The use of good quality instruments is necessary when

The use of good quality instruments is necessary when examining associations between adolescents’ neighbourhood social environments and their health and well-being. Different approaches are taken to conceptualisation, operationalisation and measurement which might explain inconsistent research findings (Sampson et al., 2002). Reviews examining the social environment and similar health outcomes (i.e. alcohol use) have found conflicting results between studies (Bryden, Roberts, Petticrew, & McKee, 2013; Jackson, Denny, & Ameratunga, 2014) which may be due to considerable heterogeneity in how the neighbourhood social environment is measured. The neighbourhood social environment is often measured at different levels. The individual level represents the survey respondent\'s perception of their neighbourhood, while the neighbourhood level represents the combined characteristics of all survey respondents in that area. Ecological neighbourhood level measures are relevant to research of neighbourhoods and health so that the researcher can address health outcomes that vary across places, independent of the resident\'s individual level characteristics (Hawe & Shiell, 2000). Moreover, neighbourhood level exposures may be mediated by the corresponding individual level measure. As social processes occur at a neighbourhood level, measurement of ecological constructs represents a collective phenomenon; consequently, neighbourhood level measures are essential to better understand what makes some places more or less healthy and inform place-based interventions (Sampson et al., 2002). The specific objectives of the systematic review are as follows:
Methods
Results The search yielded a total of 13689 unique articles. Scanning these titles and abstracts yielded 683 articles that were further assessed for eligibility through full-text screening. Inter-rater agreement in the sample of 15% of titles and abstracts double-screened was 97% which suggested good agreement between the reviewers. Outstanding disagreements were resolved by the two reviewers through discussion. Upon screening the full-texts of the 683 articles, 205 met the inclusion criteria and were further assessed for quality using the COSMIN checklist. This led to exclusion of 651 articles. Thus a total of 32 studies (containing 56 unique measures) were rated as sufficient quality to include in the narrative synthesis (Fig. 1). Of the 32 studies, the majority were conducted in the Europe or North America (United States = 21). Only 2 studies were conducted in regions outside of Europe or North America. Approximately an equal number of studies were conducted in urban and mixed areas. Only one study was conducted in a solely rural environment. One paper used item response bmn 673 to examine reliability and structural validity; all others used classical test theory methods. Moreover, only five studies derived aggregate neighbourhood measures, with four of these using school as a proxy for residential neighbourhood. Reliability of aggregate neighbourhood measures was not addressed for most of these studies. General characteristics of the measures included in this review are presented in Table 2. Of the 56 social environment measures the minimum number of items was two and the maximum was 15. The minimum Cronbach\'s alpha was 0.45 and the maximum was 0.92. It has been suggested that an alpha between .70 and .90 is desirable, as an alpha that is too high may suggest that some items are redundant (Tavakol & Dennick, 2011); just over half of the 56 measures fell within this range. As shown in Table 2, concepts relating to sense of community belonging and neighbourhood support were the most prevalent.
Discussion The aim of this paper was to review measures of the adolescent neighbourhood social environment. One of our most stark findings was how many studies were identified as having poor quality reliability and/or validity reporting. This likely exacerbates confusion surrounding the concepts related to the neighbourhood social environment, both in research and in public policy. Having good quality measurement instruments is necessary for identifying associations between the neighbourhood social environment and adolescent health outcomes; lack of methodological uniformity is, therefore, likely to be a contributing factor to inconsistent findings. Despite the finding that many studies did not meet the quality cut-off, this review identified 56 measures of the neighbourhood social environment, where studies had sufficient quality in reporting. These measurement tools represent an encouraging basis in the field of measuring the neighbourhood social environment of adolescents. However, there is a need for further development or validation of existing measures outside of the US, particularly in non-westernised countries. Moreover, very few studies extend their measure to ecological areas, and those that do often use school as a proxy for neighbourhood. This is of concern, as the questions referred to the area in which adolescents live rather than area where they attend school. Consequently, these aggregate scales suffer from issues of face validity as adolescents may not live in the same area as where they go to school. Even fewer studies reported attempts to quantify the reliability and validity of ecological measures. This finding mirrors that of Sampson et al. (2002) and highlights that many studies addressing neighbourhood characteristics examine individual perceptions, but do not extend these measure to the neighbourhood level. This limits the informative power of these studies in terms of place-based interventions. Only one study utilised item response theory techniques; these techniques are useful for non-linear items and can be extended to neighbourhood level measures and are therefore of use in future studies (Matsueda & Drakulich, 2016).