Satisfaction with indoor environmental quality in BREEAM and non-BREEAM certified office buildings

ABSTRACT This paper presents preliminary analysis of occupant satisfaction with indoor environmental quality in BREEAM (Building Research Establishment Environmental Assessment Method) and non-BREEAM certified offices in the UK. Results from cross-sectional questionnaires (N = 121) showed that BREEAM certification per se did not seem to substantively influence building and workspace satisfaction. Conversely, occupants of BREEAM offices tended to be less satisfied with air quality and visual privacy than users of non-BREEAM buildings. Lower satisfaction was also detected in BREEAM offices for occupants having spent over 24 months in their building, and for users working in open-plan spaces. To interpret these findings, a methodology for data analysis was adopted whereas responses to point-in-time surveys (N = 82) were paired with environmental measurements. Broadening the perspective for appraising occupants’ perceptions, these combined techniques led to conclude that certification schemes should balance criteria addressing energy performance with design solutions considerate of issues of privacy, proxemics and perceived control over the qualities of the indoor environment.


Introduction
In December 2015, at the UN Conference of Parties in Paris, almost 200 nations set the goal to 'accelerate the reduction of global greenhouse gas emissions' (COP21 2015), pushing energy efficiency at the core of the building industry's sustainability agenda. These ambitions reinforce the prominent role that green certification schemes such as the Building Research Establishment Environmental Assessment Method (BREEAM) and Leadership in Energy and Environmental Design (LEED) are assuming at a global level. However, although these schemes embrace a wide range of environmental issues, there is a risk that a prevailing emphasis given to energy performance may depart attention from the physical, physiological and psychological impacts that the indoor environment has on building occupants.
Indoor Environmental Quality (IEQ) can be defined as 'the quality of a building's environment in relation to the health and wellbeing of those who occupy the space within it' (NIOSH 2016). IEQ includes factors such as temperature, air quality, noise, natural and artificial lighting, views, and visual and sound privacy. In the workplace, users' IEQ satisfaction has been associated to their comfort, health, wellbeing and self-estimated job performance (Frontczak et al. 2012). Considering that occupants greatly impact on buildings' energy use (Janda 2011), a vast body of research has studied the influence of physical parameters of the indoor environment on user perception (Frontczak and Wargocki 2011), and the contribution of rating tools to occupant satisfaction. Among other studies, the In response, this paper offers a preliminary analysis of occupant IEQ satisfaction in recently built BREEAM-rated office buildings in the UK, and compares responses with those provided by users of non-BREEAM certified buildings similar in age, function, size and location. In addition, this study explores how factors that are unrelated to conventional measures of environmental quality (e.g. time spent in the building, spatial layout, etc.) might affect IEQ satisfaction in BREEAM and non-BREEAM buildings. This paper also aims to propose and test a methodology for interpreting the findings related to the evaluation of occupant IEQ satisfaction in buildings. Consistent with earlier studies, in fact, responses were primarily collected via cross-sectional (transversal) questionnaires based on the CBE survey (CBE 2016). However, to support inferences, pointin-time (right-now) surveys were also administered to occupants while basic physical measurements were taken at their workspace.

The BREEAM programme
In 1990, the UK Building Research Establishment (BRE) published a method for assessing, certifying and rating buildings based on 'sustainable values [ . . . ] ranging from energy to ecology' (BRE 2016a). Being the longest established method globally, BREEAM has awarded to date more than 550,000 certificates in 77 countries, and more than 2.2 million buildings have been registered for assessment since the scheme was launched. BREEAM had an initial focus on new office buildings at the construction stage. However, the scheme was gradually expanded to also cover inuse buildings, refurbishments and fit-outs, infrastructure, and communities (BRE 2016a).
The BREEAM system awards credits under nine categories: Energy; Health and Wellbeing; Land Use; Materials; Management; Pollution; Transport; Waste; and, Water. Further credits can be gained under the Innovation area. BREEAM encompasses both mandatory and optional credits. It is, however, a flexible system that can trade credits, that is, non-compliance in one area can be offset through compliance in another. The performance of a project assessed by BREEAM is determined by a number of elements: the scope of the assessment; the rating level benchmarks; the minimum standards required; the environmental section weightings; the BREEAM assessment 'issues' and credits, and how these elements combine to produce a BREEAM rating.
As an example, for international fully fitted non-residential new construction buildings (BRE 2016b), the Health and Wellbeing category weighs 14% of the total score attainable, and assigns credits to the following issues: visual comfort; indoor air quality; safe containment in laboratories; thermal comfort; acoustic performance; accessibility; hazards; private space; and, water quality. Some issues include minimum standards that require compliance depending on the targeted rating level: visual comfort (high frequency ballast), indoor air quality (no asbestos), accessibility, private space and water quality (minimize legionellosis risk). Other issues are not compulsory for certification, although they contribute to the final score. The BREEAM rating benchmarks achievable are: Unclassified (percentage score < 30), Pass ( ≥ 30), Good ( ≥ 45), Very Good ( ≥ 55), Excellent ( ≥ 70) and Outstanding ( ≥ 85) (BRE 2016b).

Building selection
The criteria for the selection of the buildings featured in this study required them to be comparable in terms of design brief, geographical location, size, age of construction, distribution and type of occupants' activities, function, etc., and to have received -or have applied for -certification with the BREEAM rating system. This aimed to ensure that differences in the data could be associated essentially to the buildings' BREEAM certification, and that no other physical or organizational factor affected the comparison.
Four buildings were chosen for this preliminary study, all hosting office-type activities. The buildings all included private, shared and open-plan workspaces, had a number of floors ranging between 3 and 4, a size from 3000 to 3200 m 2 , were built between 2011 and 2012, were owned by the same institution, and were located in the UK's East Midlands area. In terms of operation strategies, all buildings featured a mixed-mode ventilation system and relied on a balance between natural and artificial lighting. Although all buildings responded to the same sustainable building strategic brief, two achieved certification by BREEAM in 2013 (respectively, Outstanding and Excellent), while two marginally failed to obtain the minimum score required for the targeted Excellent BREEAM rating, and lower certification was not pursued since two mandatory credits related to commissioning and microbial contamination were found to be not achievable. The BREEAM-certified buildings received, respectively, 10 and 7 points in the Health and Wellbeing category including, among others, credits for glare control, internal and external lighting levels, thermal comfort, and acoustic performance.

Data collection
Data were collected in two successive phases, based on two methodologies: cross-sectional (transversal) questionnaires; and, point-in-time (right-now) surveys (Privitera 2016). Table 1 summarizes the two datasets used in the analysis.
Cross-sectional questionnaires (online) were sent to all the occupants of the selected buildings. Coherent with the structure of the CBE survey, the questionnaire featured an initial section enquiring about participants' sex, age, time spent in the building and at their current workspace, the location of the workspace, its orientation, proximity to windows and spatial layout (i.e. private office, shared office, cubicle, open space). The questionnaire then asked occupants to rate -on a Likert scale ranging from very dissatisfied (−3) to very satisfied (+3) with a neutral midpoint (0) -their satisfaction with: building; workspace; ease of interaction; building cleanliness; amount of  Point-in-time surveys (paper-based) were distributed to volunteering occupants for them to fill in while physical measurements of basic environmental parameters were taken at their workstation with calibrated hand-held equipment. The survey collected information on satisfaction with luminous, acoustic and thermal conditions, and perceived control over these factors, and offered users the opportunity to give comments on the characters of their workspace. While the survey was filled, a monitoring sheet was completed where measurements were recorded. Table 2 illustrates the equipment and environmental parameters used in this study. All surveys were administered in the month of June, during fully occupied working hours, between 9am and 11am. For each variable, three measurements were taken, and values were mean averaged for data analysis. Vertical illuminance was taken from the point of view of the user facing the visual task (e.g. computer screen).
To perform statistically robust comparisons between occupants' responses in BREEAM and non-BREEAM certified buildings, the two independent groups needed not only to be homogenous in terms of location, size, function and year of construction of the buildings featured in each, but also had to be similar in sample size. In addition, distribution of responses based on several non-environmental factors -that is, 'factors unrelated to environmental quality that influence whether indoor environments are considered to be comfortable' (Frontczak and Wargocki 2011) -was also considered, since earlier research had revealed that these might affect significantly the IEQ satisfaction of occupants at their workplace (Schiavon and Altomonte 2014). Table 3 presents a distribution of occupants' responses to the cross-sectional questionnaires based on consideration of non-environmental factors, showing that the two groups (BREEAM and non-BREEAM) provide comparable subsets for the purpose of this study. To be noted that, in terms of spatial layout, two of the categories of workspace type normally featured in the CBE survey -cubicles with high and low partitions -were merged together to obtain a more evenly distributed sample.

Cross-sectional questionnaires
The analysis of cross-sectional questionnaires (N = 121) initially consisted in calculating descriptive statistics (e.g. mean, standard deviation, median, interquartile ranges) of votes of satisfaction with the building, workspace and various categories of IEQ in BREEAM and non-BREEAM buildings. Exploratory inspection of the data (e.g. Q-Q plots and Kolmogorov-Smirnov tests) revealed non-normal distribution of statistical values, thus violating one of the assumptions for the adoption of parametric tests. Since data had an ordinal character, the statistical significance (NHST, Null Hypothesis Significance Testing) of the difference in median votes of satisfaction between the two independent groups ( M dn , BREEAM minus non-BREEAM) was tested with a two-tailed non-parametric Wilcoxon rank-sum test. Individual responses in each independent group were considered in the analysis instead of average building values. This was to avoid loss of information (e.g. variance) considering that, at the building level, the sample size was small. Results were declared statistically significant when the probability that a difference could have arisen by chance was below 5% (p ≤ .05). However, one of the limitations of NHST is that the p-value depends both on the size of the sample and on the size of the influence tested. Therefore, the mean ranks for each group were calculated, and the effect size was estimated for each comparison (Field 2013).
The effect size coefficient places the emphasis on the most essential element of the analysis -that is, the standardized size of the difference between groups, and not just its statistical significance -therefore providing a more reliable estimator to infer whether the differences detected have any practical relevance (Nuzzo 2014;Schiavon and Altomonte 2014). In this study, the effect size was calculated by making use of equivalence with the Pearson's r correlation coefficient, using the equation: Effect size = Z-score/ √ N, where the Z-score was provided by the Wilcoxon tests, and N was the number of observations (Field 2013). The interpretation of the outcome was derived from Ferguson (2009), where benchmarks are provided for small, moderate, and large effect sizes (r ≥ 0.20, 0.50 and 0.80, respectively). Values of r < 0.20 were considered negligible, and therefore not providing any substantive -that is, practically relevant (Field 2013) -effect. In this analysis, the interpretation of the effect size was based on its absolute value, that is, the magnitude of the effect was benchmarked irrespective of its sign. It should be noted that the detection of effect sizes of small magnitude is customary in user-assessment studies. The use of this terminology, however, should not detract from the substantive value of the outcome, and reflects the practical relevance of the effects detected (Field 2013). The same methods were adopted for consideration of differences based on distribution of responses according to non-environmental factors.

Point-in-time surveys
Measurement of environmental parameters taken in BREEAM and non-BREEAM buildings during the administration of point-in-time surveys (N = 82) were statistically compared using two-tailed non-parametric Wilcoxon rank-sum tests (data were non-normally distributed), and the effect sizes of differences were calculated (Pearson's r).
In order to correlate physical measurements with the responses provided by participants, Jonckheere-Terpstra (J-T) tests were performed. These are rank-based non-parametric tests that require independent groups divided into ranked orders to search for statistically significant trends between (continuous or ordinal) independent and dependent variables (Jonckheere 1954). Dependent variables were measured at the ordinal level based on 7-point Likert scales (e.g. from 'no discomfort' to 'a lot of discomfort'). In this case, the effect size (Pearson's r) was used to measure both the magnitude and the directionality of the trend, that is, whether there was a direct or inverse relationship (positive or negative sign) between variables. The interpretation of the outcome was again derived from Ferguson (2009). For lighting and noise, the physical readings -illuminance (lux) and sound pressure levels (dB(A)) -were directly used in the statistical tests. For thermal sensation, since the buildings were not free-running, measures of dry bulb temperature, humidity, air speed and mean radiant temperature (derived from globe temperature), were combined with estimations of metabolic rate and clothing levels to determine the Predicted Mean Vote (PMV), which was calculated via the CBE Thermal Comfort Tool web application (comfort.cbe.berkeley.edu) according to ASHRAE Standard 55 (Schiavon, Hoyt, and Piccioli 2014).
The analysis was performed with SPSS statistical software version 21. Table 4 provides the descriptive and inferential statistics from the analysis of the cross-sectional questionnaires (N = 121). For each category, the table presents the mean, standard deviation, median and interquartile ranges of occupants' satisfaction votes in BREEAM and non-BREEAM buildings, the median differences ( M dn ) and the interpretation of their two-tailed statistical significance (NHST expressed in terms of p-value), the mean ranks of independent groups, the Wilcoxon test statistic (W) and the effect sizes (r). The plotting order of categories follows the ranking presented in Altomonte and Schiavon (2013) so as to facilitate a visual comparison of results with previous work. Values in bold italic are statistically significant (p ≤ .05) and have substantive magnitude of effect (r ≥ 0.20, absolute values were considered for interpreting the practical relevance of effect sizes).

Occupant satisfaction in BREEAM and non-BREEAM buildings
Analysis of descriptive statistics in both independent groups revealed positive mean (M) and median (M dn ) scores of satisfaction with the building and with the workspace. The inferential tests showed that users of BREEAM and non-BREEAM offices expressed equal satisfaction with these two categories, as per the non-statistically significant median differences between groups ( M dn ) and the effect sizes of negligible magnitude (r = −0.16) (Table 4). For other IEQ categories, satisfaction votes in both BREEAM and non-BREEAM buildings showed positive or neutral mean and median values, except for visual privacy and sound privacy. The inferential tests revealed that BREEAM-rated buildings had equal or lower median scores of satisfaction with these IEQ categories than non-BREEAM buildings ( M dn values are, in fact, always zero or negative), although the differences detected were statistically significant only for satisfaction with amount of space, air quality, visual privacy and sound privacy.
Satisfaction with air quality showed a significant median difference with the largest practically relevant effect size (r = −0.27). This suggests a trend for higher occupant satisfaction with air quality in buildings not certified by BREEAM. Consideration of visual privacy detected higher occupant satisfaction in non-BREEAM buildings, as denoted by a statistically significant difference between groups and an effect size of substantive relevance (r = −0.20). Inferential results for amount of space and sound privacy showed tendencies for higher satisfaction in non-BREEAM buildings, this being supported by statistically significant differences, although effect sizes were slightly lower than the borderline of practical relevance (r = −0.18) ( Table 4). The results of the inferential tests are graphically summarized in Figure 1.

Influence of non-environmental factors on occupant satisfaction
Tables 5-9 present selected results of the inferential tests for the satisfaction votes expressed by occupants upon consideration of their sex (Table 5), time spent in the building (Table 6), time spent at the workspace (Table 7), distance from windows (Table 8) and spatial layout (Table 9). Data related to the age groups of users have not been reported in tables since no statistically significant differences were detected. The tables present the sample sizes of independent groups (x 0 = BREEAM, x 1 = non-BREEAM), the median and interquartile range of satisfaction votes, the median differences ( M dn ) and the interpretation of their two-tailed statistical significance (NHST), the mean ranks, the Wilcoxon test statistic (W) and the effect sizes (r).

Sex
As shown in Table 5, the inferential tests based on consideration of occupants' sex did not detect statistically significant differences in satisfaction with the building and with the  Note: Values in bold italic denote statistically significant differences (p ≤ .05) and substantive effect sizes (r ≥ 0.20, in absolute value).
n.s.: not significant. workspace for male and female users of BREEAM and non-BREEAM buildings. Analysis of differences in satisfaction for all other IEQ categories revealed that median votes given by males were often higher than females both in BREEAM and non-BREEAM buildings, and were consistently positive except for sound privacy and visual privacy. When comparing satisfaction scores given by males in BREAAM and non-BREEAM buildings, no statistically significant differences were detected. Conversely, analysis of votes from female users detected statistically significant and practically relevant higher satisfaction with amount of space (r = −0.27), air quality(r = −0.55), visual privacy (r = −0.43), temperature (r = −0.28) and sound privacy (r = −0.29) in buildings not certified by BREEAM.

Age
Analysis of satisfaction votes expressed by occupants from different age groups (under 30, 30-40, 41-50, over 50) did not show statistically significant differences between BREEAM and non-BREEAM buildings, and for this reason these data have not been reported in a table format. However, effect sizes of substantive

Time spent in the building
For occupants who spent less than 12 months in their building, the median votes of satisfaction were consistently positive for all variables considered, except for the satisfaction with temperature expressed by users having occupied their BREEAM building for 6-12 months (M dn = −0.50). Among other inferential tests, Table 6 presents the results for the satisfaction votes provided by users having spent over 24 months in their building. The data reveal in non-BREEAM offices a statistically significant and practically relevant higher satisfaction with workspace (r = −0.38),  ). An analogue tendency was detected for satisfaction with the building, although such difference had a substantive effect size (r = −0.27), but it was not statistically significant. Similar results of practically relevant but not statistically significant differences were also found for higher satisfaction with noise (r = −0.25) and sound privacy (r = −0.36) in non-BREEAM offices for users having occupied their building for 12-24 months.

Time spent at the workspace
As per the data of Table 7, participants who spent over 24 months at their workspace in a non-BREEAM building expressed statistically significant and practically relevant higher satisfaction with building cleanliness (r = −0.39), amount of space (r = −0.37), visual privacy (r = −0.49) and sound privacy (r = −0.36). For this group of users, similar tendencies were detected also for satisfaction with workspace (r = −0.28), visual comfort (r = −0.33), air quality (r = −0.23), noise (r = −0.32) and temperature (r = −0.22); these differences were not statistically significant, yet suggesting a trend for higher satisfaction in non-BREEAM buildings. For other groups of occupants, differences in satisfaction between BREEAM and non-BREEAM offices were consistently not statistically significant, with the exception of a significant higher satisfaction with air quality in non-BREEAM buildings expressed by users having occupied their workspace for 6-12 months (r = −0.52).

Distance from windows
Satisfaction votes provided by occupants whose workstation was within 4.6 m (15 feet) from windows were consistently higher than those expressed by users sitting far from the perimeter in both BREEAM and non-BREEAM buildings, as shown in Table 8. However, no statistically significant differences in satisfaction could be detected for this group of occupants between certified and non-certified offices. Conversely, users sitting further than 4.6 m from windows expressed statistically significant and substantive higher satisfaction with the building, workspace and almost all IEQ categories in non-BREEAM buildings. These significant differences in satisfaction ranged from small (colours and textures, r = −0.29) to moderate (sound privacy, r = −0.50) effect sizes. The only exceptions were represented by satisfaction with ease of interaction and temperature, which resulted in non-statistically significant differences and effect sizes marginally lower than the benchmark for practical relevance (r = −0.19 and −0.17, respectively), although following a trend for higher satisfaction in non-BREEAM buildings.

Spatial layout
Median satisfaction votes expressed by occupants of enclosed offices (private and shared) were positive in both BREEAM and non-BREEAM buildings. For these spatial layouts, inferential tests did not detect statistically significant differences, even if effect sizes of mostly practical relevance suggested higher satisfaction in non-BREEAM buildings. For users of cubicles, differences varied depending on IEQ category, but were consistently not significant. The inferential data related to the satisfaction expressed by users working in open-plan offices are presented in Table 9. For these occupants, statistically significant and substantive higher satisfaction with the building, workspace and almost all IEQ categories were detected in non-BREEAM buildings.  Note: Values in bold italic denote statistically significant differences (p ≤ .05) and substantive effect sizes (r ≥ 0.20, in absolute value). r < 0.20 = negligible; 0.20 ≤ r < 0.50 = small; 0.50 ≤ r < 0.80 = moderate; r ≥ 0.80 = large. ***p ≤ .001 = highly significant. **p ≤ .01 = significant. *p ≤ .05 = weakly significant. n.s.: not significant.

Point-in-time surveys and physical measurements
recorded (mostly ranging between 0.0 and 0.1 m/s). All differences between values measured in BREEAM and non-BREEAM buildings were non-statistically significant nor practically relevant, with the only exception of relative humidity (r = −0.50).
Since the environmental conditions of workspaces in BREEAM and non-BREEAM buildings were substantially similar, and in line with regulatory values for office spaces, the responses to the point-in-time surveys were paired by J-T tests with the physical measurements taken onsite for light, sound and thermal sensation (N = 82). This aimed to detect direct or inverse relationships between responses from users and measured data, explore differences between BREEAM and non-BREEAM buildings, and contribute to the interpretation of the results from the cross-sectional questionnaires.
Tables 11-13 present the data from the J-T tests for light, sound and thermal sensation. The tables provide uniquely the results of the tests for which statistical significance or practical relevance was detected. For each measured variable, the tables report the building group, the J-value, the test statistic (Z-score), the two-tailed statistical significance of differences (p-value) and the effect sizes (r). The estimation of statistical significance was supported by calculation of Monte Carlo simulated lower and upper 99% confidence intervals (not reported in tables). Values in bold italic denote statistically significant differences (p ≤ .05) and substantive effect sizes (r ≥ 0.20, the magnitude of the effect size was interpreted considering its absolute value).
not found in non-BREEAM buildings (Table 12). A significant and substantive inverse relationship appeared between perceived control over noise and dB(A) levels in BREEAM offices (p = .01**, r = −0.37), while a significant and practically relevant direct trend was detected in non-BREEAM buildings (p = .02*, r = 0.39).

Thermal sensation
A highly significant and practically relevant direct relationship was detected between users' description of thermal sensation (ranging from 'cold' to 'hot') and calculated PMV in BREEAM buildings (p < .001***; r = 0.51). This trend was also substantiated by results in non-BREEAM buildings, although at lower level of significance and effect size (p = .01**; r = 0.44). The tests considering the relationships between perceived control over the thermal environment and calculated PMV did not detect any significant nor substantive trend, and therefore have not been reported in Table 13.

Discussion
This study sought to provide a preliminary analysis of occupant IEQ satisfaction in BREEAM and non-BREEAM-rated office buildings, and investigate if BREEAM certification has a statistically significant and practically relevant influence on satisfaction with the building, workspace and several IEQ categories.
Although, consistent with the literature (Frontczak et al. 2012), occupants were in general reasonably satisfied with their indoor environment (i.e. mostly positive mean and median satisfaction votes), rigorous statistical analysis of the data from the cross-sectional questionnaires leads to infer that the achievement of BREEAM certification per se does not have a significant and substantive influence on satisfaction with the building and the workspace. Conversely, users of non-BREEAM buildings expressed a statistically significant and practically relevant higher satisfaction with air quality and visual privacy. Tendencies also suggested that users of non-BREEAM offices might be more satisfied with sound privacy and amount of space (Table 4). These results are coherent with previous research by the authors (Altomonte and Schiavon 2013), where the achievement of LEED certification was found not to substantively affect occupant satisfaction with the building and the workspace. Also, in line with earlier studies, satisfaction with sound privacy, visual privacy, temperature, air quality and noise corresponded to the lowest mean and median scores in BREEAM buildings. Issues related to lack of privacy are recurrent in green-buildings research , likely due to the incentive towards the design of open spaces that can support the achievement of credits for natural ventilation and daylight penetration. However, previous studies on LEED-rated buildings detected higher satisfaction with air quality (Newsham et al. 2013), a result that is not supported by our study. This could be explained by the two prerequisite credits for minimum indoor air quality performance and environmental tobacco smoke control that are compulsory for a new building to obtain LEED certification (USGBC 2016), while only one air quality credit related to the absence of asbestos is mandatory for BREEAM rating (BRE 2016a).
In terms of the influence of non-environmental factors on occupants' responses, consideration of sex did not lead to detect significant differences in satisfaction with the building and the workspace between BREEAM and non-BREEAM offices, although female users expressed higher satisfaction with various IEQ categories in buildings not certified by BREEAM. Analysis of satisfaction votes also revealed that males tended to be more satisfied with the qualities of their indoor environment than females (Table 5). These findings are consistent with those of , who found that female occupants were significantly more likely to express dissatisfaction with IEQ than males.
In line with the findings of Frontczak and Wargocki (2011), age groups could not be associated to significant differences in occupant satisfaction.
Inferential tests revealed that IEQ satisfaction tended to decrease with the increase in time spent in the building and at the workspace, this being particularly evident in BREEAMrated offices. In addition, users who spent over 24 months in their BREEAM-certified building and workspace expressed statistically significant and practically relevant lower satisfaction with their workspace and with several IEQ categories than occupants of non-BREEAM buildings (Tables 6 and 7). These results are consistent with those of Schiavon and Altomonte (2014), who concluded that users of LEED-rated offices having spent less than one year at their workplace had higher IEQ satisfaction than users who occupied their building for more than 12 months. In this context, Singh et al. (2010) suggested that IEQ satisfaction might be higher immediately after moving into a new green building, hence questioning the positive effect of green certification on occupants' perception over time. It must be emphasized that the number of study participants having occupied their building and their workspace for over 24 months was broadly similar between BREEAM and non-BREEAM buildings (Table 3). Conversely, a larger percentage of users had occupied their workspace for less than 6 months in non-BREEAM buildings (41% against 28% in BREAAM offices). This could have brought a potential source of bias in comparing occupants' assessments of the qualities of their indoor environment. However, no statistically significant differences in satisfaction between BREEAM and non-BREEAM offices were detected for users who had only recently (0-6 months) moved to their workspace.
Results related to consideration of distance from windows (Table 8) and spatial layout (Table 9) are in line with those of Leder et al. (2016), who stated that access to a window positively affects workplace experience and suggested that IEQ satisfaction is higher in enclosed offices, a conclusion that is supported by our data. In our study, the spatial layout had considerable influence on the difference in satisfaction between BREEAM and non-BREEAM buildings, although -contrary to previous research (Schiavon and Altomonte 2014) -occupants of openplan offices showed to be significantly and substantively more satisfied with almost all IEQ categories in buildings not certified by BREEAM. These results can be explained by the findings from the pairing of the point-in-time surveys with the physical measurements, as discussed below.
The J-T tests related to consideration of the luminous environment, in fact, detected no significant or practically relevant relationship in BREEAM buildings between measured illuminance levels (horizontal and vertical), users' assessment of lighting availability, their perception of control over it and reported discomfort. Conversely, direct associations were found between reported luminous qualities and measured parameters in non-BREEAM offices (Table 11). These findings lead to infer that perception of lack of control over lighting in BREEAM buildings -particularly in open-plan layouts, as per the analysis of the comments provided -could have resulted in a luminous assessment that was effectively detached from fluctuations in illuminance levels. This might have ultimately led to lower satisfaction with the qualities of the indoor luminous environment. Conversely, perception of personal control over lighting was reported in non-BREEAM buildings, allowing users to directly intervene at the occurrence of temporary visual discomfort, and therefore enhancing feelings of satisfaction with illuminance conditions.
In terms of the aural environment, in BREEAM buildings a direct relationship was found between measured acoustic parameters and users' description of noise, while an inverse trend was detected between decibel levels and perception of control over noise. Conversely, a direct relationship was found between sound measurements and reported level of control in non-BREEAM offices (Table 12). In interpreting these findings, it should be reminded that statistically and practically significant lower satisfaction with noise and sound privacy was detected in the cross-sectional questionnaires for users working in BREEAM-certified open spaces. Similar results were also found for occupants whose workstation was located further than 4.6 m from a window, this being often the case in an open-plan office (respectively, satisfaction with noise: r = −0.43; satisfaction with sound privacy: r = −0.50). This suggests that users of BREEAM offices might have been more sensitive to sound and to disturbance from noise than occupants of non-BREEAM buildings. This higher sensitivity might be more evident in open workspaces where direct control over the aural environment could be perceived as more challenging .
Finally, for the thermal environment, a direct relationship was detected between reported thermal sensation and calculated PMV in BREEAM buildings. This relationship had larger magnitude than the same tendency found in non-BREEAM offices (Table 13). This leads to infer that occupants of BREEAM-rated workspaces might have been more sensitive to changes in their thermal environment than users of non-BREEAM buildings. However, no significant trend was detected in either groups of buildings for the relationship between perception of thermal control and calculated PMV. This is in contrast with the analysis of openended comments provided by occupants of BREEAM buildings, who often related their higher dissatisfaction with temperature to a perceived lack of control. This suggests that, in rich dynamic working spaces, the complex influence of a number of physical, physiological and psychological variables should be comprehensively considered when evaluating thermal expectations and experience (Brager, Zhang, and Arens 2015;Parkinson and De Dear 2015).

Conclusions
The main conclusions to be drawn from this study are: • In the dataset analysed, BREEAM rating per se did not seem to significantly and substantively affect occupant satisfaction with the building and the workspace. • Occupants of non-BREEAM-rated buildings showed trends for significant and substantive higher satisfaction with air quality and visual privacy than users of BREEAM-certified offices. Tendencies were also detected for users of non-BREEAM buildings to be more satisfied with sound privacy and amount of space. • Lower satisfaction with most IEQ categories was detected in BREEAM offices for occupants having spent more than 24 months at their building and workspace, and for users working in open-plan layouts. • Pairing of occupants' responses with physical measurements led to infer that lower satisfaction in BREEAM buildings, particularly in open workspaces, might be associated with a perceived lack of control over the luminous, aural and thermal environments.
In interpreting these results, some limitations should be acknowledged. First of all, only a narrow sample of buildings and a limited number of responses were used for the analysis. Also, the buildings were chosen to be as similar as possible for them to be statistically comparable, and they were all located in the same geographic area, so they cannot be representative of all buildings certified by BREEAM. Moreover, only basic environmental parameters were recorded in the workspaces analysed. Finally, occupant responses have not been related to the distribution of BREEAM credits targeted or attained by buildings in the Health and Wellbeing category.
Regardless these limitations, this study has provided some useful preliminary data on which further research, on larger samples and supported by the recording of more detailed and varied environmental parameters (e.g. air quality), can be developed. In the sample used for this analysis, occupants were reasonably satisfied with their building and workspace. This is a testament to the efforts devoted by designers and green certification systems to provide comfortable working environments. However, consistent with previous research on other rating systems (e.g. LEED), the findings from this study suggest that, to improve workplace experience, BREEAM might benefit from balancing the credits that directly address criteria of visual, acoustic, air quality and thermal performance, with design solutions and spatial strategies that are considerate of issues of privacy and proxemics (e.g. amount of space), and are conducive to perceived control over the qualities of the indoor environment. The results also suggest the need for rating systems to reinforce mandatory criteria that can guarantee minimum standards in specific areas (e.g. air quality), and support the requirement for further research on the sustained benefits of certification over time.
Far from being a criticism of BREEAM or other rating schemes, studies such as that presented in this paper can provide evidence-based data to improve the standards promoted and achieved in green certification, whereas the emphasis given to energy performance should not come to the detriment of IEQ and user satisfaction. Also, they can propose and test methodologies -relatively new to green-building research -for assessing the effectiveness of certification schemes from the occupants' point of view through a combination of cross-sectional questionnaires, point-in-time surveys and physical measurements. The use of these techniques, and the application of appropriate methods of statistical testing, can reinforce the rigour of the analysis and broaden the perspective for interpreting the information provided by users. As pointed out by Allen et al. (2015), in fact, one of the strongest limitations of the research in this field is related to the frequent reliance on indirect and abstract indicators, without a direct appraisal of the factors that mostly impact on the perception that occupants have of the qualities of their indoor environment. This study has intended to offer a methodological contribution in this direction.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was supported by the International Collaboration Fund awarded by The University of Nottingham to the first author, and by the Developing Solutions Scholarship (number 15137) awarded by The University of Nottingham to the second author for the completion of her MSc in Sustainable Building Technology.