The study, published in the Journal of Population Research by Guang-Xu Wang of National University of Tainan, tackles a deceptively simple question: when we say marriage is being deinstitutionalized, which beliefs are actually driving the shift, and are they all moving together? Drawing on three waves of the Taiwan Social Change Survey conducted in 2002, 2012, and 2022, Wang assembled a pooled sample of 5,783 respondents, with model-specific samples ranging from 3,697 to 3,896, to track how attitudes toward marriage differ across birth cohorts, age groups, and historical periods.

The concept of deinstitutionalization, famously articulated by sociologist Andrew Cherlin in 2004, describes the weakening of marriage as a social institution governed by clear, taken-for-granted norms. Rather than treating this erosion as a single phenomenon, Wang decomposed it into three distinct attitudinal dimensions: skepticism about whether marriage makes people happier, acceptance of having children outside marriage, and acceptance of unmarried couples living together. If the Second Demographic Transition operates as theory predicts, all three should show a steady cohort gradient, with younger generations progressively more permissive than their elders.

The results tell a more complicated story. Only one of the three dimensions displayed the clean, monotonic cohort pattern the theory would anticipate. Attitudes toward cohabitation showed a striking divide: cohorts born before the mid-1960s fell below the overall population mean in their acceptance of unmarried cohabitation, while those born after that point rose above it. This gradient proved remarkably robust, surviving every model specification Wang examined. In other words, no matter how the statistical machinery was configured, the generational split on cohabitation remained visible, suggesting that living together before or instead of marriage has become a genuine generational marker in Taiwan.

The other two dimensions did not follow suit. The cohort gradient for accepting childbearing outside marriage was noticeably weaker, and the pattern for skepticism toward marital happiness was weaker still. This is a crucial finding, because it means the belief that marriage is no longer a prerequisite for parenthood and the belief that marriage no longer guarantees happiness are not simply earlier or later stages of the same generational process. They appear to be stratified by cohort to markedly different degrees, implying that different components of marital normativity respond to different social forces.

Methodologically, the study relies on hierarchical age-period-cohort modeling, an approach designed to untangle three effects that are notoriously difficult to separate. Age effects capture how attitudes change as individuals grow older; period effects capture shifts that affect everyone at a given moment in history, such as economic shocks or legal reforms; and cohort effects capture the distinctive outlooks that people carry from the circumstances of their upbringing. Because age, period, and cohort are linearly dependent, researchers cannot fully disentangle them from any single survey, which is why Wang employed the hierarchical mixed-model framework developed by Yang and Land, treating cohort and period as random effects within a multilevel structure.

Two secondary findings stand out. First, age deviations from the cohort and period patterns were consistently small, meaning that attitudes toward marriage in Taiwan do not change dramatically as people age; the outlook people acquire early in life appears to persist. This aligns with the classic insight of demographer Norman Ryder, who argued in 1965 that cohorts, not individuals, are the primary vehicle of social change. Second, the analysis detected an upward period shift in 2022, suggesting that something about the most recent survey moment pushed acceptance of non-traditional family arrangements higher across all generations simultaneously, a pattern that may reflect broader cultural or pandemic-era shifts that the study does not pin down definitively.

The study also identified which individual characteristics travel most consistently with permissive family attitudes. Traditional gender-role attitudes and acceptance of matrilineal surnames, a marker of willingness to depart from patrilineal family tradition, emerged as the most reliable correlates across all three attitudinal dimensions. This connection is theoretically significant for East Asia, where family systems have historically been organized around patrilineal descent and male heirs. It suggests that the loosening of marriage norms is intertwined with a broader renegotiation of gender hierarchy within the family, echoing the gender revolution framework proposed by Frances Goldscheider and colleagues, which holds that family change accelerates when gender attitudes and institutions evolve together.

Why does this matter beyond Taiwan? East Asian societies have experienced some of the world’s most dramatic retreats from marriage, with delayed marriage and very low fertility documented from Japan to South Korea. Yet scholars have debated whether these changes represent the same Second Demographic Transition seen in the West or a distinct regional trajectory in which marriage remains normative even as its timing shifts. By showing that the attitudinal components of deinstitutionalization are not uniform, the study warns against measuring family change with a single index. A society might appear traditional on one dimension, such as disapproval of non-marital childbearing, while its youngest cohorts have already embraced cohabitation wholesale. Aggregating these beliefs would obscure precisely the generational dynamics that drive long-term change.

The implications extend to forecasting as well. If cohort replacement is the engine of attitudinal change, then the strong gradient on cohabitation suggests that acceptance of unmarried partnership will continue rising in Taiwan as older cohorts pass from the population and younger ones take their place. The weaker gradients on non-marital childbearing and marital skepticism, by contrast, imply that those beliefs may be more resistant to generational turnover, potentially reflecting the enduring institutional weight of family registration systems, social stigma, and the close linkage between marriage and parenthood in East Asian welfare arrangements. The study, supported by Taiwan’s National Science and Technology Council and based on publicly available survey data archived at Academia Sinica, thus offers both a caution and a tool: researchers who want to understand the future of marriage in East Asia should stop asking whether marriage is losing its hold as a single question, and start asking which of its many norms is slipping, for whom, and how fast.

Subject of Research: Cohort patterns in attitudinal deinstitutionalization of marriage in Taiwan

Article Title: Is marriage losing its hold? Differentiated cohort patterns in the attitudinal deinstitutionalization of marriage in Taiwan

Article References: Wang, G.-X. (2026). Is marriage losing its hold? Differentiated cohort patterns in the attitudinal deinstitutionalization of marriage in Taiwan. Journal of Population Research, 43(4), Article 31. https://doi.org/10.1007/s12546-026-09439-9

Image Credits: AI Generated

DOI: 10.1007/s12546-026-09439-9

Keywords: Second Demographic Transition, deinstitutionalization of marriage, age-period-cohort analysis, cohort replacement, cohabitation, non-marital childbearing, gender-role attitudes, Taiwan, family change, East Asia, marriage attitudes, Journal of Population Research