Friendship Networks over the Life Course: A test of the Dyadic Withdrawal Hypothesis Using Survey Data on Couples

M. Kalmijn

Research output: Contribution to journalArticleScientificpeer-review

213 Citations (Scopus)
803 Downloads (Pure)


The dyadic withdrawal hypothesis argues that friendship networks become smaller when people enter a cohabiting relationship and that friendship networks become more overlapping with the partner during the course of the relationship. This hypothesis has received fragmented support in earlier research and has not been tested in The Netherlands. A nationally representative data set is analyzed which includes information on the five best friends of the respondent. A special feature of the data is that both partners were interviewed which allows us to check whether the friends reported by the respondent were also reported by the partner. In contrast to earlier studies, a broad set of life course stages is compared: single, dating, married (or cohabiting) without children, married with children, and the empty nest stage. Bivariate results and multilevel regression analyses indicate that friendship networks become smaller over the life course, although these changes primarily occur when people start dating and enter wedlock. Later changes are dominated by a simple age effect. For overlap, the models show that the percentage of shared friends and the number of joint contacts increase over the life course, both in a stepwise fashion and in a continuous fashion due to the aging process. In addition, shared friendship and joint contacts become more strongly related when people start living together. Implications for two underlying theoretical mechanisms are discussed, the competition principle and the balance principle.
Original languageEnglish
Pages (from-to)231-249
JournalSocial Networks
Issue number3
Publication statusPublished - 2003


Dive into the research topics of 'Friendship Networks over the Life Course: A test of the Dyadic Withdrawal Hypothesis Using Survey Data on Couples'. Together they form a unique fingerprint.

Cite this