Does Daily Meditation Increase Self-Compassion? A Bayesian Multilevel Diary Study

Evaluations of brief self-compassion programs usually compare phases, leaving open whether meditation and nonmeditation days differ. With few nonmeditation days, subtracting each person's observed average can distort the separation of day-level from person-level effects. This secondary analysis used the archival 28-day diary of 65 employees in a two-week online selfcompassion meditation program, with morning reports of meditation over the previous 24 hours (Li et al., 2021). Across 14 intervention days, a prespecified Bayesian two-level model treated each person's meditation propensity as latent, separating the within-person contrast between meditation and nonmeditation days from stable differences in meditation frequency. Hypotheses were higher same-day self-compassion on meditation days (primary) and lower next-day emotional exhaustion, adjusting for same-day exhaustion (secondary). Results are posterior medians, 95% credible intervals (CrIs), and directional probabilities. For same-day selfcompassion, the meditation-day slope had median = 0.00 SD, 95% CrI [−0.09, 0.09], P(b > 0) = .48; for next-day exhaustion, the slope had median = −0.03 SD, 95% CrI [−0.16, 0.11], P(b < 0) = .65. Both CrIs spanned zero. An archival intervention diary can answer an adherence question its original phase-based analysis did not ask.