the analysis of performance, the results obtained on the VA scales were first analyzed using complete design analyses, i.e., each scale was first analyzed using only the results of those pupils who had marked that scale in all conditions, and then using the ratings of all pupils, including results from those who had not marked this scale in all conditions (incomplete design).

[FIGURE 2 OMITTED]

Observational Checklists and Parental Logbooks

Each week the teachers carried out checklist observation of the children's behavior. The list included the following items (Wyon and Holmberg 1972): working very hard, looking around, closing eyes, playing with things, happy, apathetic, rocking the chair, disobedient, making too much noise, talking to a neighbor, teasing others, looking pale, supporting head with hands, and coughing/sneezing. The list was marked while pupils performed one of the tasks. For each item on the list, the number of different pupils noted by the teachers as exhibiting this behavior was counted and then grouped by condition and teacher. Parents and teachers recorded their observations of children's health, mood, and changes in behavior in logbooks. They were asked to complete the logbook every day during the experiments. In addition, the teachers recorded in their logbooks the time when each task had been applied and its duration.

Measurements of Perceived Air Quality

A sensory panel of adults was recruited to assess the air quality in the classrooms after the pupils had gone home, so as to not disturb normal school activities. The measurements were made once a week, in the afternoon, about 1-1.5 hours after the pupils had left the classrooms. The panel entered the classrooms every 2-3 minutes in groups of two to three at a time and assessed the air quality immediately upon entering; the doors were closed during these assessments and the order of assessments was balanced. Between assessments the panel members stayed in a well-ventilated rear entrance hall to the school, opening off the corridor leading to the classrooms. Subjects assessed air quality using four scales: continuous acceptability scale (Wargocki 2004), odor intensity (Yaglou et al. 1936), and two horizontal VA scales describing the freshness and dryness of classroom air. The ratings from the scales were digitized using the following coding: clearly acceptable = 1, clearly not acceptable = -1; no odor = 0, overpowering odor = 50; endpoints on VA scales were coded 0 and 100. The sensory panel approach was used only in Experiment 2T.

Statistical Analysis

A commercially available statistical software package was used to analyze the data. Shapiro-Wilk's test was used to test whether residuals were normally distributed, and if necessary the data were log transformed. When the normality assumption was fulfilled, repeated measures 2 x 2 analysis of variance (ANOVA) were performed for pupils for whom data from all four conditions were available. A general linear model (GLM) with Type V sum of squares was also used for all available data, i.e., including data from pupils for whom no data were available in some conditions. Friedman's two-way nonparametric ANOVA was used when the normality assumption was not met, for the markings on VA scales, and when analyzing the observational checklists. Wilcoxon's matched-pairs signed-ranks test was used to analyze the main effects of ventilation and temperature when the assumption of normality was not valid: performance and ratings on VA scales were averaged for one condition independently of the other one to form pairs of observations separately for each pupil, as is the case in 2 x 2 ANOVA. The averaging was carried out accepting only those pupils for whom ratings were available in all the conditions tested and then repeated accepting all ratings. The Wilcoxon test was always used to test differences between conditions in Experiment 2T. In all statistical tests, the changes in responses were analyzed at the individual level, i.e., using pupils (subjects) as their own controls. The P-level for rejection of the null hypothesis was set to 0.05 (2-tail), meaning that in 100 tests only 5 would appear to be significant by chance.

RESULTS

Experiment 1T

In Experiment 1T, the temperature was reduced and the ventilation rate was altered in a 2 x 2 design balanced for order of presentation. Table 2 shows the results of continuous measurements of temperature, relative humidity, C[O.sub.2], and window opening behavior in the classrooms in the course of the experiment. The values shown describe the periods when children were present in the classrooms (excluding even short breaks between classes). Table 2 shows that the air temperature in the classrooms was about 20[degrees]C (68[degrees]F) when cooling was provided and 23.6[degrees]C (74.5[degrees]F) in the warmer reference condition, so that the difference was 3.6 K (6.5[degrees]F). Average maximum temperatures in the two conditions differed by 3.9 K (7.0[degrees]F). Due to lower than normal outdoor air temperatures and opening of windows, the high temperature in the classrooms with high outdoor air supply rate was lower than expected. The effective ventilation rate estimated from measurements of C[O.sub.2] by using a mass balance equation was about 5 L/s per person (10.6 cfm/person) and 9.5 L/s per person (20.1 cfm/person) at the two outdoor air supply rates. The relative humidity was about 50%-55%, higher at lower air temperature, as expected. The windows were opened more often at the higher temperature, as might be expected; C[O.sub.2] was higher at the low ventilation rate, as expected. Spot measurements in empty classrooms without children showed that the air velocity was slightly higher at the low temperature, although it was still low (< 0.11 m/s [22 fpm]); noise levels did not differ much between conditions (Table 3). There are too few data to interpret the measured concentrations of airborne particles, but the few measurements performed suggest that they were lower at the high ventilation rate (Table 3). The outdoor ozone concentration was about 24[+ or -]6 ppb during the course of experiments.

The results of the schoolwork performance exercises are presented in Table 4. They are based on complete design analyses, i.e., analyzing for each exercise only the results obtained from those pupils who had taken that test in all four conditions. A detailed summary of the analyses performed for each test is given in the following text and in Table 5. Due to teacher error, the subtraction, multiplication, number checking, and addition tasks were not administered in week three in one class, so using repeated measures in a complete 2 x 2 design was not possible for these exercises. The analysis for the differences between three other conditions, excluding the one from week three, was possible, but since the design of the study does not make it possible to test the differences between single conditions (because they are confounded with any differences between weeks), the analyses of the main effects for these exercises were consequently based instead on an incomplete design, i.e., including the performance of pupils who had not performed the exercise in all four conditions (a complete class in one condition). As a measure of the average performance, adjusted (least squares) means were used.

Subtraction. Analysis of the data for pupils who performed the test in three conditions (complete design) showed that there was a significant difference between conditions (P < 0.001) regarding speed (number of units attempted) but not errors. The analyses based on an incomplete design showed that reducing temperature and increasing outdoor air supply rate significantly improved the speed with which pupils subtracted the numbers. There was also a tendency (P = 0.06) for the increased ventilation rate to reduce the percentage of errors. There was no effect of temperature on percentage errors. Speed significantly increased in the course of the experiment, but no adjustment for this learning effect could be made because of the missing data.

Multiplication. Analysis of the data for pupils who performed the test in three conditions (complete design) showed that there was a significant difference between conditions (P < 0.001) regarding speed (number of units attempted) but not errors. The analyses based on the incomplete design showed that increasing outdoor air supply rate significantly increased speed and reduced percentage errors. There were no significant effects of reduced temperature. Speed and percentage errors decreased significantly in the course of the experiment, but no adjustment for these effects could be made because of the missing data.

Number comparison. Analysis of the data for pupils who performed the test in three conditions (complete design) showed that there was a significant difference between conditions (P < 0.01) regarding speed (the number of units attempted to compare) but not errors. The analyses based on incomplete design showed that increasing the outdoor air supply rate significantly improved the speed with which pupils compared the numbers. There was a tendency for the reduced temperature to increase speed, but this effect did not reach formal statistical significance (P = 0.10). There were no significant effects on errors. Speed and percentage errors significantly increased in the course of the experiment, but no adjustment for these effects could be made because of the missing data.

Addition. Analysis of the data for pupils who performed the task in three conditions (complete design) showed that there was a significant difference between conditions (P < 0.005) regarding speed (the number of units attempted) but not errors. The analyses based on the incomplete design showed that increasing outdoor air supply rate and reducing air temperature significantly improved the speed at which pupils added numbers. There were no significant effects on errors and no significant change of speed or percentage errors in the course of the experiment.

Logical thinking. No significant effects were observed.

Acoustic proofreading. Reducing temperature significantly reduced the error rate. There were no significant effects on false positives. In the course of the experiment, errors increased while the percentage of false positives decreased significantly. Adjusting for these trends did not change the results. There were no effects of increased ventilation rate. Analyses of all available data (incomplete design) showed similar results.

Reading and comprehension. Reducing temperature significantly increased reading speed. There were no effects of increased ventilation on speed or on errors (comprehension). In the course of the experiment, reading speed increased significantly while errors decreased significantly. Adjusting for these trends did not change the results. Analyses of all available data (incomplete design) showed similar results.

Proofreading. The exercise was not completed by either class in week two and, consequently, due to missing data, the effects on either speed or percentage errors/percentage false positives could not be estimated.

The effects on the performance of individual tasks caused by reducing classroom temperature (independent of ventilation) are summarized in Figure 3, while those caused by increasing the outdoor air supply rate (independently of classroom temperature) are summarized in Figure 4. Reduced temperature significantly improved the speed at which the addition, subtraction, and reading and comprehension tasks were performed and significantly reduced the percentage of errors in the acoustic proofreading task. Increased outdoor air supply rate significantly improved the speed at which all four numerical tasks were performed and at the same time reduced the percentage of errors committed in the subtraction and multiplication tasks. Taking into account the number of statistical tests performed, only two of these effects could have occurred by chance at the selected significance level of P = 0.05. It should be noted that the effects on the performance of numerical tasks and the proofreading task were derived from the incomplete data set (one or two classes missing one experimental condition). Thus, although the effects on speed were adjusted, no adjustment was possible for the effects on errors, so the results for errors should be interpreted with caution.

[FIGURE 3 OMITTED]

Observations of the children's behavior, appearance, and working ability. When the temperature was low, independently of ventilation rate, more children were observed to play with things (P = 0.03; total observations = 24). When the outdoor air supply rate was increased, independently of the temperature, more children were observed to talk to neighbors (P = 0.03; total observations = 18). The observations of other items on the checklist were either not significantly different, were too few for it to be possible to make any sound comparisons, or were not made at all. It was not possible to analyze the parental logbooks, as the response rate was too low (35%).

Subjective assessments. Table 6 shows the results obtained on the VA scales. They are based on complete design analyses, i.e., each scale is analyzed using only the results of those pupils who had marked that scale in all four conditions. When the temperature was reduced, the pupils indicated that it was significantly less warm and more drafty, that there was less light in the classroom, and that the air was significantly more fresh. There was a tendency for the pupils to report that they were more hungry at the lower temperature, but the effect remained not significant (P = 0.059, N = 48) even when data from pupils who had not marked this scale in all four conditions were included. Pupils also indicated that the air was less dry when the temperature was reduced; this effect became formally significant (P = 0.027, N = 48) when data from pupils who had not marked this scale in all four conditions were included. When the ventilation rate was increased, pupils reported that it was significantly less quiet in the classroom. No significant effects on any other VA scale were found.

Estimated thermal state of the children. Table 7 shows the assessments of thermal conditions made by the children on VA scales at the end of the week and the estimated predicted mean vote (PMV) and predicted percentage dissatisfied (PPD) (ASHRAE 2004; Fanger 1970; ISO 2005). PMV and PPD were estimated using the measured air temperatures, relative humidity, and air velocity (Tables 1 and 2), and it was assumed that the operative temperature was equal to the air temperature, that the clothing insulation was about 0.6 clo and was similar under each experimental condition, and that the metabolic rate of children was about 1.7 met (CEN 1998; ECA 1992; Pejtersen et al. 1991). A higher metabolic rate was assumed than is usually used for adults during light office work, as it is expected that children in schools have a higher average activity level than adults in offices (running and playing every hour during the breaks between lessons) and because the production rate of C[O.sub.2] by children in the present experiments was similar to what is produced by adults (Table 1). For that reason, the metabolic rate for the children was estimated to be about 40% higher than for adults in offices considering that the body size of an average 10- to 12-year-old child is about 40% smaller than that of an average adult. Table 7 shows that at the higher temperature pupils indicated that it was only slightly too warm, while at the lower temperature they voted close to the midpoint of the scale, indicating that it was neither too warm nor too cold. These assessments are supported by the estimated PMV values.

Sensory panel assessments. No sensory assessments were made.

[FIGURE 4 OMITTED]

Experiment 2T

In Experiment 2T, the temperature was changed in a crossover design while the ventilation rate remained always at its normal low level. Table 2 shows the results of continuous measurements of temperature, relative humidity, C[O.sub.2] level, and window opening behavior in the classrooms throughout the course of the experiment. The values shown describe the periods when children were present in the classrooms (excluding even short breaks between classes). Table 2 shows that the air temperature in the classrooms was 21.6[degrees]C (70.9[degrees]F) in the low temperature condition and 24.9[degrees]C (76.8[degrees]F) in the high temperature condition, i.e., a 3.3 K difference (5.9[degrees]F); a similar but slightly lower difference of 2.3 K (4.1[degrees]F) was observed between the average maximum temperatures measured. The effective ventilation rate estimated from measurements of C[O.sub.2] by using a mass balance equation was about 3 L/s (6.4 cfm) per person; it was slightly higher in the warmer reference condition since the windows were opened more often, as might be expected. The relative humidity was about 52%. The C[O.sub.2] level was slightly higher in the warmer reference condition. Spot measurements in empty classrooms without children showed that the air velocity was slightly higher in the low temperature condition, although it was still low (< 0.15 m/s [29 fpm]); noise levels did not differ between conditions (Table 3). Measurements of settled dust indicate that more dust settled on glass plates when the temperature was low, although the measured concentration of airborne particles was lower in this condition (Table 3). The outdoor ozone concentration was about 18[+ or -]5 ppb during the experiments.

The results of the performance exercises are presented in Table 4. They are based on complete design analyses, i.e., each exercise was analyzed using only the results from those pupils who had performed that exercise in both conditions. A detailed summary of the results obtained for each exercise is given in the following text and in Table 5.

Subtraction. Reduced temperature tended to increase speed (number of units attempted) and to reduce errors. Speed significantly increased throughout the course of the experiment, although errors did not. After adjustment for this learning effect, reduced temperature significantly increased the number of units attempted (P = 0.001). These analyses were supplemented with analyses of all available data, i.e., including the performance of pupils who had not performed the exercise in both conditions (incomplete design); similar results were obtained.

Addition. Reduced temperature significantly increased speed (number of units attempted) with no significant effect on percentage errors. Speed did not increase significantly in the course of the experiment, but percentage errors tended to increase (P = 0.06). After adjustment for this trend, the effect of temperature on errors remained nonsignificant. The analyses of all available data (incomplete design) showed similar results.

Logical thinking. There was no significant effect of reduced temperature on speed or percentage errors, although both increased significantly throughout the course of the experiment. After adjustment for these trends, the effect of temperature on both speed and errors remained nonsignificant. The analyses of all available data (incomplete design) showed similar results.

Acoustic proofreading. There was no significant effect of reduced temperature on errors, but the percentage of false positives significantly increased at reduced temperature. Neither percentage errors nor the percentage of false positives changed significantly throughout the course of the experiment.

Reading and comprehension. There was no significant effect of reduced temperature on reading speed or errors (comprehension). Speed increased significantly throughout the course of the experiment, while percentage errors decreased significantly. After adjustment for these trends, the effect of temperature on both speed and errors remained nonsignificant. The analyses of all available data (incomplete design) showed similar results.

The effects on the performance of the different tasks that were caused by reducing classroom temperature are summarized in Figure 3. Reduced temperature significantly improved the speed at which the addition task was performed, and the positive effect on speed of subtraction approached significance. Reduced temperature significantly increased the percentage of false positives in the acoustic proofreading task. Taking into account the number of statistical tests performed, one such effect could have occurred by chance at the selected significance level of P = 0.05.

Observations of the children's behavior, appearance, and working ability. When the temperature was reduced, more children were observed to work hard (P = 0.005; total observations = 52). For other items on the checklist, there were either no significant differences or they were indicated too seldom for it to be possible to make any sound comparisons, or they were not indicated at all. Parents were not asked to maintain logbooks in this experiment.

Subjective assessments. Table 6 shows the results obtained on the VA scales. They are based on complete design analyses, i.e., each scale was analyzed using only data from those pupils who had marked that scale in both conditions; it was not possible to perform any analyses that included the ratings of pupils who had not marked some scales in all conditions. When the temperature was reduced, the pupils indicated they felt significantly less warm and also considered that there was significantly less light and noise in the classroom. They reported significantly less intense headaches at reduced temperature. Additionally, there was a tendency for pupils to feel less hungry when the temperature was reduced (P = 0.07). As this nonsignificant tendency is in the opposite direction to the nonsignificant tendency that was found in Experiment 1T, it seems justifiable to disregard both of them. No other significant effects were observed.

Estimated thermal state of the children. Table 7 shows the assessments of thermal conditions made by children on the VA scales at the end of the week and the PMV and PPD as estimated for Experiment 1T. Table 7 shows that at the higher temperature pupils indicated that it was only slightly too warm, while at the lower temperature they voted close to the midpoint of the scale, indicating that it was neither too warm or too cold. These assessments are supported by the estimated PMV values.

Sensory panel assessments. Table 8 shows the results of assessments of perceived air quality made by the panel of adult subjects. Reducing air temperature significantly increased the acceptability of classroom air quality; the air was also perceived to be significantly fresher. No other significant effects were observed.

DISCUSSION

The present results show that reducing the air temperature in classrooms with moderately elevated temperatures improved the performance of schoolwork by children. Performance improved both in the case of numerical tasks requiring concentration and logical thinking and in the case of language-based tasks. The speed at which the tasks were performed improved, and the effects in four cases reached statistical significance (P < 0.05), while in two cases they approached significance (P < 0.10). The percentage of errors was significantly affected in only two cases and only in the acoustic proofreading task, in which the speed of performance was imposed by the rate of dictation. The percentage of errors was reduced in Experiment 1T and the percentage of false positives increased in Experiment 2T, suggesting the exertion of increased effort to maintain speed. At the same time, in the latter experiment, the decrease in the percentage of errors committed in subtraction approached significance and pupils reported a significantly reduced intensity of headaches, suggesting that they found it easier to perform the tasks.

The effects on speed in Experiment 2T were similar to the effects observed in Experiment 1T, although only in the case of addition and subtraction were the effects significant in both experiments. This can be considered a fairly good agreement between the two independent experiments carried out one year apart. Further validation is required to examine whether the different component skills that affect the overall performance of schoolwork are similarly affected by avoiding elevated temperatures. The present results both confirm and supplement the findings of thermal effects on children's schoolwork performance that were obtained nearly four decades ago in the experiments by Schoer and Shaffran (1973) and by Wyon (1970), who, as in the present experiments, studied thermal effects on school performance in the moderate temperature range. It would be interesting in future studies to examine the effects on performance outside this temperature range and to study whether the magnitude of the effects on performance approaches what was observed in the present study (Figure 3) and by Wyon (1970) or is as low as was reported by Schoer and Shaffran (1973).

Increasing the outdoor air supply rate improved the performance of schoolwork by children. As in the case of the effects of cooling, the effects were mainly on speed and only to a very small extent on errors. The speed at which all four numerical tasks were performed was significantly improved, although no effects could be shown on the language-based tasks. In the previous two experiments in this series (Wargocki and Wyon 2007), similar effects were observed, although in the previous experiments the speed at which the language-based tasks were performed was also improved. Only in the case of subtraction and multiplication did the improvement in speed reach statistical significance in both previous and present experiments (Figure 4). No effects on errors were observed in the previous experiments (Wargocki and Wyon 2007), while in the present study the percentage of errors in multiplication and subtraction was lower when the ventilation rate had been increased. However, these effects were not adjusted for the incomplete data set from which they were derived, due to analytical limitations, and may be considered less credible. Taking the above into account, the present findings provide satisfactory validation of the previous findings, considering that the three experiments were performed at intervals of several months over the course of a full year, one in winter and two in late summer, with six different groups of children as subjects. Whether the small discrepancies between the results of the three series of experiments are caused by the differences in the effects of outdoor air supply rate on the different component skills that affect the overall performance of schoolwork or by other factors should be investigated in future experiments.

The present results were obtained using a crossover design in pairs of classrooms. The main advantage of this is that any external factors that affect performance in a given week affect the results obtained under both of the conditions that were established in that week, avoiding bias. However, the present design has some limitations; e.g., it does not prevent carryover effects, if present, and is sensitive to differences in group size. To minimize the former, the conditions were changed during each weekend, allowing two days for "recovery." It was not possible to assign an equal number of pupils to each class; however, the small differences in the number of pupils in the classrooms (Table 4) are not expected to have had much effect on the present results. It should be emphasized that testing the effects of interventions on performance separately for each class is not possible in the present design, as the effects of the environmental conditions are confounded with any difference between weeks, such as a systematic change in the performance of some tasks in the course of the experiment (due to learning, familiarity, or boredom) with uncontrolled external factors such as the weather and with differences between the test versions used on different occasions. The present design minimized the impact of these factors when the analysis was performed on pooled data from both classrooms. This is illustrated in Table 5 for the effect of learning on performance. In future experiments, multiple crossover designs (with repetitions) should be used, together with longer periods between the application of the interventions, to further reduce the possibility of bias. It would also be desirable to extend the period of each intervention beyond the period of one week used in the present experiments.

The observed effects of increased ventilation rate and reduced temperature on the performance of schoolwork by children (Figures 3 and 4) are larger than reported effects on the performance of office work by adults (Wyon and Wargocki 2006a, 2006b). This indicates that children may be more susceptible than adults to environmental conditions. The importance of the observed environmental effects on the performance of simple tasks resembling schoolwork for the entire learning process is as yet unknown, although a simple interpretation of the observed effects on speed would be that it would leave more time for learning and leisure, both of which would be expected to improve the long-term learning process. Neither the present results nor the ventilation rate experiments already reported in this series (Wargocki and Wyon 2007) provide any information on the mechanisms by which reduced temperatures or improved ventilation affect performance. It is possible that in the present series of experiments, the effects on performance were caused by the distraction of thermal discomfort (pupils reported that it was significantly warmer when cooling was off, as confirmed by the estimated PMV) and of the perception of poor air quality (pupils and the sensory panel indicated significantly lower classroom air quality when the temperature was high, as would be predicted by the models derived by Fang et al. [2000]). The existence of such mechanisms is implied by the results of the previous studies on the effects of temperature on performance that were recently reviewed by Wyon and Wargocki (2006a). Improved air quality is a possible reason performance improved when the ventilation rate was increased in previously reported experiments in the same series (Wargocki and Wyon 2007) and might be the mechanism for the effects observed in the present study.

Further experiments on possible mechanisms are required, as they are still poorly understood. These should include studies investigating the effects of interaction between temperature and ventilation rate on the performance of schoolwork. This issue could not be examined in the present series, as the interaction was not fully balanced and was confounded with any difference between weeks. Future experiments should also investigate the effects of slowly drifting temperatures on the performance of schoolwork. In the present series, classroom temperatures slowly increased when cooling was off, and this effect could contribute to the observed differences between conditions created in the classrooms, considering that slow temperature changes have been reported to cause discomfort in adults (Wyon and Wargocki 2006a).

The fans in the split units were continuously in operation, both when cooling was on and off, to create a placebo condition. The noise level and air movement generated by the split units should therefore have been similar under both conditions. Nevertheless, pupils indicated that the air was significantly less still and the classroom less noisy when cooling was applied. This may have been because pupils found the air movement generated by the split units to be more drafty at the lower temperature or because the windows were more often closed when there was cooling so that the noise level from outdoors was lower. Pupils also indicated that the lighting level was significantly less bright in the classroom when the cooling was on. It is possible that pupils associated reduced temperature with less sunny outdoor conditions. Alternatively, reporting reduced light and noise may simply be an indication that pupils felt less stressed when the temperature was reduced.

In the present experiments, the temperature in the classrooms was reduced by split cooling units installed for that purpose. Reduced temperatures can also be achieved by other means, e.g., by using sun-blinds or by operating air-conditioning units installed in the air-handling unit. The former may restrict the amount of daylight, and this in itself could affect performance, as suggested by studies performed by the Heschong Mahone Group (2003). Central air conditioning may cause adverse health effects, as suggested by studies with adults in office buildings, reviewed by Seppanen and Fisk (2002). However, as pointed out by Wargocki et al. (2002), most of these studies were not carried out in the summer and consequently could disregard the thermal benefits of air conditioning. For that reason there is a need for further studies on how different methods of avoiding high classroom temperatures affect the performance of school-work by children.

Most temperate countries provide no cooling in schools, so the present findings indicate that the moderately raised classroom temperatures that occur during warm weather make the process of educating children less efficient, as additional effort must be used to overcome the lethargy induced by raised classroom temperatures and the distraction of thermal discomfort rather than to perform more, and more difficult, schoolwork. This is especially the case for those children who find schoolwork to be difficult and are thus unable to exert additional effort. Unsuitably high classroom temperatures may occur for other reasons than an absence of cooling and at other times of year than summertime, and the present results may also apply to these cases, as they serve to confirm the results that were obtained by Wyon (1970) in experiments during the winter in which classroom temperatures were artificially raised. The present results can be generalized to other developed countries where the climate, classroom conditions, level of education, and educational approach are often quite similar to those in Denmark. It seems likely that the observed positive impact on the performance of schoolwork that can be achieved by preventing children from feeling too warm would also occur in warmer climates. However, this assumption will have to be validated by repeating the study in hotter and more humid climates.

CONCLUSIONS

* Reducing moderately high classroom air temperatures in late summer from the region of 25[degrees]C to 20[degrees]C (77[degrees]F to 68[degrees]F), by providing sufficient cooling, improved the performance of two numerical tasks and two language-based tasks resembling schoolwork. Improvement mainly occurred in terms of the speed with which these tasks were performed, with almost no effects on errors. A fairly good agreement in terms of the effects on performance was obtained between two independent experiments carried out one year apart. In both experiments, children's thermal sensation decreased from slightly too warm to neutral, and in one experiment children reported significantly less headache at the lower temperature. A panel of adults visiting the classrooms soon after the children had left found the classroom air quality significantly fresher and more acceptable at the lower temperature. These findings imply that these benefits may have been the mechanisms for the observed effects of temperature on performance.

* Increasing the effective outdoor air supply rate from 5 L/s (11 cfm) per person to 10 L/s (20 cfm) per person improved the performance of four numerical tasks resembling schoolwork by improving the speed at which they were performed, with almost no effects on errors. The effects on performance were similar to the results obtained in other experiments previously reported in this series, providing independent validation.

* The present results require further validation with children in other age groups and in other countries, including those located in warm climates. The importance of the observed effects on simple tasks resembling schoolwork for the entire learning process remains to be demonstrated.

ACKNOWLEDGMENTS

The cost of these experiments was borne by ASHRAE Contract RP-1257, Indoor Environmental Effects on the Performance of School Work by Children, and by a grant from the Danish Technical and Scientific Research Council (STVF), which currently supports the International Centre for Indoor Environment and Energy at the Technical University of Denmark (DTU).

Some of the results obtained in this paper have been presented in short conference communications by Wargocki et al. (2005a, 2005b), with the kind permission of ASHRAE. A final report on RP-1257 will eventually be available from ASHRAE and will describe these and other experiments in the same series.

The authors wish to thank the technical section of the local authority responsible for the school in which the experiments were carried out for their cooperation in setting up the experiments; the teachers, parents, and pupils of the school for their participation in them; Sophie Irgens and Bartlomiej Matysiak, who acted as research assistants in Experiment 1T, and Mateusz Komenda, who acted as research assistant in Experiment 2T, reporting them either as their Midterm Project Report or MSc thesis; and Cristina Pirvu and Henry Cahyadi Willem, for their contributions to the execution of Experiment 2T. Many thanks are due to Bent Michael Nielsen for his help during selection of the school for the study. The authors also thank the anonymous reviewers for their detailed comments that made possible an improvement of the original manuscript.

REFERENCES

ASHRAE. 2004. ANSI/ASHRAE Standard 55-2004, Thermal Environmental Conditions for Human Occupancy. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Baddeley, A.D. 1968. A 3-min. reasoning test based on grammatical reasoning. Psychonomic Science 10:341-42.

CEN. 1998. CEN CR 1752, Ventilation for buildings: Design criteria for the indoor environment. Brussels: European Committee for Standardization.

DHBA. 1995. Building Regulations. Copenhagen, Denmark: Danish Ministry of Housing, Danish Housing and Building Agency.

ECA. 1992. Guidelines for ventilation requirements in buildings, Report no. 11, EUR 14449 EN, for the European Concerted Action "Indoor Air Quality and Its Impact on Man." Luxembourg: Office for Publications of the European Communities.

Eriksson, B.E., S. Mandorff, and A. Boysen. 1967. Larare bedomer klassrumsklimatet--en enkat under-sokning (Teachers' opinions of classroom climate--A questionnaire survey), in Swedish. BFR Report 31:1967. Stockholm: Building Research Council.

Fang, L., G. Clausen, and P.O. Fanger. 2000. Temperature and humidity: Important factors for perception of air quality and for ventilation requirements. ASHRAE Transactions 106(2):503-10.

Fanger, P.O. 1970. Thermal Comfort. Analysis and Applications in Environmental Engineering. Copenhagen: Danish Technical Press.

Heschong Mahone Group. 2003. Windows and Classrooms: A Study of Student Performance and the Indoor Environment, Technical Report. California Energy Commission.

Holmberg, I., and D.P. Wyon. 1969. The dependence of performance in school on classroom temperature. Educational and Psychological Interactions 31. Malmo, Sweden: School of Education.

Holmberg, I., and D.P. Wyon. 1972. Systematic observation of classroom behaviour during moderate heat stress. Educational and Psychological Interactions 37. Malmo, Sweden: School of Education.

ISO. 2005. ISO 7730, Ergonomics of the thermal environment--Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria. Brussels: European Committee for Standardization.

McIntyre, D.A. 1980. Indoor Climate. London: Applied Science.

Mendell, M.J., and G.A. Heath. 2005. Do indoor pollutants and thermal conditions in schools influence student performance? A critical review of the literature. Indoor Air 15:27-52.

Pejtersen, J., G. Clausen, J. Sorensen, D. Quistgaard, G. Iwashita, Y. Zhang, T. Onishi, and P.O. Fanger. 1991. Air pollution sources in kindergartens. Proceedings of Healthy Buildings--IAQ '91, pp. 221-24.

Ryd, H., and D.P. Wyon. 1970. Methods of evaluating human stress due to climate. Document D6/70, National Swedish Institute for Building Research. Stockholm: Building Research Council.

Schoer, L., and J. Shaffran. 1973. A combined evaluation of three separate research projects on the effects of thermal environment on learning and performance. ASHRAE Transactions 79:97-108.

Seppanen, O., and W.J. Fisk. 2002. Association of ventilation system type with SBS symptoms in office workers. Indoor Air 12:98-112.

Taylor, R. 1997. An Introduction to Error Analysis. The Study of Uncertainties in Physical Measurements. Sausalito, CA: University Science Books.

Wargocki, P. 2004. Sensory pollution sources in buildings. Indoor Air 14(Suppl 7):82-91.

Wargocki, P., J. Sundell, W. Bischof, G. Brundrett, P.O. Fanger, F. Gyntelberg, S.O. Hanssen, P. Harrison, A. Pickering, O. Seppanen, and P. Wouters. 2002. Ventilation and health in nonindustrial indoor environments. Report from a European Multidisciplinary Scientific Consensus Meeting. Indoor Air 12:113-28.

Wargocki, P., D.P. Wyon, L. Jark, and M. Schaub Hansen. 2005a. The effects of outdoor air supply rates in classrooms on the performance of schoolwork by children. Proceedings of CLIMA 2005, on CD-ROM.

Wargocki, P., D.P. Wyon, B. Matysiak, and S. Irgens. 2005b. The effects of classroom air temperature and outdoor air supply rate on performance of school work by children. Proceedings of Indoor Air 2005 I(1):368-72.

Wargocki, P., and D.P. Wyon. 2007. The effects of outdoor air supply rate and supply air filter condition in classrooms on the performance of schoolwork by children (1257-RP). HVAC & R Research 13(2):165-91.

Wyon, D.P. 1969. The effects of moderate heat stress on the mental performance of children. Document D8/69, National Swedish Institute for Building Research. Stockholm: Building Research Council.

Wyon, D.P. 1970. Studies of children under imposed noise and heat stress. Ergonomics 13(5):598-612.

Wyon, D.P., and I. Holmberg. 1972. Systematic observation of classroom behaviour during moderate heat stress. Proceedings of the CIB (W45) Symposium "Thermal Comfort and Moderate Heat Stress," September 13-15, Garston, Watford, England.

Wyon, D.P., and P. Wargocki. 2006a. Room temperature effects on office work. In D. Croome (ed.), Creating the Productive Environment, 2d ed., pp. 181-92. London: Taylor & Francis.

Wyon, D.P., and P. Wargocki. 2006b. Indoor air quality effects on office work", In D. Croome (ed.), Creating the Productive Environment, 2d ed., pp. 193-205. London: Taylor & Francis.

Yaglou, C.P., E.C. Riley, and D.I. Coggins. 1936. Ventilation requirements. ASHVE Transactions 42:133-62.

Pawel Wargocki, PhD

David P. Wyon, PhD

Member ASHRAE

Received February 14, 2006; accepted June 7, 2006

Pawel Wargocki and David P. Wyon are with the International Centre for Indoor Environment and Energy, Department of Mechanical Engineering, Technical University of Denmark, Lyngby, Denmark. Wargocki is also vice-president for research of the International Society for Indoor Air Quality.
Table 1. The Partially Balanced Design of Experiments 1T and 2T

 Experiment 1T, Summer
Week Classroom 1 Classroom 2

1 Ventilation rate high Ventilation rate low
 Air temperature low Air temperature high
2 Ventilation rate high Ventilation rate low
 Air temperature high Air temperature low

 Week break due to excursion
3 Ventilation rate low Ventilation rate high
 Air temperature high Air temperature low
4 Ventilation rate low Ventilation rate high
 Air temperature low Air temperature high

 Experiment 2T, Summer
Week Classroom 1 Classroom 2

1 Ventilation rate low
 Air temperature high Air temperature low
2 Ventilation rate low
 Air temperature low Air temperature high
3
4

Table 2. Average Values of Continuous Measurements Made When Classrooms
Were Occupied by Children and Estimated Ventilation Rates

 Experiment 1T, Summer
 Ventilation Rate
 Low
 Air Temperature
Parameter High Low

Classroom temperature, 24.6[+ or -]1.3 19.2[+ or -]0.8
 mean[+ or -]sd, [degrees]C (a)
Classroom temperature 20.6-26.7 17.7-21.5
 range, min-max, [degrees]C
Classroom RH, 52[+ or -]8 54[+ or -]4
 mean[+ or -]sd, % (a)
Classroom C[O.sub.2], 952[+ or -]232 1049[+ or -]154
 mean[+ or -]sd, ppm (a)
Average classroom 1138 1218
 peak C[O.sub.2], ppm
Total time with 1 or more 13.4 5.4
 windows opened, h
Estimated effective 468[+ or -]157 402[+ or -]103
 ventilation rate, [m.sup.3]/
 h (b)
Average number of 22+1 22+1
 pupils + teacher, per class
Estimated effective 5.7[+ or -]2.1 4.7[+ or -]1.3
 ventilation rate,
 L/s per person (b)

 Experiment 1T, Summer
 Ventilation Rate
 High
 Air Temperature
Parameter High Low

Classroom temperature, 22.5[+ or -]0.9 20.8[+ or -]0.9
 mean[+ or -]sd, [degrees]C (a)
Classroom temperature 20.6-25.7 18.5-23.0
 range, min-max, [degrees]C
Classroom RH, 49[+ or -]8 56[+ or -]8
 mean[+ or -]sd, % (a)
Classroom C[O.sub.2], 744[+ or -]176 809[+ or -]148
 mean[+ or -]sd, ppm (a)
Average classroom 831 843
 peak C[O.sub.2], ppm
Total time with 1 or more 11.8 6.7
 windows opened, h
Estimated effective 881[+ or -]176 800[+ or -]198
 ventilation rate, [m.sup.3]/
 h (b)
Average number of 24+1 23+1
 pupils + teacher, per class
Estimated effective 9.9[+ or -]1.6 9.3[+ or -]2.2
 ventilation rate,
 L/s per person (b)

 Experiment 2T, Summer
 Ventilation Rate
 Low
 Air Temperature
Parameter High Low

Classroom temperature, 24.9[+ or -]1.7 21.6[+ or -]1.6
 mean[+ or -]sd, [degrees]C (a)
Classroom temperature 20.8-28.1 19.0-25.8
 range, min-max, [degrees]C
Classroom RH, 52[+ or -]4 52[+ or -]6
 mean[+ or -]sd, % (a)
Classroom C[O.sub.2], 1230[+ or -]325 1462[+ or -]412
 mean[+ or -]sd, ppm (a)
Average classroom 1424 1721
 peak C[O.sub.2], ppm
Total time with 1 or more 13.7 4.8
 windows opened, h
Estimated effective 332[+ or -]122 247[+ or -]41
 ventilation rate, [m.sup.3]/
 h (b)
Average number of 24+1 24+1
 pupils + teacher, per class
Estimated effective 3.7[+ or -]1.1 2.7[+ or -]0.4
 ventilation rate,
 L/s per person (b)

a. Experiment 1T: outdoor temperature = 17.4[degrees]C[+ or -]
1.8[degrees]C, RH = 69%[+ or -]12%; C[O.sub.2] = 400[+ or -]17 ppm;
Experiment 2T: outdoor temperature = 15.7[degrees]C[+ or -]
1.8[degrees]C, RH = 68%[+ or -]8%; C[O.sub.2] = 378[+ or -]11 ppm.
b. Experiment 1T: average production rate of C[O.sub.2] = 17.7[+ or -]
1.3 L/h per person (10-year-old children + teacher);
Experiment 2T: average production rate of C[O.sub.2] = 17.5[+ or -]
1.5 L/h per person (11-year-old children + teacher).

Table 3. Average Conditions (Mean[+ or -]SD) in Empty Classrooms, as
Determined by Spot Measurements at the End of Each Intervention

 Experiment 1T,
 Summer
 Ventilation Rate
 Low
 Air Temperature
Parameter High Low

Air velocity [m/s] 0.04[+ or -]0.03 0.11[+ or -]0.07
Noise level [dB(A)] 36 35
Particles
> 0.02 [micro]m (ultrafines) 3435 8383
 [counts/[cm.sup.3]]
 > 0.75 [micro]m [counts/ 308 1673
 1000 [cm.sup.3]]
 > 1.0 [micro]m [counts/ N/A
 1000 [cm.sup.3]]
 > 2.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 3.5 [micro]m [counts/
 1000 [cm.sup.3]]
 > 5.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 7.5 [micro]m [counts/
 1000 [cm.sup.3]]
 > 10.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 15.0 [micro]m [counts/ 2 2
 1000 [cm.sup.3]]
Settled dust N/A
[% covering of a glass plate]

 Experiment 1T,
 Summer
 Ventilation Rate
 High
 Air Temperature
Parameter High Low

Air velocity [m/s] 0.09[+ or -]0.05 0.11[+ or -]0.06
Noise level [dB(A)] 39 38
Particles
> 0.02 [micro]m (ultrafines) 2700 2030
 [counts/[cm.sup.3]]
 > 0.75 [micro]m [counts/ 717 256
 1000 [cm.sup.3]]
 > 1.0 [micro]m [counts/ N/A
 1000 [cm.sup.3]]
 > 2.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 3.5 [micro]m [counts/
 1000 [cm.sup.3]]
 > 5.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 7.5 [micro]m [counts/
 1000 [cm.sup.3]]
 > 10.0 [micro]m [counts/
 1000 [cm.sup.3]]
 > 15.0 [micro]m [counts/ 2 1
 1000 [cm.sup.3]]
Settled dust N/A
[% covering of a glass plate]

 Experiment 2T,
 Summer
 Ventilation Rate
 Low
 Air Temperature
Parameter High Low

Air velocity [m/s] 0.05[+ or -]0.02 0.15[+ or -]0.09
Noise level [dB(A)] 44[+ or -]3 44[+ or -]4
Particles
> 0.02 [micro]m (ultrafines) N/A
 [counts/[cm.sup.3]]
 > 0.75 [micro]m [counts/ 2479 1192
 1000 [cm.sup.3]]
 > 1.0 [micro]m [counts/ 1480 639
 1000 [cm.sup.3]]
 > 2.0 [micro]m [counts/ 644 262
 1000 [cm.sup.3]]
 > 3.5 [micro]m [counts/ 247 106
 1000 [cm.sup.3]]
 > 5.0 [micro]m [counts/ 83 46
 1000 [cm.sup.3]]
 > 7.5 [micro]m [counts/ 27 21
 1000 [cm.sup.3]]
 > 10.0 [micro]m [counts/ 12 10
 1000 [cm.sup.3]]
 > 15.0 [micro]m [counts/ 3 3
 1000 [cm.sup.3]]
Settled dust 1.89[+ or -]0.24 2.73[+ or -]0.62
[% covering of a glass plate]

Table 4. Performance of Schoolwork by Children (Mean[+ or -]se);
Significant Differences Shown in Bold

 Experiment 1T, Summer
 Ventilation Rate
 Number of Low
Performance Performance Pupils Total Air Temperature
Exercise Metric (Class 1/2) High

Subtraction Attempted 44 (20/24) 2.53[+ or -]0.40 (a)
 units per min (1.94[+ or -]0.31) (b)
 Percentage 17.38[+ or -]5.37 (a)
 errors
Multiplication Attempted 45 (21/24) 1.66[+ or -]0.31 (a)
 units per min (1.49[+ or -]0.11) (b)
 Percentage 33.42[+ or -]5.81 (a)
 errors
Number Attempted 44 (19/25) 14.34[+ or -]0.89 (a)
Comparison units per min (11.81[+ or -]0.97) (b)
 Percentage 1.77[+ or -]0.43 (a)
 errors
Addition Attempted 42 (18/24) 4.52[+ or -]0.48 (a)
 units per min (3.70[+ or -]0.16) (b)
 Percentage 6.38[+ or -]1.41 (a)
 errors
Logical Attempted 32 (16/16) 7.43[+ or -]0.36
Reasoning units per min
 Percentage 24.62[+ or -]3.92
 errors
Acoustic Percentage 34 (16/18) 46.82[+ or -]3.88
Proofreading errors
 Percentage 3.02[+ or -]0.60
 false
 positives
Reading and Attempted 40 (18/22) 1.31[+ or -]0.07
Comprehension units per min
 Percentage 31.71[+ or -]2.73
 errors
Proofreading Attempted 40 (18/22) N/A (due to missing data)
 units per min
 Percentage 48.28[+ or -]3.29
 errors
 Percentage 9.15[+ or -]1.58
 false
 positives

 Experiment 1T, Summer
 Ventilation Rate
 Low High
Performance Performance Air Temperature
Exercise Metric Low High

Subtraction Attempted 2.71[+ or -]0.26 2.44[+ or -]
 units per min 0.18
 (2.69[+ or -]0.16) (b) (2.43[+ or -]
 0.17) (b)
 Percentage 17.46[+ or -]3.47 11.84[+ or -]
 errors 2.53
Multiplication Attempted 1.43[+ or -]0.14 1.78[+ or -]
 units per min 0.18
 (1.46[+ or -]0.96) (b) (1.81[+ or -]
 0.06) (b)
 Percentage 31.39[+ or -]4.22 23.59[+ or -]
 errors 3.62
Number Attempted 14.73[+ or -]0.66 15.30[+ or -]
Comparison units per min 0.90
 (14.61[+ or -]0.50) (b) (15.24[+ or -]
 0.52) (b)
 Percentage 3.87[+ or -]0.83 4.78[+ or -]
 errors 0.99
Addition Attempted 4.05[+ or -]0.33 4.39[+ or -]
 units per min 0.34
 (4.06[+ or -]0.11) (b) (4.40[+ or -]
 0.10) (b)
 Percentage 10.68[+ or -]3.01 6.88[+ or -]
 errors 1.35
Logical Attempted 7.07[+ or -]0.49 6.86[+ or -]
Reasoning units per min 0.45
 Percentage 21.87[+ or -]3.97 21.80[+ or -]
 errors 3.44
Acoustic Percentage 39.33[+ or -]3.94 42.18[+ or -]
Proofreading errors 3.18
 Percentage 2.09[+ or -]0.59 1.55[+ or -]
 false 0.53
 positives
Reading and Attempted 1.65[+ or -]0.11 1.32[+ or -]
Comprehension units per min 0.09
 Percentage 25.49[+ or -]2.70 27.29[+ or -]
 errors 3.31
Proofreading Attempted N/A (due to missing data)
 units per min
 Percentage 54.00[+ or -]3.19 (a) 52.64[+ or -]
 errors 4.63 (a)
 Percentage 8.64[+ or -]1.96 (a) 7.55[+ or -]
 false 2.65 (a)
 positives

 Experiment 1T, Summer
 Ventilation Rate Difference Between
 High Conditions
Performance Performance Air Temperature
Exercise Metric Low Test

Subtraction Attempted 3.06[+ or -]0.25 2 x 2
 units per min (3.06[+ or -] ANOVA
 0.16) (b)
 Percentage 13.65[+ or -]2.76 Wilcoxon
 errors
Multiplication Attempted 1.77[+ or -]0.15 2 x 2
 units per min (1.78[+ or -] ANOVA
 0.06) (b)
 Percentage 26.11[+ or -]3.95 Wilcoxon (a)
 errors
Number Attempted 14.08[+ or -]0.78 2 x 2
Comparison units per min 13.94[+ or -] ANOVA (c)
 0.51) (b)
 Percentage 3.66[+ or -]0.72 Wilcoxon
 errors
Addition Attempted 4.51[+ or -]0.32 2 x 2
 units per min (4.55[+ or -] ANOVA
 0.10) (b)
 Percentage 6.50[+ or -]1.37 Wilcoxon (a)
 errors
Logical Attempted 7.44[+ or -]0.45 Wilcoxon
Reasoning units per min
 Percentage 20.93[+ or -]3.61 Wilcoxon (e)
 errors
Acoustic Percentage 40.90[+ or -]3.05 2 x 2
Proofreading errors ANOVA (f)
 Percentage 2.14[+ or -]0.57 Wilcoxon (e)
 false
 positives
Reading and Attempted 1.62[+ or -]0.10 2 x 2
Comprehension units per min ANOVA (f)
 Percentage 27.71[+ or -]2.87 Wilcoxon (e)
 errors
Proofreading Attempted N/A (due to missing data)
 units per min
 Percentage 45.58[+ or -]2.68 Wilcoxon
 errors
 Percentage 7.03[+ or -]1.64 Wilcoxon
 false
 positives

 Experiment 1T, Summer Experiment
 Difference Between 2T, Summer
 Conditions Number of
Performance Performance Temperature Ventilation Pupils Total
Exercise Metric P < P < (Class 1/2)

Subtraction Attempted N/A (c) N/A (c) 42 (18/24)
 units per min (0.006)# (b) (0.014)# (b)
 Percentage N/A (c) N/A (c)
 errors (0.14) (d) (0.06) (d)
Multiplication Attempted N/A (c) N/A (c) N/A
 units per min (0.41) (b) (0.013)# (b)
 Percentage N/A (c) N/A (c)
 errors (0.56) (d) (0.05)# (d)
Number Attempted N/A (c) N/A (c) N/A
Comparison units per min (0.10) (b) (0.043)# (b)
 Percentage N/A (c) N/A (c)
 errors (0.24) (d) (0.75) (d)
Addition Attempted N/A (c) N/A (c) 42 (18/24)
 units per min (0.041#) (b) (0.001#) (b)
 Percentage N/A (c) N/A (c)
 errors (0.90) (d) (0.23) (d)
Logical Attempted 0.62 0.49 41 (18/23)
Reasoning units per min
 Percentage 0.27 0.47
 errors
Acoustic Percentage 0.043# 0.96 43 (19/24)
Proofreading errors
 Percentage 0.84 0.19
 false
 positives
Reading and Attempted 0.001# 0.78 43 (19/24)
Comprehension units per min
 Percentage 0.41 0.58
 errors
Proofreading Attempted N/A (due to missing data) N/A
 units per min
 Percentage N/A (g) N/A (g)
 errors
 Percentage N/A (g) N/A (g)
 false
 positives

 Experiment
 2T, Summer
 Ventilation Rate
 Low Wilcoxon
Performance Performance Air Temperature Test
Exercise Metric High Low P <

Subtraction Attempted 4.26[+ or -] 5.00[+ or -] 0.06
 units per min 0.40 0.32
 Percentage 10.43[+ or -] 6.47[+ or -] 0.08
 errors 1.96 0.95
Multiplication Attempted N/A
 units per min
 Percentage
 errors
Number Attempted N/A
Comparison units per min
 Percentage
 errors
Addition Attempted 4.37[+ or -] 4.68[+ or -] 0.002#
 units per min 0.28 0.27
 Percentage 5.89[+ or -] 4.29[+ or -] 0.86
 errors 2.23 0.75
Logical Attempted 7.89[+ or -] 8.25[+ or -] 0.60
Reasoning units per min 0.40 0.47
 Percentage 26.00[+ or -] 26.37[+ or -] 0.56
 errors 3.43 3.17
Acoustic Percentage 29.93[+ or -] 29.22[+ or -] 0.76
Proofreading errors 2.78 2.93
 Percentage 2.26[+ or -] 4.52[+ or -] 0.007#
 false 0.51 0.87
 positives
Reading and Attempted 1.53[+ or -] 1.66[+ or -] 0.59
Comprehension units per min 0.07 0.13
 Percentage 26.13[+ or -] 24.00[+ or -] 0.71
 errors 3.71 3.30
Proofreading Attempted N/A
 units per min
 Percentage
 errors
 Percentage
 false
 positives

a. Test results from one class only.
b. The results from GLM analysis, including missing data showing
adjusted means and P-values.
c. Due to missing data in one condition for one class, the test on a
complete balanced design was not possible.
d. Wilcoxon test for all data, including pupils who missed some
conditions.
e. Difference between conditions tested using Friedman ANOVA: percentage
errors in logical reasoning (P = 0.64); percentage false positives in
acoustic proofreading (P = 0.24); percentage errors in reading and
comprehension (P = 0.56).
f. After transforming data using the base-10 logarithm.
g. Analyses not available due to missing data under two conditions in
two classes.

Note: Significant Differences Shown in indicated with #.

Table 5. Summary of Intervention Effects on Performance (a)

 Adjusting for Experiment 1T, Summer
 Significant Data from Pupils Who Had
 Increase of Speed Performed the Exercise in
 in the Course of All Four Conditions
 the Experiment (Complete Design)
Performance Independently of Number
Exercise the Interventions of pupils Temperature Ventilation

Subtraction Not adjusted 44 N/A N/A
(Speed) Adjusted N/A N/A
Multiplication Not adjusted 45 N/A N/A
(Speed) Adjusted N/A N/A
Multiplication Not adjusted 45 N/A N/A
(Percentage Adjusted N/A N/A
 errors)
Number Not adjusted 44 N/A N/A
Comparison
(Speed) Adjusted N/A N/A
Addition Not adjusted 42 N/A N/A
(Speed) Adjusted N/A N/A
Reading and Not adjusted 40 [up arrow] NS
Comprehension
(Speed) Adjusted [up arrow] NS
Acoustic Not adjusted 34 [up arrow] NS
Proofreading
(Percentage Adjusted [up arrow] NS
 errors)
Acoustic Not adjusted 34 NS NS
Proofreading
(Percentage Adjusted NS NS
 false
 positives)

 Experiment 1T, Summer
 Adjusting for All Available Data, i.e.,
 Significant Including Pupils Who
 Increase of Speed Had Not Taken the Task
 in the Course of in All Four Conditions
 the Experiment (Incomplete Design)
Performance Independently of Number
Exercise the Interventions of Pupils Temperature Ventilation

Subtraction Not adjusted 48 [up arrow] [up arrow]
(Speed) Adjusted N/A N/A
Multiplication Not adjusted 48 N/S [up arrow]
(Speed) Adjusted N/A N/A
Multiplication Not adjusted 49 NS [up arrow]
(Percentage Adjusted N/A N/A
 errors)
Number Not adjusted 48 NS [up arrow]
Comparison
(Speed) Adjusted N/A N/A
Addition Not adjusted 48 [up arrow] [up arrow]
(Speed) Adjusted N/A N/A
Reading and Not adjusted 48 [up arrow] NS
Comprehension
(Speed) Adjusted [up arrow] NS
Acoustic Not adjusted 47 NS NS
Proofreading
(Percentage Adjusted [up arrow] NS
 errors)
Acoustic Not adjusted 49 NS NS
Proofreading
(Percentage Adjusted NS NS
 false
 positives)

 Adjusting for Experiment 2T, Summer
 Significant Data from Pupils Who
 Increase of Speed Had Performed
 in the Course of the Exercise in
 the Experiment Both Conditions
Performance Independently of (Complete Design)
Exercise the Interventions Number of Pupils Temperature

Subtraction Not adjusted 42 NS
(Speed) Adjusted [up arrow]
Multiplication Not adjusted N/A
(Speed) Adjusted
Multiplication Not adjusted N/A
(Percentage Adjusted
 errors)
Number Not adjusted N/A
Comparison
(Speed) Adjusted
Addition Not adjusted 42 [up arrow]
(Speed) Adjusted N/A
Reading and Not adjusted 43 NS
Comprehension
(Speed) Adjusted NS
Acoustic Not adjusted 43 NS
Proofreading
(Percentage Adjusted N/A
 errors)
Acoustic Not adjusted 43 [down arrow]
Proofreading
(Percentage Adjusted N/A
 false
 positives)

 Adjusting for Experiment 2T, Summer
 Significant All Available Data, i.e.,
 Increase of Speed Including Pupils Who Had
 in the Course of Not Performed the Exercise
 the Experiment in Both Conditions
Performance Independently of (Incomplete Design)
Exercise the Interventions Number of Pupils Temperature

Subtraction Not adjusted 47 NS
(Speed) Adjusted [up arrow]
Multiplication Not adjusted N/A
(Speed) Adjusted
Multiplication Not adjusted N/A
(Percentage Adjusted
 errors)
Number Not adjusted N/A
Comparison
(Speed) Adjusted
Addition Not adjusted 49 [up arrow]
(Speed) Adjusted N/A
Reading and Not adjusted 49 NS
Comprehension
(Speed) Adjusted NS
Acoustic Not adjusted N/A N/A
Proofreading
(Percentage Adjusted N/A
 errors)
Acoustic Not adjusted N/A N/A
Proofreading
(Percentage Adjusted N/A
 false
 positives)

a. The table shows the effects on speed, errors, or false positives for
exercises in which significant differences between conditions were
observed. An arrow pointing up ([up arrow]) indicates that performance
significantly improved (i.e., speed increased or percentage errors or
false positives decreased) when the outdoor air supply rate increased or
a temperature was reduced. An arrow pointing down ([down arrow])
indicates that performance significantly reduced (i.e., speed decreased
or percentage errors or false positives increased). NS = not
significant.

Table 6. Results Obtained on the Visual Analogue Scales Marked Each Week
by the Children (Mean[+ or -]se, Median); Significant Differences Shown
in Bold

 Experiment 1T, Summer
 Ventilation Rate
 Low
Perceptions/Symptoms Number Air Temperature
(Coding) of Pupils High Low

Too cold (0)-- 39 66.9[+ or -]3.1 46.1[+ or -]3.8
Too warm (100) 60.9 40.1
There is a draft (0)-- 39 73.7[+ or -]3.8 56.7[+ or -]4.2
The air is still (100) 85.0 48.0
The air is humid (0)-- 35 61.2[+ or -]4.1 54.3[+ or -]3.6
The air is dry (100) 52.0 52.3
The air is poor (0)-- 35 61.1[+ or -]5.7 64.6[+ or -]5.2
The air is fresh (100) 67.9 74.9
Too little light (0)-- 37 50.1[+ or -]2.1 40.8[+ or -]2.9
Too much light (100) 48.9 47.7
Too noisy (0)-- 38 69.7[+ or -]4.2 61.2[+ or -]4.1
Quiet (100) 75.5 55.5
Nose blocked (0)-- 39 69.5[+ or -]4.9 62.4[+ or -]5.2
Nose fine (100) 78.3 59.9
Throat dry (0)-- 38 56.8[+ or -]5.9 58.6[+ or -]6.2
Throat fine (100) 54.6 69.1
Lips dry (0)-- 38 48.1[+ or -]5.9 41.0[+ or -]5.8
Lips not dry (100) 42.4 29.5
Skin dry (0)-- 37 65.4[+ or -]4.7 62.7[+ or -]5.8
Skin not dry (100) 62.7 71.9
Eyes hurt (0)-- N/A
Eyes not hurt (100)
Very hungry (0)-- 39 41.5[+ or -]4.7 40.2[+ or -]4.7
Full (100) 40.4 35.2
Slept badly last night 39 74.7[+ or -]3.9 75.7[+ or -]4.3
 (0)--
Slept well last night 83.5 85.0
 (100)
Feeling very tired 39 54.3[+ or -]4.9 51.4[+ or -]5.4
 (0)--
Not feeling tired at 47.4 49.2
 all (100)
Slept too little last 39 59.5[+ or -]4.2 64.3[+ or -]4.9
 night (0)--
Slept too long last 54.1 68.2
 night (100)
Sleepy (0)-- N/A
Awake (100)
Does not feel like 39 49.7[+ or -]5.0 46.2[+ or -]5.6
 working today (0)--
Feels like working 48.3 45.6
 today (100)
Having a headache N/A
 (0)--
Not having a headache
 at all (100)

 Difference
 Experiment 1T, Summer between
 Ventilation Rate Conditions
 High (Friedman
Perceptions/Symptoms Ait Temperature ANOVA)
(Coding) High Low P-value

Too cold (0)-- 54.9[+ or -]3.0 50.3[+ or -]3.2 <0.001#
Too warm (100) 48.3 48.0
There is a draft (0)-- 70.1[+ or -]4.0 58.6[+ or -]3.5 <0.001#
The air is still (100) 74.9 50.2
The air is humid (0)-- 59.2[+ or -]3.9 54.9[+ or -]3.5 0.99
The air is dry (100) 48.9 49.9
The air is poor (0)-- 58.0[+ or -]5.9 72.3[+ or -]4.4 0.022#
The air is fresh (100) 50.5 76.5
Too little light (0)-- 49.2[+ or -]2.8 45.4[+ or -]2.1 0.16
Too much light (100) 48.3 48.0
Too noisy (0)-- 51.6[+ or -]4.2 55.3[+ or -]4.0 0.021#
Quiet (100) 49.1 50.3
Nose blocked (0)-- 63.6[+ or -]5.3 73.8[+ or -]4.9 0.29
Nose fine (100) 70.6 89.9
Throat dry (0)-- 53.5[+ or -]5.9 56.3[+ or -]6.0 0.98
Throat fine (100) 48.9 60.7
Lips dry (0)-- 40.9[+ or -]6.1 43.4[+ or -]5.6 0.55
Lips not dry (100) 34.4 41.4
Skin dry (0)-- 59.1[+ or -]5.7 63.8[+ or -]5.1 0.67
Skin not dry (100) 59.3 66.4
Eyes hurt (0)-- N/A
Eyes not hurt (100)
Very hungry (0)-- 43.3[+ or -]5.1 32.2[+ or -]4.0 0.21
Full (100) 38.5 35.8
Slept badly last night 74.4[+ or -]4.4 71.9[+ or -]4.3 0.83
 (0)--
Slept well last night 86.9 80.1
 (100)
Feeling very tired 50.2[+ or -]5.5 56.2[+ or -]5.2 0.50
 (0)--
Not feeling tired at 46.2 49.2
 all (100)
Slept too little last 59.6[+ or -]4.6 52.6[+ or -]5.0 0.38
 night (0)--
Slept too long last 53.8 49.9
 night (100)
Sleepy (0)-- N/A
Awake (100)
Does not feel like 46.2[+ or -]5.8 52.0[+ or -]5.6 0.31
 working today (0)--
Feels like working 36.1 50.0
 today (100)
Having a headache N/A
 (0)--
Not having a headache
 at all (100)

 Experiment 1T, Summer
 Main Effects
 (Wilcoxon Test)
Perceptions/Symptoms P-value Experiment 2T, Summer
(Coding) Temperature Ventilation Number of Pupils

Too cold (0)-- <0.001 (a)# 0.21 41
Too warm (100)
There is a draft (0)-- <0.001 (a)# 0.89 40
The air is still (100)
The air is humid (0)-- 0.26 (a) 0.64 40
The air is dry (100)
The air is poor (0)-- 0.028 (a)# 0.60 41
The air is fresh (100)
Too little light (0)-- 0.009 (a)# 0.66 41
Too much light (100)
Too noisy (0)-- 0.55 0.009 (a)# 40
Quiet (100)
Nose blocked (0)-- 0.53 0.29 41
Nose fine (100)
Throat dry (0)-- 0.64 0.62 41
Throat fine (100)
Lips dry (0)-- 0.61 0.52 40
Lips not dry (100)
Skin dry (0)-- 0.40 0.96 41
Skin not dry (100)
Eyes hurt (0)-- N/A 41
Eyes not hurt (100)
Very hungry (0)-- 0.063 (a) 0.34 41
Full (100)
Slept badly last night 0.50 0.50 40
 (0)--
Slept well last night
 (100)
Feeling very tired 0.21 0.93 41
 (0)--
Not feeling tired at
 all (100)
Slept too little last 0.83 0.23 40
 night (0)--
Slept too long last
 night (100)
Sleepy (0)-- N/A 40
Awake (100)
Does not feel like 0.97 0.99 41
 working today (0)--
Feels like working
 today (100)
Having a headache N/A 41
 (0)--
Not having a headache
 at all (100)

 Experiment 2T, Summer
 Ventilation Rate
 Low Wilcoxon
Perceptions/Symptoms Air Temperature Test
(Coding) High Low P-value

Too cold (0)-- 62.6[+ or -]2.9 50.3[+ or -]2.4 0.002#
Too warm (100) 63.2 49.1
There is a draft (0)-- 65.6[+ or -]3.6 60.5[+ or -]3.4 0.23
The air is still (100) 65.9 57.0
The air is humid (0)-- 59.2[+ or -]3.6 56.0[+ or -]2.7 0.44
The air is dry (100) 55.0 49.0
The air is poor (0)-- 55.3[+ or -]4.2 56.3[+ or -]3.5 0.91
The air is fresh (100) 49.6 60.2
Too little light (0)-- 51.8[+ or -]2.0 44.6[+ or -]2.5 0.017#
Too much light (100) 50.4 47.7
Too noisy (0)-- 52.0[+ or -]4.3 63.1[+ or -]4.9 0.008#
Quiet (100) 47.2 67.8
Nose blocked (0)-- 59.5[+ or -]5.2 67.7[+ or -]4.6 0.39
Nose fine (100) 64.7 79.7
Throat dry (0)-- 52.3[+ or -]4.8 53.9[+ or -]5.0 0.74
Throat fine (100) 50.7 51.6
Lips dry (0)-- 51.0[+ or -]4.7 51.0[+ or -]4.4 1.0
Lips not dry (100) 48.8 49.1
Skin dry (0)-- 60.2[+ or -]4.1 60.0[+ or -]4.5 0.96
Skin not dry (100) 53.7 50.7
Eyes hurt (0)-- 66.4[+ or -]4.7 70.2[+ or -]4.7 0.86
Eyes not hurt (100) 65.9 82.2
Very hungry (0)-- 23.2[+ or -]3.6 30.0[+ or -]3.2 0.07
Full (100) 19.6 29.4
Slept badly last night 78.5[+ or -]4.0 73.3[+ or -]4.4 0.36
 (0)--
Slept well last night 91.4 79.6
 (100)
Feeling very tired 59.0[+ or -]4.6 52.7[+ or -]4.5 0.26
 (0)--
Not feeling tired at 55.5 45.4
 all (100)
Slept too little last 58.6[+ or -]5.0 54.1[+ or -]4.4 0.38
 night (0)--
Slept too long last 55.8 50.7
 night (100)
Sleepy (0)-- 62.6[+ or -]4.7 64.3[+ or -]4.2 0.64
Awake (100) 59.6 67.2
Does not feel like 39.6[+ or -]4.8 44.4[+ or -]4.3 0.40
 working today (0)--
Feels like working 40.7 47.7
 today (100)
Having a headache 61.0[+ or -]4.8 75.2[+ or -]4.6 0.025#
 (0)--
Not having a headache 52.8 88.2
 at all (100)

a. In the analysis of all ratings including pupils who did not mark the
scales in all conditions (N = 48), the effects were as follows: Too
cold--Too warm (P < 0.001); There is a draft--The air is still (P <
0.001); The air is humid--The air is dry (P = 0.027); The air is poor--
The air is fresh (P = 0.053); Too little light--Too much light (P =
0.023); Too noisy--Quiet (P < 0.001); Very hungry--Full (P = 0.059).

Note: Significant Differences Shown in indicated with #.

Table 7. Thermal State of the Children

 Experiment 1T, Summer
 Ventilation Rate
 Low High
 Air Temperature
 High Low High

Visual Analogue Scale: Too cold (0)--Too warm (100)
Mean[+ or -]SE 66.9[+ or -]3.1 46.1[+ or -]3.8 54.9[+ or -]3.0
Median 60.9 40.1 48.3
(25th percentile, (49.8; 84.7) (30.0; 53.5) (45.3; 63.6)
75th percentile)

Estimated Thermal Sensation (a)
PMV 0.81 -0.25 0.38
PPD 19% 6% 8%

 Experiment 1T,
 Summer Experiment 2T, Summer
 Ventilation Rate Ventilation Rate
 High Low
 Air Temperature Air Temperature
 Low High Low

Visual Analogue Scale: Too cold (0)--Too warm (100)
Mean[+ or -]SE 50.3[+ or -]3.2 62.6[+ or -]2.9 50.3[+ or -]2.4
Median 48.0 63.2 49.1
(25th percentile, (39.1; 59.3) (47.5; 78.0) (45.9; 56.1)
75th percentile)

Estimated Thermal Sensation (a)
PMV 0.06 0.87 0.11
PPD 5% 21% 5%

a. Assumed: clo = 0.6, met = 1.7; operative temperature = air
temperature.

Table 8. Adult Sensory Panel Assessments of Classroom Air Quality
(Mean[+ or -]se, Median); Significant Differences Shown in Bold

 Experiment 2T, Summer
 Ventilation Rate
 Low
Perceptions/Symptoms Experiment Number of Air Temperature
(Coding) 1T, Summer Subjects High

Acceptability of air quality N/A 28 0.143[+ or -]
(Clearly not acceptable = 0.049
-1 - Clearly acceptable = 1) 0.158
Odor intensity (No odor = 0 - 28 16.7[+ or -]1.1
Overpowering odor = 50) 14.8
Air freshness (Air stuffy = 28 43.6[+ or -]2.7
0 - Air fresh = 100) 40.5
Perceived dryness of air 28 53.8[+ or -]2.5
(Air humid = 0 - Air dry = 51.5
100)

 Experiment 2T, Summer
 Ventilation Rate
 Low
Perceptions/Symptoms Air Temperature Wilcoxon Test
(Coding) Low P-value

Acceptability of air quality 0.385[+ or -]0.061 < 0.001#
(Clearly not acceptable = 0.523
-1 - Clearly acceptable = 1)
Odor intensity (No odor = 0 - 18.6[+ or -]2.2 0.92
Overpowering odor = 50) 14.3
Air freshness (Air stuffy = 68.2[+ or -]3.6 < 0.001#
0 - Air fresh = 100) 77.3
Perceived dryness of air 48.9[+ or -]2.3 0.25
(Air humid = 0 - Air dry = 49.3
100)

Note: Significant Differences Shown in indicated with #.
COPYRIGHT 2007 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

 Reader Opinion

Title:

Comment:



 

Article Details
Printer friendly Cite/link Email Feedback
Author:Wargocki, Pawel; Wyon, David P.
Publication:HVAC & R Research
Geographic Code:4EUUK
Date:Mar 1, 2007
Words:16619
Previous Article:The effects of outdoor air supply rate and supply air filter condition in classrooms on the performance of schoolwork by children (RP-1257).
Next Article:Calibrating detailed building energy simulation programs with measured data--Part I: general methodology (RP-1051).
Topics:


Related Articles
Classroom health report hardly a breath of fresh air.
Improving indoor office environment boosts productivity.
Some quantitative relations between indoor environmental quality and work performance or health.
Kindergarten to 1st grade: classroom characteristics and the stability and change of children's classroom experiences.
Will work for air.
Monitored indoor moisture and temperature conditions in humid-climate US residences.
The effects of outdoor air supply rate and supply air filter condition in classrooms on the performance of schoolwork by children (RP-1257).
Experimental evaluation of a downsized residential air distribution system: comfort and ventilation effectiveness.
Should we expect more from CFD analysis?

Terms of use | Copyright © 2014 Farlex, Inc. | Feedback | For webmasters