|dc.description.abstract||People’s mental state can be affected by their perception of their surroundings, either propagating in other health outcomes or enhancing well-being. This perspective has driven a new discussion about the importance of research on perceived environments, particularly green spaces. The standard method in this field is the use of online surveys, gathering information on people’s perceptions based on Google Street View or the Flicker database. This approach, however, does not question if these assessments limited by the online environment are an accurate representation of reality, with some elements of the actual environment not being carried over. Here, we address whether people’s perceptions differ when comparing their answers from an online survey to answering the same questions in a real-life setting.
The study consisted of three typologically different locations in the city of Utrecht, the Netherlands. In total, 22 participants completed the online survey and subsequently did the assessment in the real-life setting under a researcher’s supervision. The survey was based on 360-degree pictures of these locations with 20 variables to be compared. By recruiting another group of 170 and 228 participants for the online survey, it was possible to examine how different people perceive the same pictures. The pictures were analysed using a deep learning algorithm for object recognition. These data were compared to the participants’ perceptions of natural-like and urban environments assessed in the real-life setting and online, resulting in calculating percentage agreements.
Six variables (Social Connection, Entertainment, Uniqueness, Colours, Cared For, Water) showed a significant (p<0.05) difference between the two settings. The percentage agreements exceeded consistently 50% and were as high as 88% for ‘urban elements’ online. Several perceptual variables developed links with the most prevalent feelings, i.e. a clustering effect of these variables, although not all held against the data from the real-life setting. This pilot study was the first attempt to examine the differences between the real-life setting and online surveys. Discovered differences do not reject using online surveys to study perceived environments, as they were less critical in the overall result interpretation. The main limitations seemed to be the temporal discrepancy of the locations captured by a camera and the assessment in the real-life setting, and the small size of the pictures in the survey layout combined with the lack of other stimuli. These limitations may be resolved by design improvements suggested in this study, yet more robust experiments are needed.||