Interesting statistics about divorce rates

Today a conversation was struck up regarding divorces and it got me thinking, how do people actually calculate divorce rates? Does the "almost 50% of marriages end in divorce" argument actually hold up? Or was it just a faulty statistic that was taken way out of proportion? What I came across was the complexity of calculating divorce rates and where this "almost 50%" statistic comes from. We can start with the simplest method and the problems it causes and how these problems are overcome.

Crude Divorce Rate

This is the number of divorces per 1,000 persons. From the onset, you can already see the problems that this might cause, namely; the number of persons includes everybody, including children and singles who are not at risk of divorce. Although it is some measure of divorce rates, it doesn't really give you much because the denominator is not standardized.

Example

Imagine 2 states, both with a population of 1,000, but State 1 has 200 married persons (100 married couples) whilst State 2 has 500 (250 married couples). Both states recorded 50 divorces last year. Under the crude divorce rate, both states would technically be considered to have the same divorce rate, regardless of the composition of family structure. This is not accurate. Then, 15 years later, as State 2 ends up with more children given the higher proportion of married couples, the composition of divorces, again, changes (and possibly switches) between the two states. Hence, anything that changes the composition of the family structure will distort the crude divorce rate.

Refined Divorce Rate

This is the number of divorces per 1,000 married women. This improves on the crude divorce rate because it only includes married persons, hence taking into account the family structure. This definitely helps answer some questions, but does not answer the question of how many marriages actually end in divorce? The refined divorce rate measures divorce at a certain period in time and does not consider the proportion of individuals who have been married a certain amount of time. (i.e. what if people married for 10 years are less likely to divorce than people married for 5 years?) This, again, would still be influenced by the age structure of a given population (i.e. younger people, on average, would be married for less years than older people, given that they are still married). Hence, again, this measure would not necessarily be effective in calculating the proportion of marriages that would actually end in divorce.

Cohort Divorce Rate

The cohort divorce rate considers a cohort and calculates the proportion of marriages that end in divorce (i.e. how many people who married in 1980 are still married 40 years later, in 2020). This gives you a divorce rate of a cohort, accounting, to an extent, the time factor of marriage. The problem with this, however, is that it is not forward-looking. Just because certain people are not divorced now does not mean they won't be divorced in the future. This is especially considering the lack of data of younger cohorts (i.e. there is not enough data for people who have only been married for one year, as they may be divorced in year 2, or year 3, or year 50, or year 60).

Life Table Method

To account for the limitation of the cohort divorce rate, we finally come to the life table method. This takes into account the proportion of individuals and how long they have been married. Basically, they consider the proportion of couples who have been married for one year ended up divorced in the last year, then the proportion of couples married for 2 years, and how many of them divorced in the last year, then the proportion of couples married for 3 years and so on and so forth. After this, the final number is aggregated to finally determine a 'duration specific' divorce rate  (i.e. how long a couple was married and how many divorced within that time). This measure, however, assumes that the proportion of divorces in each cohort remains the same over time, but still provides a decent estimate of the current divorce rate.

Now, taking all of this into consideration and using a refined life table method, does the 'almost 50% of marriages end in divorce' stack up?

Yes. Yes it does.

NB

It's important to note here that the divorce rate under the life table method is NOT the number of divorces divided by the number of marriages. 2 couples getting married and 1 couple (already married) getting divorced is NOT the same as 1 couple getting divorced of the 2 couples currently married.