Measures

The key outcome of interest is whether engagement leads to affective interethnic ties; a key indicator of positive interethnic contact (Lolliot et al. 2015), and a well-established strong predictor of prejudice-reduction (Pettigrew 1998). This follows the proposed model that engagement should trigger increased opportunities for optimal interethnic contact with strong “acquaintanceship potential”, leading to the formation of interethnic ties. Affective interethnic ties are captured using the following question: “Now, think about people you know who you would feel happy getting in touch with to ask for advice or a favor. How many are from a different race or ethnicity to you?” Responses on a 4-point Likert scale included (0) “none of them”, (1) “hardly any of them”, (2) “some of them”, and (3) “many of them”. This question is designed to capture social ties with a strong, affective component, which involve the provision of support, confiding/disclosure and reciprocity (Pettigrew 1998).

As discussed, the scheme is designed so that the composition of participants aligns with the composition of the youth population in the wider area (Local Authority) which young people participate in. Therefore, all contextual-level variables are measured at the Local Authority-level. A multi-group measure of segregation is applied to better capture the complexity of residential segregation among groups in an area, using the multi-group entropy index (H Index) (Iceland 2004). Segregation is measured between five ethnic groups: White, Black, Asian, Mixed and Other⁴. The Output-Area (mean n = 300 people) forms the lower area unit nested within the Local Authority. The H-Index captures how (un)evenly ethnic groups are distributed across all the Output Areas that compose a Local Authority. Values range from 0 (perfectly integrated) to 1 (perfectly segregated):

where t_i refers to the total population of OA i. T is the Local Authority area population. n is the number of OAs. E_i represent OA i’s diversity (entropy) and E represents Local Authority area diversity (entropy) (see below for calculation of entropy scores).

The H-Index falls into the class of segregation indices tapping the (un)eveness with which ethnic groups are distributed across an area (Massey and Denton 1988). The calculation of such indices are largely unaffected by the size of groups in an area⁵, instead capturing the relative distribution of those groups that are present across the area (Laurence et al. 2019b). This results in a greater spread of different levels of segregation across different levels of diversity (the two are correlated at r = 0.1). This provides opportunities to test whether, after controlling for the size of groups in an area, the unevenness with which those groups are distributed across it affects the impact of participation.

To measure the level of ethnic diversity in a Local Authority the Entropy Score is used for harmonization with the H Index. The Entropy score captures how far the ethnic composition of an area departs from perfect homogeneity (correlating with other measures of diversity e.g. Simpson’s: r = 0.99). A maximum score is when all groups have equal representation in the area:

where Π_r refers to a particular ethnic group’s proportion of the whole Local Authority population. Logarithmic calculations use the natural log (Iceland 2004).

Complete data on the ethnic composition of the teams in which young people participate is not available to link into the data. However, as discussed, the scheme aims to make the ethnic composition of participants within any given Local Authority representative of the ethnic composition of the youth Local Authority population. Best available estimates put the correlation between participants’ team ethnic diversity and Local Authority ethnic diversity at r = 0.76⁶. Local Authority ethnic diversity is therefore also used as a proxy for participants’ opportunities for interethnic contact on the scheme.