Reproducibility of indoor environment characteristics obtained in a walk through questionnaire. A pilot study

Walk through questionnaires may be feasible tools to obtain data on the indoor environment in community studies. However, limited information is available regarding the reproducibility of the data obtained through these questionnaires. In this pilot study, two construction engineering students inspected ten dwellings twice by means of a standardized walk through questionnaire. The two engineering students inspected the dwellings independent of each other within two months. Time between the two visits varied between 14 and 40 days. The variables presented were based on information from the residents and inspection. The continuous variables recorded included number of rooms in the dwellings, room volume, length of filled shelves and textile area. The inter- and intra-observed variabilities were poorer for number of rooms and shelf factor than for the other variables. For the 3 of the 9 categorical variables based on inspection, the inter-observer comparison showed complete agreement with a kappa statistics of 1.0, these variables being condition of the window frames and construction of outside walls and roof. One of the categorical variables showed a kappa statistics of < 0.5, these variables being presence of basement and presence of condensation at windows. This study indicates a wide variation in observer variability between various items of a walk through questionnaire. Clear definitions of all the parts of the questionnaire are needed, as well as thorough training of observers.