We have collected and organized elements of a comparative statistical dataset on legal recognition of same-sex couples through marriage and various forms of registered partnership. The objective is to determine the frequency of the use that same-sex couples have been making of the possibilities for legal recognition that European countries have introduced for them. The database will be open to international comparisons in order to test whether legal contexts are factors of differentiation in behaviors regarding access to marriage or registered partnership. Whenever possible, separation of registered partners or divorce of married couples has also been documented statistically. In all cases, statistics have been broken down by gender, in order to measure behaviors in gay and lesbian couples. The material collected also considers the possibility of comparing same-sex couples with different-sex couples.

The basic element collected is the annual number of partnership registrations or marriages of gay and lesbian couples. The statistics are generally made available by national statistical institutes as a component of population change, together with births, deaths and different-sex marriages. But some difficulties may arise. They have been listed and analyzed in Cortina & Festy (2014b). For the sake of comparison, we have also recorded numbers of different-sex marriages in the countries that offer legal recognition to same-sex couples. For France and the Netherlands, we have also added information on different-sex couples choosing for Pacs unions and registered partnerships, respectively, since these are open to different-sex as well as same-sex couples.

The simplest indicator of frequency in the dataset  is a crude rate: ratio of marriages or registered partnerships to total population. The number of male marriages or registered partnerships is divided by midyear total male population; that of female marriages or registered partnerships by midyear total female population. Similarly, the number of different-sex marriages or registered partnerships is divided by midyear total population, males and females.

A more meaningful indicator (which is also available in the dataset) is obtained by dividing the number of male marriages or registered partnerships by the number of male couples still unmarried or unregistered. This ratio – and its female equivalent – relates events to the population most directly “exposed to risk”, a standard procedure in demography. The main difficulty is to obtain estimates of this population. We explained this in Cortina & Festy (2014a). The symmetrical indicator for different-sex couples is much easier to calculate since statistics of unmarried different-sex couples are routinely published in many countries, since the 1990s or even earlier.

Crude rates are imperfect indicators of frequencies, but they may be taken as acceptable proxies if the objective is to evidence time trends or international comparisons. They are meaningless if the objective is to compare same-sex couples to different-sex couples, which is only possible with refined rates.