Does Deliberative Education Increase Civic Competence? Results from a Field Experiment

1,283 students in 59 classes across 25 upper-secondary schools, 36 teachers

Swedish upper-secondary students (16-17 years old) in the mandatory social science course, western and southern Sweden, 2015/16 academic year.

Cluster-randomized at class level, blocked and balanced within teacher, so the same teacher typically taught both a deliberative and a teacher-centred class - which removes teacher quality as a confound. Both arms received seven researcher-written lessons covering identical content (human rights; democratic decision-making) with identical reading, differing ONLY in whether students worked through 'deliberation scenario dilemmas' in small groups or reflected individually under an initiate-respond-evaluate teacher-led sequence. That is an ACTIVE-ALTERNATIVE control of unusual quality: content, texts and time-on-task are held constant. Manipulation check passed - treated students rated their classroom climate as significantly more open on the IEA CivEd items. Implementation check passed - over 90% of teachers reported being able to follow the instructions. Randomization checks on sex, immigrant origin and SES passed. Follow-up survey ~7 months later. Data, code and materials deposited on Harvard Dataverse.

Nothing moved. Immediately after seven lessons: political interest -0.012 (SE 0.012), democratic values -0.003 (0.008), political knowledge +0.068 (0.143) on a 0-7 index, political discussion +0.009 (0.009); n=661-1,092 across outcomes, 59 classes. At the end of the school year (~7 months later, 50 classes): interest -0.002 (0.017), values -0.000 (0.009), knowledge +0.307 (0.201), discussion +0.006 (0.013). Course grades were slightly higher in deliberation classrooms but not significantly. The authors single out the long-run knowledge estimate (0.307 of one extra correct answer out of seven) as 'of some substantial interest' despite being non-significant - which is a fair note, and also the only positive thing in the paper. They caution that a seven-lesson dose may be too small and that subgroup effects were not powered.

Low. Class-level randomization blocked within teacher; balance verified on sex, immigrant origin and books-at-home. This is a null from a design that would have detected a moderate effect.

This is an explicit, better-powered replication attempt of the small experimental literature on deliberative teaching (Bogaards and Deutsch 2015; Deslauriers et al. 2011; Latimer and Hempson 2012; Andersson 2015, six classes), and it fails. The authors say so: 'in contrast to previous research, we find little evidence that deliberative education significantly increases civic competence'. The prior positives came from one-day interventions or tiny samples. This also bears directly on the 'open classroom climate' observational literature (Campbell 2008; Torney-Purta; Kahne et al.), which is the most-repeated positive claim in civic education - the manipulation check confirms treated students perceived a significantly more open climate, and no civic outcome moved.