Revisión de los fundamentos del análisis de clases latentes y ejemplo de aplicación en el área de las adicciones

Abstract

Aims

Latent class analysis (LCA) is a technique especially relevant in the drug field, where much of the variables are categorical. The methodology on which the LCA is based allows identifying typologies of use of substances instead of to focus of exclusive way in pathological patterns, so it can detect other consumer problems. The objectives of this study are to review the basics of LCA and to provide an application example in area of addictions.

Method

Data of a study conducted by the Center of Sociological Research on consumer habits associated to tobacco in adults over 18 years old from Spain (N = 2002) were used for the exemplification of LCA. Of 513 people who reported being smokers, 498 completed sufficient data to be included in the LCA, 55.2 % men and 44.8 % women (M = 41.71 years old, SD = 13.75).

The analysis was based on the following variables: number of cigarettes smoked per day, intention to quit, self-perception of smoking and smoking frequency of parents as a child.

Results

In order to evaluate the fit of the model estimated by LCA were considered mainly the information criteria (BIC, AIC and AIC3). Preliminary results showed the models 2 and 4 classes as the most plausible. After bootstrapping conditional the model of 4 latent classes was chosen.

The classes recognized were: “smoking remarkable” = 27.91 % of the sample, “light smokers” = 26.21 %, “moderate smokers” = 23.75 %, and “serious smokers” = 22.13 %.

Conclusions

The results and usefulness of LCA to detect patterns of substance use are discussed.