Fig. 1: Item-response probabilities estimated from a latent class analysis model with 19 personal care products and 6 latent classes, SELF (2013–2018), N = 1562.

The frequency distribution of each product for the entire cohort is shown in the last row (Overall). Latent class labels were assigned based on comparing probabilities of usage frequency categories in each latent class to probabilities of usage frequency categories in the overall sample. Latent classes: (1) Lower Overall (18%): lower/less frequent use of products across all categories; (2) Higher Nailcare (12%): lower/less frequent use of most products but higher/more frequent use of nail products; (3) Higher Skincare (16%): lower/less frequent use of most products but higher/more frequent use of body and hand creams; (4) Moderate Overall (25%): class closest to the cohort average; (5) Higher Makeup/Haircare/Skincare (16%): higher/more frequent use of makeup, skincare, and haircare products but lower/less frequent use of nailcare products; (6) Higher Overall (13%): similar to the Higher Makeup/Haircare/Skincare class but with additional higher/more frequent use of nailcare products.