From a crisis of results to a crisis of wellbeing – education reform and the declining sense of school belonging in Sweden

Figures & data

Figure 1. Changes in school belonging 2003–2015 across EU and OECD countries. Data from PISA (waves 2000, 2003, 2012, 2015, 2018). Only countries with complete data on school belonging from all survey waves (2000–2018) included.

Figure 2. Trends in school belonging 2000–2018. Data from PISA (waves 2000, 2003, 2012, 2015, 2018). Only countries with complete data on school belonging from all survey waves (2000–2018) included.

Figure 3. School belonging in Sweden compared to synthetic Sweden. Data from PISA (waves 2000, 2003, 2012, 2015, 2018). Only countries with complete data on school belonging from all survey waves (2000–2018) included.

Figure 4. Gap in school belonging between each country and its synthetic counterpart. Data from PISA (waves 2000, 2003, 2012, 2015, 2018). Only countries with complete data on school belonging from all survey waves (2000–2018) included.

Table 1. Two-way fixed effects regression models with school belonging as the outcome.

	Column 1	Column 2	Column 3	Column 4	Column 5	Column 6
Average effects
Treatment: Swedish education reform	−0.171* (0.072)	−0.165*** (0.017)
Subgroup effects
Reading achievement: top 90%			−0.168 (0.073)
Reading achievement: bottom 10%			−0.210 (0.081)
Difference			−0.042
Mathematics achievement: top 90%				−0.219 (0.050)
Mathematics achievement: bottom 10%				−0.355 (0.068)
Difference				−0.136**
Parental class: top 90%					−0.159 (0.072)
Parental class: bottom 10%					−0.262 (0.073)
Difference					−0.103***
Swedish-born						−0.166 (0.072)
Foreign-born						−0.254 (0.076)
Difference						−0.088***
N pupils	857094	460509	857094	735203	857094	857094
N country	40	40	40	40	40	40

*p-value < 0.05.

**p-value < 0.01.

***p-value < 0.001.

Data from PISA (2000, 2003, 2012, 2015, 2018). All OECD/EU countries with data on school belonging from at least one survey wave included. Table shows results of linear regression models with dummy variables for countries and year, and controlled for gender, age, achievement in reading and mathematics, parental socio-economic class, migration background, disciplinary climate, GDP, country educational attainment, social expenditure and unemployment rate. Cluster robust standard errors in parentheses.