LCA analysis
To identify the best-fitting model for LCA (i.e. between 1-5 models), we chose the minimum adjusted Bayesian Information Criteria (aBIC)25,26 as this is known to be more robust with smaller datasets (<400 participants) than Bayesian Information Criterion (BIC)26,27 and Akaike’s Information Criterion (AIC)26,28. Entropy R2 was used as the second criterion to determine the separation between the models using the highest value between models29.