Complexity of the Task and the Committee’s Approach
The study committee was charged to review and assess existing empirical and theoretical studies on the possible connections between characteristics of green schools and the health and performance of students and teachers. A number of factors make this task more complex than might first be evident. These include lack of a clear definition of what constitutes a green school; variations in current green school guidelines; the difficulty of measuring educational and productivity outcomes; the variability and quality of the research literature; and confounding factors that make it difficult to separate out the effects of building design, operations, and maintenance from the effects of other variables. Each of these factors is discussed below to provide context for the committee’s approach to meeting its charge.
COMPLEXITY OF THE TASK
Lack of Clear Guidance About the Characteristics of Green Schools
Ideally, the committee’s evaluation would be based on a single clear definition of a green school and/or guidance about the characteristics of green schools that differentiate them from conventional schools. However, as noted in Chapter 1, green school-related definitions and guidelines vary in the objectives to be achieved, individual elements, and level of detail. Some green school guidelines, such as California’s, go beyond environmental objectives, specifying objectives for occupant health and productivity. This variation in definitions and guidelines is understandable given
the grass roots nature of the green building movement and the fact that the guidelines were developed by a diversity of organizations. However, it did complicate the committee’s task.
In reviewing examples of green school objectives and guidelines for California and Washington State and draft guidelines for Massachusetts, the committee determined that such documents have two complementary, but not identical, goals: (1) to support the health and development (physical, social, and intellectual) of students, teachers, and other staff by providing a healthy, safe, comfortable, and functional physical environment and (2) to have positive outcomes for the environment and the community. Thus, research on green schools might be conceptualized as having two quite different outcomes: improved student and staff health and development or improved environment and community. Because they were first developed to minimize adverse environmental impacts, current green school guidelines place less emphasis on supporting human health and development. Accordingly, and in line with its charge, the committee focused on outcomes associated with student and teacher health, learning, and productivity.
Measuring Educational and Productivity Outcomes
An assessment of the effects of a school building on student learning and health and teacher health and productivity must be set in context. First, time spent in school is, at most, 40 to 50 hours per week, and other environments, including home, neighborhood, recreational, cultural, and religious settings, could equally affect health and performance. In addition, differing populations and individuals may have differing sensitivities and responses to features or attributes of the built environment.
Education, which is the transfer of knowledge and skills to people (learning), is also difficult to measure directly. Learning is influenced by many factors, including the quality of curriculum, teacher education/ experience, parental support, peer support, student background, quality of the administration, instructional materials, laboratory equipment, and educational standards. Policy researchers suggest that teaching and learning might be shaped by various state policies and their implementation regarding teacher education, licensing, hiring, and professional development and by national policies such as the No Child Left Behind Act (P.L. 107-110).
Measuring productivity is equally difficult. Productivity for an individual or an organization has been defined as the quantity and/or the quality of the product or service delivered (Boyce et al., 2003). Productivity is influenced by both the individual and the system within which he or she works. Increasing evidence is available to indicate that the built envi-
ronment can influence both individual and organizational productivity, which can be measured, in some cases, by the number of units manufactured or the number of words typed correctly in a given amount of time. Absenteeism is often used as a surrogate for productivity, the rationale being that people who are absent are less productive than people who are on the job.
For work like teaching that cannot be measured in units, productivity is more difficult to define and measure. Productivity is closely linked to the “quality” of teachers, both individually and collectively. Researchers have used various measures such as educational preparation, certification/licensure, and scores on teacher examinations (Praxis is one) to try to objectively evaluate the quality of teaching staff. These measures can be quantified to a certain extent and are used quite efficiently in research studies. They have not been tested sufficiently, however, to enable a great deal of confidence in their validity. For instance, a teacher’s years of experience is to a degree a measure of quality, because a teacher is usually evaluated in conjunction with a decision about the granting of tenure. To that extent, an administration may judge a teacher’s quality to be high enough, given his or her satisfactory performance during the preceding years, that it decides to grant tenure. Conversely, there is no guarantee that someone gaining tenure is doing anything more than meeting the minimum standards. Also, the basis for awarding tenure is not consistent across school districts. Thus, the same level of performance might gain a teacher tenure in one school district but not in another.
Researchers cannot directly access an individual’s personal experiences but must gain access to them indirectly by measuring outcomes such as learning, productivity, or attendance or asking them to reveal their preferences by means of surveys. In the studies reviewed by the committee, multiple and quite varied measures were used. One measure, absenteeism, was used as a surrogate measure for student health, teacher health, student learning, and teacher productivity (Table 2.1).
Because of the multiplicity of variables that can influence learning, research that evaluates educational interventions does not often find a large effect on student achievement outcomes. Interventions such as intensive in-service education for teachers, major revisions of curriculum, and addition of curricular units on substance abuse prevention or economics tend to show small gains in student achievement at best.
Similarly, it is very difficult to draw firm inferences about cause-and-effect relationships between a physical intervention—say, school design—and health and performance outcomes for individuals. Thus, it is not unexpected if research on building characteristics or operation practices finds only a small effect on health, learning, or productivity.
TABLE 2.1 Examples of Indicators Used to Measure Relevant Outcomes
Cold/flu and other respiratory diseases
Cold/flu and other respiratory diseases
Development (learning and productivity)
Standardized test scores
Demonstration of concepts
Teacher examination scores (Praxis)
The Variability and Quality of Research
The committee’s task was further complicated by gaps in the research literature. For example, much of what is known about the impact of the physical environment on health and performance is based on studies of adult populations. Some research is quite narrow because it makes an effort to control for all possible extraneous variables. Other research is so broad that confounding variables are possible, if not likely.
In general, much less is known about the impact of the school environment—green or conventional—on children’s health and learning as compared with the impact of workplace environments on adults’ health and productivity. Extrapolating from studies of adults to draw conclusions about a much younger population can be suspect because children are still developing physically, mentally, and emotionally. In addition, environmental factors in a school may interact with genetic factors to determine the degree to which a child develops language skills or displays asthma symptoms.
An additional challenge is the great variability in the characteristics and standards used by different disciplines to conduct research. For example, medical research may employ clinical trials and intervention studies in which various factors can be controlled for and their results directly evaluated using established protocols. Often clinical trials of drugs include placebos administered according to the same protocol as the drug of interest. Epidemiological studies often rely on statistical significance as a
quantitative measure of the extent to which chance or sampling variation might be responsible for an observed association between an exposure and an adverse event. In these studies, quantitative estimation is firmly founded in statistical theory on the basis of repeated sampling.
Social science research testing links between the built environment and the behaviors of occupants, in contrast, cannot set up strictly controlled trials or easily manipulate variables to test for statistical significance. Studies attempting to link health, students’ learning, and teachers’ productivity with school environments cannot control for the effects of other, nonschool environments. In addition, there are no standard protocols for conducting building-related research, although some studies have used similar methodologies or evaluation methods, including multiple regression analyses and measures of statistical significance.
The committee was asked to consider the possible influence of confounding, bias, error, and chance in the relevant literature. Confounding, in particular, poses a major challenge to researchers and those evaluating their work. Confounding occurs when the evaluation of a relationship between two features or groups is biased by a third factor. For example, in any epidemiological study comparing an exposed group with a non-exposed group, a simple comparison of the groups may exaggerate the true difference or hide it, because it is likely that the two groups will differ with respect to factors that are also associated with the risk of the outcome of interest, such as socioeconomic status. Said differently, a simple comparison of the incidence of health outcomes among exposed and nonexposed groups may exaggerate an apparent difference, because socioeconomic status is also thought to influence the incidence of several health problems.
A variety of confounding factors will be present in any study that attempts to link features of school buildings with student and teacher health and development. Age differences among students will influence the outcomes of research into health and learning: Young children inhale more air per pound of body weight than teenagers; their tissues and organs are actively growing; they are still developing language and cognitive skills; and they spend more time in school than anywhere else but home. Extrapolating findings from studies of adults and applying those findings to children and teenagers can also be problematic.
Research studies that include measures of student learning in making comparisons to other variables must try to control extraneous variables. For instance, when comparing student achievement test scores with the condition of the physical environment, a researcher would typically
attempt to control other variables that influence outcomes, including the socioeconomic status of the students or teachers and the curriculum. Student test scores will also be influenced by conditions in the classroom on the day of the test and other factors, like the amount of time spent on the task.
Another confounding factor is that school buildings themselves are not standardized, making direct comparisons between school environments problematic. Unlike the houses in tract developments, most schools are designed as unique structures, whose features depend on the resources available, construction methods, curriculums, populations, and building codes in effect at the time. They may have one or multiple stories, may accommodate a hundred or several thousand students, may be 5 or 100 years old; may be designed with open plans or traditional classrooms; may use central air handling systems or natural ventilation.
The condition of the school buildings is also a factor: School conditions diverge widely from one part of the country to another or even within localities. Those school systems with fewer financial resources and a greater percentage of students from low-income households seem to have school buildings in worse condition than those school systems that have a high tax base and the financial resources to solve the problems of providing good school buildings. In addition, it is difficult to find data about the condition of school buildings: The Educational Writers’ Association (1989) found that few states were able to properly evaluate school buildings because departments of education had few or no personnel to conduct such evaluations. Consequently, data collection was erratic.
Another confounding factor is that building systems and characteristics operate in an integrated fashion to effectively deliver (or not) overall building performance, whose components include thermal comfort; air, visual, acoustic, and spatial quality; and long-term building integrity. Building performance will be affected by the operation and maintenance of these integrated systems over time and by the occupants of buildings and their activities.
Interactions between people and built environments are numerous and difficult to account for in research studies. For instance, the physical and psychological health of teachers may be affected by a school building’s characteristics and conditions. When teachers’ well-being suffers, so too might their instruction and their participation in school activities. In that sense, teachers can be seen as both an outcome of building conditions (they might be healthier or not, more motivated or not) and a mediating variable (they might teach differently or interact with students differently depending on their health and well-being) in explaining students’ experiences. They may also positively or negatively alter a school’s physical setting by
adjusting the temperature of a room, opening or closing windows, and other actions.
Similarly, student learning and health might be directly or indirectly related to a school building’s condition: If there is high absenteeism at a school among teachers because of poor air quality, the quality of instruction may suffer. If students are absent because of poor air quality, they have less opportunity to receive instruction. And, just as teachers can alter the school environment, so too can students.
A third population that can affect and be affected by a school’s environment is the administrators (principals, financial staff, counselors, librarians) and support staff (building operations and maintenance personnel, cleaning crews, and kitchen workers). These groups may spend as much time in a school building as teachers and students and sometimes at different times of the day (before and after classes, on weekends, during school breaks and summer vacation). The quality of the support staff training may significantly affect the performance of building systems, the timeliness and quality of maintenance and repair, and cleaning practices.
With outcomes as complex as student health and learning—influenced as they are by many individual, family, and community factors—it might be theoretically possible to design research that controls for all potentially confounding variables, but it would be difficult to conduct such research.
THE COMMITTEE’S APPROACH
The line of reasoning inherent in this study’s task—mapping connections from physical environments to student and teacher outcomes—poses significant challenges, as described above. An additional challenge is the directionality of relationships. Green schools might have positive effects on student health, but it might also be that students who live in communities inclined to build green schools are healthier to begin with. Another challenge is that the effects of physical environments might be trumped in some way by other forces—teacher quality, parental involvement, or financial resources, for instance.
To help inform the committee about mechanisms by which the physical environment might affect student learning, teacher productivity, or the health of students and teachers, the committee developed a conceptual model for evaluating links between green school buildings and outcomes for learning, health, and productivity (Figure 2.1).
In the conceptual model, one underlying assumption is that a green school’s location and design (site, orientation, envelope, heating, ventilation, air conditioning, acoustics, lighting) will result in an indoor environment with appropriate (or inappropriate) levels of moisture, ventilation,
air quality, noise, light, and other qualities. It also assumes that the indoor environment will be modified by season (e.g., presence of airborne pollen) and over time (e.g., mold growth from chronic water leakage) and by operational, maintenance, repair, and cleaning practices. Finally, the model assumes that the indoor environment can affect student learning and health and teacher health and productivity.
The committee’s initial review of the literature focused specifically on identifying studies that purport to address the connections between green school design, student learning and health, and teacher productivity and health. No well-designed, evidence-based studies concerning the overall effects of green schools on the health or development of students and teachers were identified. The committee also did not identify any evidence-based studies that analyze whether green schools differ from conventional schools in regard to occupant health and productivity. A few studies were identified that examined specific building features often emphasized in green school design and the effects of these features on health and performance. Among those are a series of studies on daylighting and student achievement produced between 1999 and 2003. These studies are discussed in Chapter 5. For the most part, however, the
literature on green schools, health, and productivity consists of anecdotal information and case studies of varying quality.
This lack of well-designed, evidence-based studies specifically related to green schools is understandable, because the concept of green schools is relatively new and evidence-based studies require a significant commitment of time and resources. Furthermore, other research on the factors that influence student learning does not typically look at factors related to building design and maintenance.
A much more robust body of scientific evidence is available that looks at building characteristics emphasized in green school design—envelope, mechanical and engineering systems, lighting, acoustics—and their relationships to occupant health, development, and productivity. Typically, such studies look at a single system or a very limited number of variables, and their quality varies. For example, a review of the literature on building characteristics, dampness, and health effects identified 590 epidemiological studies addressing these topics. Of these, only 61 met standards for strong study design and the provision of useful information (NORDAMP, 2002). Relatively few studies look at two or three building systems and their effects. With these caveats in mind, the committee determined it should review the scientific literature relating to those characteristics that are typically emphasized in definitions of and guidelines for green schools, such as indoor air quality, and building characteristics for which significant scientific research was available.
However, a review of all research literature that touches on some aspect of buildings and their potential impacts on occupant health, learning, or productivity is an undertaking beyond the resources of this study. Where rigorous reviews of a particular aspect of interest have been conducted by the National Research Council or other organizations or researchers, the committee relied on that work (e.g., IOM, 2004; Mendell and Heath, 2004). Where research is fairly limited but important to the study, the committee conducted its own review. Ultimately, the scope of the literature review was based on the committee’s collective judgment as to where its efforts should be concentrated to best address the task statement and meet the sponsors’ requirements. In all cases, the committee describes the source of the literature reviewed, the research methodology used, and the basis for the committee’s conclusions.
The committee’s examination of the evidence included assessing the relevance and validity of individual studies, integrating the overall evidence within and between diverse types of studies and across studies of different building characteristics, and then formulating its findings and recommendations. Because those tasks required thoughtful consideration of the evidence and could not be accomplished by adherence to a narrowly prescribed formula, the committee’s approach evolved throughout the
study process and was determined to some degree by the nature of the available evidence. Ultimately, the committee used its collective best judgment with regard to evaluating the published research and the plausibility of explanations of physiological mechanisms, and then integrated the results of those studies found to be useful (even if flawed) into its findings, conclusions, and recommendations.
In its deliberations, the committee determined that given the complexity of interactions between people and their environments, it may never be possible to categorically establish a causal relationship between an attribute of a school building and its effect on students, teachers, and staff. The effects of the built environment will necessarily appear to be small, given the large number of variables. Nor may it be possible to quantify the effects of one feature, such as acoustics, on student learning. However, the committee believes that empirical measures do not necessarily capture all relevant considerations that should be applied when evaluating research results. Qualitative aspects of the environment are also important. Thus, in the committee’s collective judgment, there is value in attempting to identify design features and building processes and practices for green schools that may lead to improvements in learning, health, and productivity for students, teachers, and support staff, even if the empirical results are less than robust.
FINDINGS AND RECOMMENDATION
Finding 2a: Given the complexity of interactions between people and their environments, establishing cause-and-effect relationships between an attribute of a green school or other building and its effect on people is very difficult. The effects of the built environment may appear to be small given the large number of variables and confounding factors involved.
Finding 2b: The committee did not identify any well-designed, evidence-based studies concerning the overall effects of green schools on human health, learning, or productivity or any evidence-based studies that analyze whether green schools are actually different from conventional schools in regard to these outcomes. This is understandable because the concept of green schools is relatively new, and evidence-based studies require a significant commitment of resources.
Finding 2c: Scientific research related to the effects of green schools on children and adults will be difficult to conduct until the physical characteristics that differentiate green from conventional schools are clearly specified. With outcomes as complex as student and teacher health, student learning, and teacher productivity influenced by many individual
family and community factors, it may be possible in theory to design research that controls for all potentially confounding factors, but difficult in practice to conduct such research.
Recommendation 2: The attributes of a green school that may potentially affect student and teacher health, student learning, and teacher productivity differently than those in conventional schools should be clearly specified. Once specified, it may be possible to design appropriate research studies to determine whether and how these attributes affect human health, learning, and productivity.