Are consensual unions more gender equal than marriages?

1. Are consensual unions more gender equal than marriages? By A. Pinnelli, F. Fiori and A. Testini Department of Demography University of Rome La Sapienza The study is part of the project “Family and health between gender system and institutions”, coordinated by Antonella Pinnelli and partially financed by the Ministry of University and Scientific Research Canterbury 2005- BSPS Annual Conference

2. Aim of the study: versus CONSENSUAL UNIONS MARRIAGES We intend to compare the two typologies of union, according to various characteristics describing the gender contract into force within the couple. The hypothesis is that consensual unions are characterised by less asymmetrical gender roles “Are consensual unions more gender equal than marriages?,” by Pinnelli, Fiori, Testini

3. WARNING! The work is still in progress What we will be illustrating are only results from first analyses carried out to test the validity of our hypotheses Main intent is therefore to obtain suggestions and advices to improve our work and to further develop the study, both from a theoretical and a statistical point of view. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

4. Some considerations One of the major demographic change investing Europe since the Seventies has been the increase in the proportion of women and men living together as a couple without formalising their union with marriage. Both if considered as a real alternative to marriage or merely as a trial period, prelude of a formalised union, informal cohabitations are often associated to changes in the status of women and to their gained independence and they seem to be an option particularly attractive for people sharing more liberal and gender-egalitarian attitudes. Many studies demonstrated that those who choose to cohabit are, on average, more liberal, less religious and more favourable to more egalitarian relations and to less traditional family roles. What is new in our work, compared to previous studies on differences between cohabitations and marriage, is the explicit accent on gender differences and the attempt to take into account all the dimensions of gender inequality within the couple. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

5. Hypotheses For a generic couple, a set of characteristics pointing at a more egalitarian gender system may increase the probability of being a consensual union, while when those characteristics point at a more unfair (or less favourable to women) gender balance, then the probability of being a married couple is higher. Specifically, we consider the gender inequality between partners to be made of: Gender differences in age, education and income; Gender division of roles (work, household tasks, care duties) Gender differences in the amount of available free time and in the frequency of social contacts. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

6. Data The study is carried out on data coming from the ECHP, wave 2001. Information was taken mainly from personal file. Using the relationship file we managed to individuate partners and to reconstruct the couples, which we used as statistical units of analysis. We excluded from the sample couples with man aged more than 55 or woman aged more than 50. According to what suggested by ECHP user manual, couples were weighted using household weights, corrected by an inflation factor in the case of multi-country analyses (divided for 1000 to obtain comparable sample size for multi and single country analyses) “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

7. Countries • We consider first Europe as a whole (15 countries, 47869 couples, of which 13.6 cohabiting) • Then, we grouped the 15 countries into 3 macro regions, following a clusterization obtained by previous studies (Pinnelli and Di Giulio 2003) according to levels of gender equality and the diffusion of consensual unions: Northern Europe: Denmark, Sweden and Finland (2162 couples, of which 41.8 % cohabiting) Southern Europe: Italy, Greece, Spain and Portugal (15044 couples, of which 3.6% cohabiting) Western Europe: Germany,The Netherlands, Belgium, United Kingdom, Ireland, Luxembourg, France, Austria (30663 couples, of which 16.5 cohabiting) • Finally we focus on Italy (7123 of which 3.1 % cohabiting) and Great Britain (6729 couples of which 21.1 cohabiting) “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

8. Variables The choice of the variables directly comes from the hypotheses we aim to test We try to include in the models at least one variable for each single aspect of gender inequality Moreover we add some “control” variables to take into account the influences of context, time, and social class. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

9. Control Variables • MACROREGIONS: to take into account the influence of the context surrounding individual behaviours, we utilised the subdivision of Europe into Northern, Southern and Western Europe, and for Italy the division into Centre-North and South and Island. • WOMAN AGE: we include woman age (in class) to account for the temporal dimension of the phenomenon. The diffusion of cohabitation is relatively recent in many countries, where we hypothesize it to be more widespread in the younger age class • WOMAN EDUCATIONAL LEVEL and MAN INCOME: both variables are control variable for social class They were included in the model to assess whether higher social status lead more openness towards modern behaviours, or whether on the contrary cohabitation may be seen as ‘poor people marriage’, i.e. an option preferred by less fortunate couples in terms of human and material capital. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

10. Gender variables • GENDER DIFFERENCES IN AGE: The variable included in the model is: Older partner (1: man, 2: same age, 3: woman) We hypothesised a less traditional assortative mating by age (ITEMS 2-3) for cohabiting couples • GENDER DIFFERENCES IN EDUCATION: The variable included in the model is: Partner with higher educational level (1: man, 2: same educational level, 3: woman) We hypothesised an higher probability of cohabiting rather than being married when the woman is the more educated partner • GENDER DIFFERENCES IN INCOME: The variable included in the model is: Partner with higher income from work (1: man, 2: same income, 3: woman) We expect higher odds of cohabitation for couples with a fairer income distribution (or more favourable to the woman) between partners “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

11. Gender variables • WORK To describe the participation of the two partners to the labour market we included in the model two variables, one sinthesising the extent of role specialisation within the couple and one accounting for the two partners’ degree of satisfaction for the main work activity performed. Specifically the variables used were: Employment status (1: both working, 2: only man working, 3: other arrangements) Partner more satisfied with his/her main activity (1: man, 2: same satisfaction, 3: woman) • CARE AND HOUSEHOLDS TASKS No information was available onthe extent to which each partner perform households task or care activities. We built a variable which tell us if household tasks are perceived by women as an obstacle to work the amount of time they wish to and one describing how partners share care duties. Woman not working or working less than she wants because of household tasks or care of family members (1: yes, 0:no) Partner caring for children of other relatives (1: woman only, 2: only man or both, 3: neither man nor woman) “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

12. Gender variables • FREE TIME AND SOCIAL CONTACTS Last aspect we consider of relevance in determining the gender balance within the couples regarded the amount of free time left to each partner after work, household tasks and care activities. We hypothesised partners in consensual unions to have a fairer sharing of free time than married partners We included in the models the variables: Partner more satisfied with the amount of free time (1:man, 2:same satisfaction, 3. woman) Partner meeting more often friends and relatives (1:man, 2: same frequency, 3: woman) “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

13. Method The two typologies of union were compared through means of simple logistic regressions (Dependent variable:1= consensual union, 0=marriage) We utilised the step-forward procedure so that only statistically significant variables were included in the final models. We estimated a total of 7 models: • Model Europe • Model Northern Europe, Model Southern Europe, Model Western EuropeModel Italy, Model Northern Italy • Model Great Britain “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

14. Main results: • For reasons of time economy, i will not show tables with model outcomes but i will only sinthesisize main results, highlighting statistically significant associations, strength and direction of the observed relations. • On the whole, omnibus tests of coefficients based on likelihood ratio tests show that models perform satisfactory well. • The explained variability, on the basis of Nagelkerke R square, is lower for model Southern Europe and Model Italy and higher for model Europe and model Northern Europe “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

15. The models

16. Model 1: Europe • RESIDENCE: residence in the Southern Europe (ref. Northern Europe) reduces significantly the probability for couple to be cohabiting rather than married; same effect but with a weaker influence is observed regarding residence in western Europe. • AGE: the probability of cohabiting rather than being married decreases with age, and this effect is particularly strong as we move from age class 16-25 to age class 26-30. Moreover we observe that the probability is also significantly much lower for couples with more traditional age mating by age (same age or man older). The two variables turned out to be statistically significant and with the same effect on the probability of cohabiting for all the implemented models • EDUCATION: taking as reference women educational level = high, we observe a slightly lower probability of cohabiting for couples where the woman holds a medium educational level, and a slightly higher probability for couples where the women holds a low educational level. As to gender differences in education between partner, the probability of being in a consensual union rather than in a marriage decreases if partner have the same educational level or if it is the man to be more educated, compared to the couples where the woman is the more educated partner. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

17. Model 1: Europe • INCOME: higher man incomes increases the probability for a couple to be cohabiting rather than married. As to gender differences in income, we observe a significantly higher probability of cohabiting for couples in which the income distribution between partner is fair or favourable to the woman. • OBSTACLES: if households tasks or care of family members do not represent an obstacle for women to work, or to work the amount of time they wish to, the probability of cohabiting rather than be married is significantly higher. • CARE: compared to situations in which the woman alone is responsible for the caring of children and family members, couples where neither man nor woman are involved in such tasks are more likely to be cohabiting than married. • SATISFACTION: woman being more satisfied of her main activity increase the probability of being cohabiting rather than married compared to couples characterised by man being the most satisfied partner. • EMPLOYMENT STATUS, FREE TIME and FREQUENCY OF MEETING FRIENDS AND RELATIVES did not enter the model “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

18. Model 2: Northern Europe • AGE: we find the same pattern observed for Europe as a whole. • EDUCATION: same tendencies observed for the model Europe as to gender differences in education. The probability of being cohabiting rather than married is higher the lower the woman educational level is; particularly if woman holds a low educational profile the couple is around 3.5 times more likely to be cohabiting. • CARE: as observed in model Europe. • SOCIAL CONTACTS: the probability of cohabiting is higher for couples with the woman meeting more often friends and relatives. • INCOME, EMPLOYMENT STATUS, OBSTACLES, SATISFACTION AND FREE TIME did not enter the model “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

19. Model 3: Southern Europe • AGE: again we observe the same pattern seen for Europe and for Northern Europe. But this time the decreases of exp(b) from an age class to the other are more gradual. • EDUCATION: woman educational level is significant in predicting the probability of cohabitation rather than marriage. Particularly, if the woman holds a medium educational level the probability of cohabitation decreases compared to couples where women have a high educational level. • INCOME: same as in model Europe • OBSTACLES: same as in model Europe. • CARE: if man are involved in caring of family members, the couple is more likely to be married than cohabiting, respect to couples where the woman alone is in charge of care duties. On the other hands, if neither woman nor man are involved in such tasks, then the couples is more likely to be cohabiting. • GENDER DIFF IN EDUCATION, EMPLOYM STATUS, SATISFACTION, FREE TIME and MEETING FRIENDS AND RELATIVES did not enter the model “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

20. Model 4: Western Europe • The model Western Europe is very similar to the model Europe, Particularly: • AGE, EDUCATION,INCOME, CARE, OBSTACLES and SATISFACTION show significant associations to the probability of being married rather than cohabiting and their effects go in the same direction observed in the model Europe. • EMPLOYMENT STATUS, FREE TIME, and FREQUENCY OF MEETING FRIENDS AND RELATIVES did not enter the model. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

21. Model 5:Italy • RESIDENCE: living in the South and Islands instead of in the Centre-North implies a much lower probability of cohabiting rather than be married. • AGE: the observed tendency is similar to that registered in the rest of Europe; however the association is significant only for the age class 36-40 onwards. Particularly low is the probability of cohabiting rather than living in a marriage if the woman is in the age group 36-40. As to gender differences, once again traditional assortative mating by age (same age or man older) is significantly associated to a lower probability of cohabitation vs. marriage. • OBSTACLES: contrary to what observed in the previous regressions, when households tasks or care of family members do not represent an obstacle for women to work, or to work the amount of time they wish to, the probabilities of cohabiting rather than be married are significantly lower • CARE: when the man is involved in the care of family members, the probability of cohabiting rather than be married is significantly lower. • EDUCATION, INCOME, EMPLOYM STATUS, FREE TIME, SATISFACTION, FRIEND AND RELATIVES not significant “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

22. Model 6: Northern Italy • AGE: same as in model Italy • INCOME: while man income turned out to be not significant, gender differences in income distribution show a significant association with the odds of cohabiting; particularly if partners have the same amount of income, the odds of cohabiting to be married are 2.1 to 1. • CARE: same as in Southern Europe and Italy • EMPLOYMENT STATUS: this is the only case in which we observe a significant association with partners working arrangements. Specifically, the male breadwinner model appears significantly associated to a lower probability of living in a consensual union compared to the couples with both partners working. • EDUCATION, OBSTACLES, FREE TIME, SATISFACTION and RELATIVES and FRIENDS not significant “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

23. Model 7: Great Britain • RESIDENCE: we included in the model more than one variable accounting for variability due to place of residence. We used the Government Office Regions (9 + Scotland), then we utilised the division of GB into England (both as a whole and distinguishing between South east and rest of England, due to considerations on sample size), Scotland and Wales, and finally we tried to include the degree of urbanization. However, none of this variable turned out to be significant in predicting the odds of cohabiting. • AGE: same pattern observed in the previous models • CARE: compared to couples where the woman alone is in charge of care duties, couples where neither man nor woman are involved show a odds of being in a informal union of 2.3 to 1. • All the other variables not significant “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

24. To sum up: • Assortative mating by age, educational level and income more favourable to women turned out to be positively associated to higher odds of cohabitation. • With the only exception of Italy (opposite direction), when households tasks or care of family members do not represent an obstacle for women to work, or to work the amount of time they wish to, the probabilities of cohabiting rather than be married are significantly higher. • As to involvement in care duties, the general pattern is that when neither man nor woman are involved in such duties the probability of being cohabiters are higher. In Southern Europe and Italy, however, if the man is involved in care activities, this reduces the probability of cohabiting rather than being married • Employment status is significant only for Northern Italy, where male breadwinner model decreases the probability of cohabitation • When significant (only Europe and Western models), woman being more satisfied for her main work activity shows a positive association with the probability of cohabiting rather than be married. • In the model Northern Europe the material aspects related to work and income turned out to be not relevant in discriminating between marriage and cohabitation, whereas woman’s quality of life appeared to be much more important. • In Southern Europe and Italy cohabitation is still not widespread and not clearly defined as a modern behaviour, whereas Northern Italy shows more similarities with Western Europe. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

25. Conclusions • Even though every single dimension of the gender inequality does not always play a role to the same extent and in the same direction in all the observed contexts, we may assert that obtained results provide a first proof of the validity of our hypotheses. • Cohabitations are often associated to reduced gender differences in age, education and income, and to situations in which households tasks and caring activities do not represent an obstacle for women to work. “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini

26. Further developments of the study • First, we aim at improving logistic regressions, taking more into account regional specifities. Particularly we are interested in finding suitable gender predictors for the case of Great Britain. • Second, we aim at a better exploitation of the longitudinal dimension of the ECHP. Particularly, inspired by the work of Moore (2003)1, we will follow couples from first to last wave and we will turn our attention to the transition cohabitation->marriage, distinguishing between stayers and incomers. The intent is to assess whether the couples who experience the transition from an informal cohabitation to a formal marriage reduce their level of ‘gender equality’, or, conversely, whether the experience of a cohabitation stably defines egalitarian gender relations within the couple. 1 (Moore, G. 2003 “Estimating the Reciprocal Effect of Gender Role Attitudes and Family Formation: A Log-linear Path Model with Latent Variables” European Journal of Population 19: 199-221 ) “Are consensual unions more gender equal than marriages?”, by Pinnelli, Fiori, Testini