Communication with family and friends across the life course

Each stage of the human life course is characterized by a distinctive pattern of social relations. We study how the intensity and importance of the closest social contacts vary across the life course, using a large database of mobile communication from a European country. We first determine the most likely social relationship type from these mobile phone records by relating the age and gender of the caller and recipient to the frequency, length, and direction of calls. We then show how communication patterns between parents and children, romantic partner, and friends vary across the six main stages of the adult family life course. Young adulthood is dominated by a gradual shift of call activity from parents to close friends, and then to a romantic partner, culminating in the period of early family formation during which the focus is on the romantic partner. During middle adulthood call patterns suggest a high dependence on the parents of the ego, who, presumably often provide alloparental care, while at this stage female same-gender friendship also peaks. During post-reproductive adulthood, individuals and especially women balance close social contacts among three generations. The age of grandparenthood brings the children entering adulthood and family formation into the focus, and is associated with a realignment of close social contacts especially among women, while the old age is dominated by dependence on their children.

💡 Research Summary

This paper investigates how the intensity and importance of an individual’s closest social contacts evolve across the adult life course, using a massive mobile‑call dataset from a European country. The authors first infer the most probable relationship type for each call (parent–child, romantic partner, same‑gender friend, opposite‑gender friend) by combining the ages and genders of caller and recipient with call‑frequency, duration, and directionality metrics. A multivariate logistic‑regression model, supplemented by random‑forest classification, achieves an accuracy of roughly 80 % when validated against a small survey sample, demonstrating reliable relationship labeling at scale.

With relationships assigned, the adult population is divided into six canonical stages: (1) young adulthood (18‑24 y), (2) early family formation (25‑34 y), (3) middle adulthood (35‑49 y), (4) pre‑/post‑menopause (50‑64 y), (5) grandparenthood (65‑74 y), and (6) old age (75 y+). For each stage the authors compute the proportion of total call minutes devoted to each relationship type, stratify by gender, and test differences with ANOVA and post‑hoc Tukey tests (all p < 0.001).

Key findings:

1. In young adulthood, calls to parents steadily decline while same‑gender friend calls rise, reflecting a shift from family‑centric to peer‑centric networks.
2. During early family formation, communication with a romantic partner peaks (≈55 % of total minutes) and parental contact reaches a minimum, indicating the emergence of a nuclear‑family core.
3. Middle adulthood shows a resurgence of parental contact (≈35 % of minutes), especially among women, who also exhibit the highest same‑gender friend interaction. The authors interpret this as evidence of alloparental care, where adult children rely on parents for support while parents may assist with grandchildren.
4. In the pre‑/post‑menopausal and grandparent phases, women balance contacts across three generations—children, parents, and grandchildren—maintaining high call volumes with each. Men’s communication remains more child‑focused.
5. Old age is dominated by child‑centric communication (≈70 % of minutes), confirming a dependency reversal where adult children become the primary source of social and instrumental support.

The study’s strengths lie in its use of unobtrusive, high‑resolution digital trace data and its rigorous statistical treatment of relationship inference and life‑stage comparisons. Limitations include the exclusive reliance on voice calls (omitting SMS, social‑media, and video calls), the cultural specificity of a single European nation, and the probabilistic nature of relationship labeling, which may miss nuanced or multiplex ties.

Overall, the paper provides a quantitative, life‑course perspective on how human social networks re‑configure with age and gender, offering valuable insights for demographers, sociologists, and policymakers interested in inter‑generational support systems and the design of age‑sensitive communication services.