Comparison of two approaches in multichannel sequence analysis using the Swiss Household Panel

Abstract

Sequence analysis is an established approach to study life courses. When several life domains are considered simultaneously, multichannel sequence analysis (MSA) and the extended alphabet (EA) approach are the most frequently used strategies. We compare these two methods using real data composed of four life domains (cohabitational status, children, professional status, health), and we focus on clustering since sequence analysis usually aims to identify typical patterns in sequences. As professional status trajectories, and potentially their relationship with other domains, proved to be different between men and women, the analyses were run separately by sex. We describe step by step the approach followed and the different criteria to judge the relevance of a typology. Neither of the two approaches is clearly superior, and the typologies obtained with both methods are often close. However, even if MSA is generally easier to use and applies to a broader range of situations, EA can provide original typologies in specific cases and we therefore propose guidelines for choosing between the two approaches depending on the context.