BUILDING SURVEYS | BEST PRACTICE MEASURE OF SUCCESS So you want to conduct an occupant survey in a nondomestic building? It requires more thought and skill than you might think, and good design may prevent you jumping to the wrong conclusions, says Roderic Bunn Scores for control ventilation by window location Response count: non-window 15 Sample: 42 Mean score: 2.79 Median: 2.50 Variance: 2.56 10 5 0 1 2 3 4 Response count: window 15 5 6 7 Sample: 43 Mean score: 3.65 Median: 4.00 Variance: 2.85 10 5 0 1 2 3 4 5 6 7 Control of ventilation 1: No control, 7: Full control Figure 1: A good-practice example of how to show data in stacked histogram format to illustrate differences in ventilation control perceptions between window and non-window seats. All statistics are given for total clarity. Note that spatial differences can lead to skewed or at (non-normal) distributions. This is not a failing, rather a normal characteristic of occupant perceptions. An understanding of what the data is saying requires the researcher to have a full grasp of the buildings physical context P eople are a problem. The humourist Guy Browning wrote that most problems in life are people problems, and most people problems are communication problems. If you want to solve a communication problem, he said, go give someone a damn good listening to. The key to understanding what people experience in buildings is to ask them clear, unbiased questions. The more abstract, enigmatic, or technical a question, the greater the variation in the answers. This is compounded by the fact that humans are emotional, diverse and perverse. This makes accurate surveying of building occupants tricky. Hence the market for environmental instrumentation. Reliable, low-cost, and accurate digital devices can measure most environmental parameters. Their drawback is an inability to record conditions as perceived by humans. Nor can they rank trade-offs that humans make, such as between ventilation and external noise, or ventilation and draught from an open window. Ultimately, it is human perception of conditions that determines comfort and motivates behaviour, such as opening or closing a window, not an instrumented value. Like an ill-fitting suit, instrumentation and perception data tend to fit only where they touch. Its why proof of causal links between measured environmental parameters and productivity remains elusive. Anyone can design a survey, but the difference between a bad questionnaire and a good one is the difference between a novice and a virtuoso. Most people can bash out Chopsticks on a piano, but performing a concerto is another level. When it comes to creating a good survey, there is simply no substitute for formal study and experience. Coversely, a badly designed survey risks leading you to jump to the wrong conclusions. Creating a questionnaire Rule One: read a book on how to do it. There are many guides on designing social surveys. Rule Two: Be clear what your survey is about. Do you want it to be specific or cover a wide range of comfort factors? The best surveys are those that have a particular research focus, be it thermal comfort, acoustics or health (see Figure 2 A map of typical occupant surveys). Its possible to develop a general built environment questionnaire, but it takes talent to avoid breaking Rule Three, which is: keep it short. Respondents start to suffer survey fatigue after three A4 pages or web screens. Rule Four: Test your survey before unleashing it on the world. Practise it with volunteers and ask them what they thought the questions were about. Only a couple of people in any one survey might express misgivings about a particular question youve posed, which is why it takes 10 or 20 surveys before a consistent pattern of confusion emerges. Once realisation dawns, youll 36 February 2019 www.cibsejournal.com CIBSE Feb19 pp36-38 Occupant survey JR.indd 36 28/01/2019 09:30