By: Cesira Urzì Brancati

In the first blog, we investigated sexual desire and sexual activities among English people aged 65+ and revealed how, even though they both decreased with age, they still matter for a significant proportion of older people.

In the second blog of the series, we investigate another burning subject: love and marriages in later life. In other words, we try to answer the following questions: how many people aged 50+ are likely to remarry after experiencing divorce/widowhood? Do people’s chances of finding love decrease with age? What differentiates people who remarry? And finally, does sexual desire play a role?

To this end, we carried out some empirical analyses using 10 years’ worth of data from the English Longitudinal Study of Ageing (ELSA) 2002-2012.

Between 2002 and 2012, approximately 1 in 10 English people aged 50+ classify themselves as divorced, while about 1 in 6 as a widow/er. Among them, a relatively low proportion, approximately 1 in 20, managed to find a new partner during the period considered, with the numbers increasing over time (see figure 1) for a total of approximately 185,000 new relationships in 2012.

Figure 1: Divorcées/Separated or Widow(er)s finding a new partner in later life

Source: Own elaborations from ELSA 2002-2012

Unsurprisingly, finding a new partner is easier for divorcées than for widows and widowers, who may still be grieving for the loss of their partner; however, between 2002 and 2012, nearly 50,000 widows and widowers remarried (or started cohabiting) and among them, at least half were aged 65+.

How does the likelihood of finding a new partner changes with age?

In Figure 2 we plot the relationship between ageing and the likelihood of finding a new partner for men and women. The results are somewhat surprising and reveal a profound gender gap.

In general, finding a new partner appears to be harder for women than men at any age, but it is next to impossible for women past the age of 75. The likelihood of starting a new relationship is even lower for widows and widowers, and it drastically drops with age, even after taking into account any potential health factors.

Figure 2: How is ageing associated with the probability of finding a new partner?

Source: Own elaborations from ELSA 2002-2012

Who remarries?

We then turn to analyse the socio-demographic characteristics of divorced/widowed people aged 50+ who manage to find a new partner[1], and discover not only that women are less likely to find a new partner than men, but also that healthy people are more likely to find a new partner than those in bad health. Being highly educated, i.e. having a degree, only has a positive impact for widows, while it has no association with the probability that a divorcée remarries.

Quite interestingly, high sexual desire (i.e. claiming to think about sex more than once a week) only seems to have an impact on widows, while it does not affect the likelihood that divorcée finds a new partner.[2]

How can we explain the gender gap in romantic success?

Many reasons can explain the different success rate between men and women in finding a new partner, and the simplest one is probably that the number of “suitable” men – i.e. men who are not part of a couple – is much smaller. Indeed, nearly two thirds of divorcées and three quarters of widowed aged 50+ are women, which leaves men with a much larger sample to choose from.

And yet, only a tiny proportion of unmarried older people begins a new relationship, and, from our previous blog, we learned that at least half of them still has sexual desire, so presumably would benefit from finding a partner.

If that were the case, then initiatives aimed at increasing awareness about dating later in life would be surely welcome.

[1] Results from two probit regressions where the dependent variable is the probability of a transition from divorced/widow to being part of a couple and the independent variables are gender (dummy for female), age, health and education.

[2] Estimates on the impact of sexual desire are not very robust, since the sample size is extremely small.