Contextual resources are important for the positive development of children and youth. Growing up in socio-economically vulnerable neighborhoods has been associated with, for instance, children´s health outcomes (Brzoska & Razum, 2015), anti-social behavior (Odgers et al., 2009) and school achievement (Leventhal & Brooks-Gunn, 2000). However, the role of physical neighborhood characteristics, like safety, orderliness and well-kept buildings, and their importance to child development is understudied (Christian et al., 2015; Villanueva et al., 2016).
Assessment of the physical environment has been made through, often subjective, in-person observation and coding of contextual conditions (Clarke et al., 2010; Mooney et al., 2014). With support of recent digital geographical tools such as Google Street View (GSV), Systematic Social Observation (SSO) can to a large extent be performed from a distance and with fewer resources than in person observations. Virtual SSO is considered to be a reliable and cost-effective method to study neighborhood conditions (Bader et al., 2017; Brunton-Smith, 2018; Odgers et al., 2012), and has previously been used in the United States and United Kingdom. However, this method has not yet been used in Sweden - a context which may be expected to differ from those previously studied using SSO.
Sweden is a social welfare state, with a social and political fabric that aims to support the rights of children and families to thrive under the best living conditions that are practically possible. Yet, there are variations in contextual factors within contemporary Swedish neighborhoods that are not well studied nor understood. Given the need to understand how contextual conditions relate to children’s learning and development, e.g., their socio-emotional competence, new tools that can measure contextual conditions, and that are culturally relevant and rooted in Swedish neighborhoods are needed. Thus, the study aim was to determine if virtual SSO is a reliable and valid method that may provide assessment of meaningful neighborhood contextual factors that are relevant and reflective of life in a Swedish context.
In study 1, two raters performed in-person and virtual data collection with the same assessment protocol, in the same 24 neighborhoods within four postal code areas. The research question was to establish if the in-person and GSV could be reliable indicators across raters, and if the GSV and in-person data collection were comparable in this sample of Swedish suburban neighborhoods. On an item level, we analyzed inter-rater reliability with Fleiss Kappa, Intra Class Correlation (ICC) and percentage of agreement, and used Pearson correlations to estimate concurrent validity across methods. Results showed high consistency between raters (on in person and GSV items) and high consistency across methods, on the included items. GSV was thus regarded as a method that was comparable with in-person data collection, and possible to use as an index of neighborhood conditions in this Swedish context. Thereafter, at the scale level, we developed virtual SSO scales for neighborhood contextual conditions. Scales for Physical Decay, Neighborhood Dangerousness and Physical Disorder, proved to be reliable, with high consistency across methods, and high internal consistency (Cronbach’s alpha). In study 2, we wanted to establish if the virtual SSO measures developed in Study 1, could be linked to levels of income in the same Swedish neighborhoods that were rated using the GSV method and scales. Virtual data collection using GSV was performed by two raters, for a total of 137 neighborhoods (the 24 neighborhoods in Study 1 were included) in 22 postal code areas. We estimated internal consistency (Cronbach’s alpha) for virtual SSO measures in this larger sample, and results showed that scales for Physical Decay and Neighborhood Dangerousness had high internal consistency. Concurrent validity was estimated through correlation between virtual SSO ratings of neighborhoods aggregated to postal code level, and level of household income of all residents at postal code level. An independent t-test revealed that mean values of scales for observed Physical Decay, Neighborhood Dangerousness and a single item measuring signs of garbage or litter in the streets were significantly higher in low-income areas than in high-income areas. We conclude that virtual SSO with GSV is a reliable and valid measure of several key objective neighborhood contextual conditions and that the GSV scales distinguish between postal code areas that have residents with high or low income.
Neighborhood contextual conditions have been difficult to measure objectively and reliably, and the nature of measures have limited the ability to capture the actual contextual conditions in children’s daily life. Virtual SSO is an observational tool that offers a possibility to objectively assess not only neighborhood physical conditions, but also provide information about neighborhood assets and contextual factors critical for academic achievement as well as socio-emotional development. This could, in turn, contribute to a deeper understanding of how Specific contextual factors influence children´s potential for learning and development, especially for children in most need.