Data used

9The "Triple Biographie" survey has been presented in greater detail elsewhere ([4], [11]). In this article we will study all those interviewed (4,601 individuals), separately for each sex, and divided into broad cohorts. We thus obtain a sample of 2,105women who on their 14^th birthday lived in a non-metropolitan, i.e. little-urbanized, area, of whom 1,278 were born between 1911 and 1925 and 827 between 1926 and 1935, and a sample of 446 women who lived in a metropolitan area at their 14^th birthday. The small number in this last group prevents any sub-division into smaller cohorts. For men, the figures are 1,704 who lived in a non-metropolitan area on their 14^th birthday, of whom 929 were born between 1911-1925 and 775 between 1926-35, and 346 men who lived in a metropolitan area on their 14^th birthday.

Table A

Variables

Index of Variables nuptiality women’s men women fertility — none 0 0 0 — CEP only 1 0 1 Educational Level — certificate of apprenticeship (CAP) or technical baccalauréat 2 1 3 — at least baccalauréat or higher education 2 2 2 — eldest 1 1 Eldest Child — not eldest 0 0 Siblings number number Parents’residence during subject’s childhood — metropolitan area 1 1 — non-metropolitan area 0 0 Nationality — naturalized or foreign 1 1 — born French 0 0 Father s occupation — farmer or farm worker 1 1 — other 0 0 women spouses First or previous occupation — never worked 0 0 0 0 — farmer or farm worker 0 1 1 0 — manual or unskilled worker 1 2 2 1 — skilled worker, professional, commerc. and indust. management 2 33 2 — middle management and other employees 2 3 4 3 Type of dwelling — in parents’ home or home owner 0 0 — tenant or other 1 1

Variables

10The variables used in the semiparametric analysis are indicated in Table A. Some are coded differently according to the sex of the subject and the event studied. This choice was based on analyses of earlier INSEE surveys ([7], [12]) which involved far larger samples, and on our own tests. Here only a small number of variables can be studied given the limited nature of our sample.

Methods of analysis

11We shall only present a schematic diagram to show the links between marriage and migration, and the same scheme is used for the links between migration and the birth of successive children. In Diagram 1 we show the various possible states in which a person can be living, and the transition intensities between these states, which we need to estimate.

Diagram 1

Different states in which an individual may be living and the hazard rates of transition from one to another

Different states in which an individual may be living and the hazard rates of transition from one to another

12For example, at time t, the migration hazard rate for the N^c_r (t) individuals, still single and living in non-metropolitan areas, will be h^c_ru (t) and the marriage hazard rate will be h^cm_r (t).

13When out-migration and marriage occur during the same year, the individuals are excluded from the population at risk, as if they had ceased to be observed on the date when each of these events occurred. In consequence, simultaneous occurrences were excluded, but it is useful to include them in certain cases in order to represent the individual behaviour patterns more clearly. When this was done we state it explicitly.

14Thus, we can compare the marriage hazard rate of those individuals who remained in the non-metropolitan area, h^cm_r (t), with the rates for those who had already migrated to a metropolitan area h^cm_ru (t). This last hazard rate will also depend on the age of the individual at the time of migration (θ), and should therefore be written h^cm_ru (t∣θ). Here, we have assumed that this dependence will only affect the hazard rate slightly.

15Under these conditions it is possible to test the equality of the different nuptiality and migration hazard rates, by using methods which have been presented in an earlier article [6].

16To show the effect of the various characteristics on these hazard rates we use the following semiparametric model :

17

18where β₁, β₀ and β₂ are the coefficients to be estimated by using variables Z (intervening before the modifying phenomenon) and Z’ (intervening after the modifying phenomenon, and which may be equal to or different from z), H (x) is the Heaviside function (H (x) = 0 for x ≤ 0, H (x) = 1 for x >0), u is the date of the modifying phenomenon, h (t) the nonparametric baseline hazard. In this model it is assumed that the different variables exercise a multiplicative effect on the estimated hazard rate.

19We also have tests that enable us to judge whether a group of variables affects the phenomenon under study, and these are presented in detail in [6].

Marriage and mobility

21In Figure 1 we show the cumulated nuptiality hazard rates of women who lived in metropolitan and non-metropolitan areas at their 14^th birthday. In both cases a slight delay in timing of marriage can be seen amongst migrants, most significantly between ages 21 and 31 for women from non-metropolitan areas and at various ages between 20 and 31 for the others. This delay is made up after the age of 40. The proportion never married at this age comes to 8 % for women from non-metropolitan areas, irrespective of whether they remain there or emigrate, and to 12 % for those from metropolitan areas. This result confirms other observations ([12], p. 102), which have shown that the likelihood of remaining single is higher in cities than elsewhere. It is interesting that there is a slight delay in marriage amongst migrants, linked to a period of adaptation on arrival in the new environment, but this delay is entirely made up later.

Figure 1

Cumulative nuptiality hazards for migrant and non-migrant women according to their origin

Cumulative nuptiality hazards for migrant and non-migrant women according to their origin

22Men’s behaviour is different. In Figure 2 we show the cumulated nuptiality hazard rates of men who lived in the metropolitan and non-metropolitan areas at their 14^th birthday. In both cases, the proportion never married at their 45^th birthday is lower among migrants than among those who remained in their place of origin. This is true for men from non-metropolitan areas from their 30^th birthday, and for those from metropolitan areas from their 25^th birthday. In the first group, this behaviour may be linked with that of men working in the agricultural sector [6] who constitute a considerable proportion of men living in non-metropolitan areas. However, we may note that those migrating out of large towns do not follow the behaviour patterns of those working in agricultural occupations and that marriage rates among them are higher than in the original rural population.

Figure 2

Cumulative nuptiality hazards for migrant and non-migrant men according to their origin

Cumulative nuptiality hazards for migrant and non-migrant men according to their origin

23We next consider how marriage influences the mobility of those who lived in non-metropolitan areas at their 14^th birthday. In Figure 3 we show the cumulated mobility hazard by marital status of men and women. The behaviour of both sexes is the same, but it is accentuated amongst women : once married, their chance of moving to metropolitan areas is divided by a factor of two, compared with only 1.5 for men. This consolidates what we had already observed elsewhere [5] : we showed that whereas marriage delayed changes of residence, it hardly affected moves between "départements" ; we shall now show that it greatly affects migration to the city. This strong dependence of mobility on marital status more than balances any effect of marriage on migration.

Figure 3

Cumulative mobility hazards for migrations towards the metropolitan areas by men and women, with simultaneous occurrences either included or excluded

Cumulative mobility hazards for migrations towards the metropolitan areas by men and women, with simultaneous occurrences either included or excluded

24It must be noted that cases when marriage and migration occur simultaneously are rare : only 3.3 % of women of non-metropolitan origin (3.8 % of marriages but 23 % of moves to the city) and 1.9 % of men (2.1 % of marriages and 10.7 % of moves to the city). It is interesting to note how including these people among moves of married people to the city modifies the relative position of the married and unmarried. These results are shown in Figure 3. Cases of simultaneity relate mainly to young people under the age of 26, so that the two curves are at the same level before this age. This result confirms our observations on agricultural populations [6].

25We have not taken into account results relating to those who migrated out of the metropolitan areas, because of the small numbers observed, and of the low rates of migration to less urbanized areas.

The effect of social and occupational characteristics

26To extend our analysis we must now turn to a semiparametric approach which will make it possible to study the effect of various characteristics on an individual’s behaviour.

27In Table 1 we show an analysis of the nuptiality of men and women who lived in non-metropolitan areas at their 14^th birthday, separating those who remained there from those who migrated. There are two distinct cohorts – those born between 1911-1925 and those born between 1926-1935. The results for the two sexes differ widely.

28The educational level, measured by qualifications obtained, clearly plays an important role. Men’s ability to find a wife increases with their educational level. Men in older cohorts found one more easily before moving to the big city [1], but in the more recent cohorts men found a wife more easily after having moved. For women, however, academic qualifications delay marriage. In the older cohorts this delay was found both before and after migration, in more recent cohorts it was limited to the period before moving.

29The occupation of the individual surveyed and that of his or her father has an equally important effect.

30The probability of marriage for farmers’ sons was higher when they left this sector of the economy. This advantage was maintained after migration in cohorts born between 1911-25, but was lost in later cohorts. The effect of occupation before migration was greater in the first cohort than in the second – those working in the agricultural sector married later than manual or unskilled workers, who in turn married later than others. This confirms results of other studies ([12], p. 72). However, when migration occurs, this effect is entirely eliminated and occupation after migration does not seem to affect nuptiality.

31Amongst farmers’ daughters from the cohorts 1911-25, their father’s occupation delayed their marriage even longer when they left the agricultural community and took up different employment. Migration to the city does not alter this effect. Women in more recent cohorts seem freer from family influence, since their marriages were not delayed when their father was engaged in agriculture. The part played by the first occupation was the same in both cohorts : non-working women married sooner than women who worked in agriculture, who in their turn married sooner than did skilled women workers, who married sooner than women in other occupations. Again, this result agrees with other studies on the subject ([12], p. 73). Migration to a large city in no way alters this situation for women, unlike in the case of men.

32Amongst the different family characteristics which we have studied, the effect of being the eldest child differs at different times and between the sexes.

33Being an eldest daughter does not affect the probability of marriage before migration to a large city. After such a move, the chance of marriage will be higher for the eldest daughters in cohorts born before 1925, whereas in later cohorts marriage will be delayed. On the other hand, being an eldest son does influence nuptiality before migration to a large city : for cohorts born before 1925, the effect is favourable, for those born later it is unfavourable. In earlier cohorts, migration cancels out the effect which is still found in later cohorts. A sociological survey on the status of the eldest child would be necessary to explain these apparently significant results.

34The various characteristics considered here are not independent : educational level and occupation, for example, are closely linked. In Table 1, we have illustrated the effect of different characteristics – the educational level, father’s occupation, the subject’s first occupation and the fact of being an eldest child – taken simultaneously. It is clear, especially for women, that the effect of education and occupation on marriage, although related, is separate. This result confirms what had already been illustrated in previous family studies in which different methods had been used [1] : for people with the same level of education, occupation has an effect, and for those in the same occupation, the level of education is significant.

Table 1

Nuptiality of men and women of non-metropolitan origin, modified by migration to metropolitan areas

Cohorts born between 1911-1925 Cohorts born between 1926-1935 Variables taken separately Variables taken simultaneously Variables taken separately Variables taken simultaneously β1 β0 β2 β1 β0 β2 β1 β0 β2 β1 β0 β2 Variables Men Migration to metropolitan area 0.115 0.271 0.133 0.079 Educational level 0.112 ** 0.252 – 0.160 0.085 * – 0.163 0.126 ** – 0.167 0.283 * 0.049 0.461 ** Eldest child 0.107 * 0.247 * – 0.318 * 0.076 – 0.364 * – 0.129 * 0.172 – 0.098 – 0.167 ** – 0.001 No. of siblings – 0.010 – 0.027 0.048 0.003 0.200 – 0.021 Parents resident in met. area 0.137 0.030 0.362 0.083 0.166 – 0.237 Foreign or naturalized – 0.163 * 0.086 0.346 0.075 0.139 – 0.072 Father in agriculture 0.240 ** – 0.072 0.214 0.227 ** 0.517 * 0.293 * 0.388 ** – 0.435 ** 0.157 * – 0.250First socio-occupational group 0.091 ** 0.315 * – 0.210 * 0.022 – 0.217 0.232 ** 0.307 * – 0.257 ** 0.170 ** – 0.322 * Tenant at the time of move 0.699 – 0.673 ** 0.059 0.088 Socio-occ. group on moving 0.168 – 0.048 0.205 – 0.065 Maximum log-likelihood – 5152.30 (father in agriculture) – 5148.65 – 4115.44 (first socio-occ. group) – 4109.09 Variables Women Migration to metropolitan area – 0.083 – 0.158 0.051 0.478 ** Educational level – 0.339 ** – 0.038 – 0.071 – 0.306 ** – 0.173 – 0.231 ** 0.023 0.119 – 0.218 ** 0.270 Eldest child 0.043 – 0.263 * 0.263 ** 0.094 * 0.777 ** – 0.021 0.205 – 0.512 ** 0.043 – 0.500 * No. of siblings 0.017 * 0.097 – 0.051 0.005 – 0.221 0.071 * Parents resident in met. area – 0.031 – 0.017 – 0.582 – 0.063 0.035 0.199 Foreign or naturalized – 0.031 – 0.720 ** 0.548 ** – 0.066 0.033 0.182 Father in agriculture – 0.118 ** 0.051 – 0.186 – 0.017 – 0.310 0.069 0.196 – 0.230 0.155 ** – 0.188 First socio-occupational group – 0.107 ** – 0.136 0.034 * – 0.059 ** 0.105 – 0.071 ** 0.267 – 0.133 – 0.050 * – 0.147 Tenant at the time of move – 0.162 0.096 0.664 – 0.639 Socio-occ. group on moving 0.006 – 0.078 – 0.045 0.082 Maximum log-likelihood – 7616.56 (academic qualif’s) – 7608.56 – 4564.76 (academic qualif’s) – 4559.72

Nuptiality of men and women of non-metropolitan origin, modified by migration to metropolitan areas

^* Result significant at 10 % level.
^** Result significant at 5 % level.
β₁ = Principal effect ; β₀ = Modification ; β₂ = Interaction.

35In Table 2 we show the results of an analysis of the nuptiality of men and women who lived in metropolitan areas at their 14^th birthday. Because of the small numbers involved, we were unable to distinguish between the two earlier cohorts, and thus our results are less clear. Nevertheless, the effect of the different characteristics is similar to that observed outside the metropolitan areas. Educational level and occupation still have opposite effects for men and women. On the other hand, none of the family characteristics has a significant effect on nuptiality.

36Let us now consider the converse – the effect of marriage and other characteristics on migration to the metropolitan areas. In Table 3 we show the results of this analysis, and in this case the behaviour of men and women is similar. We will, therefore, consider them together.

37Marriage clearly delays migration, and as the nonparametric analysis showed, this delay is greater for women than men.

38Family origin, on the other hand, has the same effect for men and women. If the father worked in the agricultural sector, the chance of migration is higher, both for the single and the married. If an individual spent some years in a city during childhood, he or she is likely to return there as an adult, especially if he or she marries someone born in a city or who lived there before marriage. The city also appears attractive for those born abroad ; Italians, who accounted for the highest proportion of foreign immigrants at the time, were particulary attracted by the urban "départements" nearest their homeland, i.e. near the Mediterranean coast.

39The effect of an individual’s position in the family differs for each sex. Eldest daughters seem to experience difficulties in moving to the city, but the effect of being the eldest is never really significant. However, the number of siblings born after the eldest daughter is significant and always works in the same direction : the larger the family, the greater the probability that she will migrate to a metropolitan area. This effect continues, although to a lesser extent, even after marriage. Being an eldest son is only significant in the first cohort, and increases the probability of migration to a city. However, once the eldest son has married, this effect no longer applies. The number of his sibs will have an effect especially after marriage, and the probability of migration will increase with the number of siblings.

40Finally, we shall consider the influence of occupational status, which will depend on the educational level achieved and first occupation, and which has the same effect on both cohorts. The higher the educational level, the greater the attraction of the city, both before and after marriage. Furthermore, the spouse’s educational level will have the same effect. The effect of a man’s first occupation is noticeable before marriage, whereas for women it begins to operate after marriage. Those working in agriculture are less attracted to the city than other manual or unskilled workers, who in turn migrate less than do members of other groups. After marriage, this effect is generally combined with the influence of the subject’s new occupational group and that of his or her spouse.

Table 2

Nuptiality of men and women of metropolitan origin, modified by migration to non-metropolitan areas ; migration to non-metropolitan areas modified by marriage

Variables taken separately Variables taken simultaneously Variables taken separately Variables taken simultaneously β1 β0 β2 β1 β0 β2 β1 β0 β2 β1 β0 β2 Variables Men Women Nuptiality modified by migration to non-metropolitan areas Migration to non-metropolitan area 0.400 ** – 0.131 0.085 0.179 Educational level 0.023 0.345 * 0.039 – 0.040 0.036 – 0.340 * 0.140 0.064 – 0.342 ** – 0.055 Eldest child 0.068 0.403 ** – 0.013 0.054 0.090 – 0.007 0.033 0.111 0.074 0.235 No. of siblings – 0.032 0.394 ** 0.000 – – 0.024 0.175 – 0.035 – – Parents resident in met. area 0.009 0.387 * 0.030 – – 0.071 0.192 – 0.227 – – Foreign or naturalized – 0.528 ** 0.401 ** – 0.138 – – 0.257 0.076 0.155 – – Father in agriculture 0.180 – 0.157 0.666 * 0.068 1.031 ** 0.134 0.219 – 0.155 0.257 * – 0.312 First socio-occupational group 0.122 * 0.481 ** – 0.068 0.126 – 0.343 * – 0.090 ** 0.029 0.022 – 0.043 0.107 Tenant at the time of move 0.443 ** – 0.050 – – 0.218 0.512 * – – Socio-occ. group on moving 0.187 0.202 * – 0.142 – 0.034 – – Maximum log-likelihood – 1629.41 (foreigner naturalized) – 1627.84 – 2179.50 (academic qualif’s) – 2177.31 Variables Migration to non-metropolitan areas modified by marriage Marriage 0.035 0.446 – 0.286 * – 0.213 Educational level 0.466 ** 0.356 – 0.267 0.588 ** – 0.348 0.231 * – 0.265 0.139 – 0.258 * – 0.064 Eldest child 0.182 0.030 0.000 – – 0.204 – 0.286 0.001 – – No. of siblings – 0.016 – 0.027 0.023 – 0.016 0.022 0.061 – 0.114 – 0.071 0.087 * – 0.056 Parents resident in met. area – 0.567 ** 0.145 – 0.179 – 0.507 * – 0.151 – 0.752 ** – 0.213 – 0.056 – 0.715 ** 0.078 Foreign or naturalized 0.093 0.041 – 0.636 0.083 – 0.480 0.452 – 0.219 – 1.153 * 0.411 – 1.035 * Father in agriculture – 0.180 – 0.173 0.242 – 0.382 0.738 – 0.230 – 0.855 ** 0.642 – 0.042 0.503 First socio-occupational group 0.005 0.320 – 0.239 – 0.171 – 0.273 0.001 – 0.601 * 0.157 0.060 – 0.009 Spouse’s education – 0.021 0.151 0.064 – 0.649 ** 0.314 ** 0.165 Spouse born abroad 0.030 0.081 – 0.153 – 0.263 – 0.273 – 0.518 Spouse born in met. area 0.392 * – 0.818 ** – 0.840 ** 0.100 – 0.855 ** – 0.816 ** Spouse urban resident before mar. 0.688 ** – 0.847 ** 0.220 – 0.605 ** Spouse’s socio-occ. group 0.384 – 0.218 ** – 0.558 ** 0.200 * Socio-occ. group after marriage 0.386 – 0.285 * – 0.470 ** 0.095 Tenant or housed by empl. after mar. – 0.432 0.577 * – 0.856 ** 0.739 ** Maximum log-likelihood – 798.55 (parents resident in met. area) – 791.89 – 829.92 (parents resident in met. area) – 817.13

Nuptiality of men and women of metropolitan origin, modified by migration to non-metropolitan areas ; migration to non-metropolitan areas modified by marriage

^* Result significant at 10 % level.
^** Result significant at 5 % level. β₁ = Principal effect ; β₀ = Modification ; β₂ = Interaction.

Table 3

Migration to metropolitan areas modified by marriage

Cohorts born between 1911-1925 Cohorts born between 1926-1935 Variables taken separately Variables taken simultaneously Variables taken separately Variables taken simultaneously β1 β0 β2 β1 β0 β2 β1 β0 β2 β1 β0 β2 Variables Men Marriage – 0.449 ** – 1.977 ** – 0.732 ** – 2.030 ** Educational level 0.708 ** – 0.529 ** 0.029 0.500 ** 0.027 0.369 ** – 0.891 ** 0.103 – 0.274 ** 0.034 Eldest child 0.347 ** – 0.289 – 0.429 * – – – 0.070 – 0.729 ** – 0.017 – – No. of siblings – 0.067 ** (1) – 0.696 ** (1) 0.078 * (1) – 0.025 0.073 0.001 – 1.125 ** 0.106 ** 0.027 0.108 ** Parents resident in met. area – – – – – 1.769 ** – 0.599 ** – 1.297 ** 1.983 ** – 1.796 ** Foreign or naturalized – 0.133 – 0.638 ** 1.486 ** 0.089 1.507 ** 1.079 ** – 0.907 ** 0.497 * 1.321 ** – 0.015 Father in agriculture 0.573 ** – 1.292 ** 0.938 ** 0.001 0.955 ** 0.326 * – 0.888 ** 0.164 – 0.151 0.767 ** First socio-occupational group 0.592 ** – 0.769 ** 0.222 0.370 ** 0.002 0.332 ** – 0.730 ** – 0.096 0.200 – 0.128 Spouse’s education – 0.612 ** 0.750 ** 0.275 * – 0.909 ** 0.572 ** 0.440 ** Spouse born abroad – 0.564 ** 1.031 ** 0.341 0.941 ** 1.166 ** 0.777 ** Spouse born in metropolitan area – 0.580 ** (1) 1.443 ** (1) 1.217 ** – 0.884 ** 1.959 ** 1.866 ** Spouse urban resident before mar. – – – 0.870 ** 1.714 ** – Spouse’s socio-occ. group – 0.739 ** 0.213 ** – 0.799 ** 0.045 – Socio-occ. group after marriage – 1.016 ** 0.490 ** – 0.567 ** – 0.167 – Tenant or housed by employer after mar. – 0.747 ** 0.438 ** – 0.709 ** – 0.034 – Maximum log-likelihood – 1137.54 (first socio-occ. group) – 1106.37 – 975.94 (foreign or naturalized) – 931.84 Variables Women Marriage – 1.000 ** – 1.717 ** – 0.999 ** – 2.520 ** Educational level 0.200 ** – 1.058 ** 0.404 ** 0.255 ** 0.111 0.126 – 1.057 ** 0.271 0.175 – 0.153 Eldest child – 0.190 – 1.035 ** 0.085 – – – 0.248 – 1.104 ** 0.291 – – No. of siblings 0.065 ** – 0.864 ** – 0.046 0.075 ** – 0.016 0.065 ** – 0.871 ** – 0.036 0.064 ** – 0.003 Parents resident in met. area 1.158 ** – 1.009 ** 0.085 1.168 ** – 0.370 0.661 ** – 1.003 ** 0.149 0.682 ** 0.257 Foreign or naturalized 0.685 ** – 1.344 ** 0.305 0.624 ** 0.002 0.265 – 1.256 ** 1.359 ** 0.234 0.872 ** Father in agriculture 0.505 ** – 1.423 ** 0.622 ** 0.468 ** 0.332 0.140 – 1.415 ** 0.561 ** 0.208 0.239 First socio-occupational group – 0.012 – 1.446 ** 0.287 ** – 0.100 0.151 – 0.061 – 1.512 **0.308 ** – 0.114 0.217 * Spouse’s education – 1.384 ** 0.464 ** 0.183 * – 1.576 ** 0.564 ** 0.506 ** Spouse born abroad – 1.095 ** 0.838 ** 0.679 ** – 1.264 ** 1.413 ** 1.115 ** Spouse born in metropolitan area – 1.146 ** 1.759 ** 1.550 ** – 1.124 ** 1.444 ** 1.474 ** Spouse urban resident before mar. – 1.195 ** 1.894 ** – – 1.184 ** 1.989 ** – Spouse’s socio-occ. group – 1.034 ** 0.034 – – 1.105 ** 0.099 – Socio-occ. group after marriage – 1.234 ** 0.166 ** – – 1.191 ** 0.126 – Tenant or housed by employer after mar. – 1.532 ** 0.764 ** – – 1.173 ** 0.253 – Maximum log-likelihood – 1712.18 (academic qualif’s) – 1646.99 – 1075.47 (foreign or naturalized) – 1051.24

Migration to metropolitan areas modified by marriage

^* Result significant at 10 % level.
^** Result significant at 5 % level.
⁽¹⁾ After 10 iterations the estimation does not converge.
β₁ = Principal effect ; β₀ = Modification ; β₂ = Interaction.

41The fact that we have considered the effect of different variables simultaneously does not affect the essence of our previous results. This approach accentuates the delaying effect of marriage and underlines the significant influence of the first stage in the family cycle on mobility towards the city : once married in a non-metropolitan area, individuals are much less likely to leave that area, whatever the couple’s characteristics.

42We shall look briefly at the modifying effect of marriage on migration out of the city. This is illustrated in Table 2, and we shall discuss only two significant results. First, the effect of educational level which, surprisingly, acts in the same way as it did on migration out of the non-metropolitan areas. Educational level, therefore, seems to be an important spur in both directions. Spouse’s educational level, especially for women, works in the same way. On the other hand, those who lived in a metropolitan area during childhood, and those who married someone born in such an area, were less likely to leave the city, as might be expected.

Births and mobility

44In Figure 4 we show the cumulated fertility rates for different birth orders of women who lived in a non-metropolitan area at their 14^th birthday, depending on whether they later migrated or not. In Figure 5 the rates for women from the cities are shown.

45The contrast between the two series of curves is striking. For women from outside the metropolitan areas, migration to an urban area reduces their fertility after the birth of their second child by one-third. Conversely migration to a non-metropolitan area brings about a sharp rise in fertility after the birth of their first child [2] for women from the cities. After migration, the hazard rates are 1.4 to 2 times higher than for those who remained. It seems likely that women adapt quickly to the fertility prevailing in their area of destination.

Figure 4

Cumulative order-specific fertility hazards for women of non-metropolitan origin according to whether they migrated out or not

Cumulative order-specific fertility hazards for women of non-metropolitan origin according to whether they migrated out or not

Figure 5

Cumulative order-specific fertility hazards for women of metropolitan origin according to whether they migrated out or not

Cumulative order-specific fertility hazards for women of metropolitan origin according to whether they migrated out or not

46However, it is possible that the behaviour of women who ultimately migrate may have been different from that of women who stayed in their area of origin throughout their lives. This is why, in Figures 4 and 5, we have added the cumulated fertility hazard rates of these future migrants before they moved.

47For migrations to metropolitan areas (414 women), the results in Figure 4 confirm the previous hypothesis : women (less than 35 years old at their third birth) who migrate to these areas timed their fertility before migration differently from that of women who remained in non-metropolitan areas. Moreover, their timing was the same as that of women who had already moved to these areas. However, the behaviour of women who had their third birth after the age of 35 was similar before migration for those who migrated and for those who did not. In such cases we are concerned with very late migrants.

48Figure 5 shows a contrasting effect for the 144 women who moved to non-metropolitan areas : the fertility of women of urban origin who later migrated to less urbanized areas was not markedly different from that of those who remained in the city. In their case, fertility increased significantly only after migration.

49Figure 6 gives the cumulated hazard rates for mobility to the city of women with no, one, or two, children depending on whether they moved before or after their nth child. In Figure 7 the same rates are shown for women who left the cities. Even though the latter group is small, the two graphs present a marked contrast.

50Migration to the city, depending on the number of children already born or yet to come, follows the same pattern as that observed earlier for marriage. The mobility rates of individuals before a birth of order n are reduced to approximately two-thirds of their value thereafter. These results are very significant. The opposite pattern appears for migration in the reverse direction to a non-metropolitan area, though the effect is less significant because of the small numbers in the sample. Nonetheless, it may be concluded that the mobility rates of those living in the city are larger after a birth of order n than before, even though this increase is smaller than the reduction observed for moves in the opposite direction.

51Thus whilst there seems to be a certain attachment to non-metropolitan areas, which increases with the number of children, there is also mobility out of metropolitan areas which likewise increases with the number of children.

Figure 6

Cumulative mobility hazards for migrations to metropolitan areas by women with zero, one or two children, according to whether they migrated out before or after the n^th birth (after age 22)

Cumulative mobility hazards for migrations to metropolitan areas by women with zero, one or two children, according to whether they migrated out before or after the n^th birth (after age 22)

Figure 7

Cumulative mobility hazards for migrations to non-metropolitan areas by women with zero, one or two children, according to whether they migrated out before or after the n^th birth (after age 27)

Cumulative mobility hazards for migrations to non-metropolitan areas by women with zero, one or two children, according to whether they migrated out before or after the n^th birth (after age 27)

The effect of social and occupational characteristics

52We shall now consider in greater detail the interactions between mobility and the birth of the third child, taking into account some of the individuals’ other characteristics.

53In Table 4 we show an analysis of two cohorts of women who came from a non-metropolitan area, and in Table 5 that for women from metropolitan areas. We believe that it is important to present these two tables together, because the effect of all variables other than migration and the possibility that their parents may have lived in the city during their childhood is completely independent of their origin.

54When we consider the way in which the birth of the third child is affected by different characteristics, five important effects become apparent. Being the daughter of someone engaged in agriculture will have an effect, irrespective of whether the woman lived in a metropolitan area or not : before migration, this would delay the birth, but after having moved the effect will disappear. This is the only one of the five variables on which mobility has an effect. If a woman has a large number of sibs, she will be more likely to have a third child, irrespective of whether she comes from a rural or an urban background. A woman’s educational level will also affect her fertility : women without any qualifications are the most likely to have a third child, followed by those with only the primary school-leaving certificate (CEP), and finally those with the baccalauréat or higher qualifications. Women with a technical or professional diploma are least likely to have a third child. The woman’s first occupation will also have an effect which is independent of her origin. Women who have never worked are most likely to have a third child, followed by women in the farming sector, manual workers and women in the professions and senior management. Office workers and those employed in middle-management are least likely to have a third child. Finally, women born abroad and naturalized women are more likely to have a third child.

55These characteristics, even if considered simultaneously (see Tables 4 and 5), will affect the chance of having a third child, and the effect of the last four of the characteristics mentioned will be independent of the woman’s place of residence and whether she has moved or not. Only migration from one area to another introduces the significant difference shown in the nonparametric study, and which these characteristics do not permit us to explain : migration to the city induces a marked reduction in the chance of giving birth to a third child, whereas migration to less urbanized areas raises fertility.

Table 4

Birth of the third child amongst women of non-metropolitan origin, modified by migration to a metropolitan area ; migration to the metropolitan areas modified by the birth of the third child

Cohorts born between 1911-1925 Cohorts born between 1926-1935 Variables taken separately Variables taken simultaneously Variables taken separately Variables taken simultaneously β1 β0 β2 β1 β0 β2 β1 β0 β2 β1 β0 β2 Variables Birth of the third child modified by migration to a metropolitan area Migration to metropolitan area – 0.267 ** – 0.688 ** – 0.449 ** – 1.053 ** Educational level – 0.360 ** – 0.363 ** 0.208 – 0.231 ** 0.141 – 0.135 ** – 0.351 – 0.105 – 0.038 – 0.184 Eldest child – 0.079 – 0.483 ** 0.545 – – – 0.088 – 0.478 ** 0.074 – – No. of siblings 0.085 ** – 0.151 – 0.041 0.058 ** – 0.019 0.057 ** – 0.510 * 0.007 0.044 ** 0.014 Parents resident in met. area – 0.058 – 0.404 ** 0.867 ** – 0.218 – 0.398 ** – 0.764 Foreign or naturalized 0.293 ** – 0.265 ** – 0.154 0.138 – 0.412 ** – 0.344 Father in agriculture – 0.222 ** – 0.679 ** 0.584 * 0.023 0.553 * – 0.028 – 0.990 ** 0.713 * 0.126 0.739 * First socio-occupational group – 0.189 ** – 0.316 ** 0.049 – 0.117 ** – 0.011 – 0.151 ** – 0.535 ** 0.057 – 0.137 ** 0.073 Tenant at the time of move – 0.361 ** 0.109 – 0.734 ** 0.310 Socio-occ. group on moving – 0.233 – 0.021 – 0.446 ** – 0.002 Maximum log-likelihood – 3622.12 (academic qualifications) – 3608.64 – 2061.02 (first socio-occ. group) – 2055.83 Variables Migration to metropolitan areas modified by birth of the third child Birth of third child – 0.213 0.017 – 0.387 * – 1.032 Educational level 0.226 ** – 0.556 ** 0.518 ** 0.254 ** 0.429 ** 0.114 * – 0.412 0.057 0.101 0.303 Eldest child – 0.219 * – 0.364 * 0.387 – 0.120 – 0.366 – 0.069 No. of siblings 0.065 ** 0.288 – 0.158 ** 0.096 ** – 0.112 * 0.054 ** – 0.459 0.012 0.060 ** 0.022 Parents resident in met. area 1.206 ** – 0.212 0.029 1.183 ** – 0.636 0.757 ** – 0.373 * – 0.193 0.742 ** – 0.213 Foreign or naturalized 0.863 ** – 0.242 0.016 0.825 ** – 0.518 0.822 ** – 0.623 ** 1.027 ** 0.867 ** 0.036 Father in agriculture 0.734 ** – 0.727 ** 0.743 ** 0.574 ** 0.459 0.355 ** – 0.342 – 0.076 0.335 ** 0.005 First socio-occupational group 0.099 ** – 0.555 ** 0.237 ** 0.077 ** – 0.647 * 0.158 Spouse’s education – 0.357 * 0.176 * – 0.460 * 0.077 Spouse born abroad – 0.383 ** 1200 ** 1.496 ** – 0.874 ** 1.951 ** 1.645 ** Spouse born in metropolitan area – 0.254 ** 0.768 – 0.419 ** 0.854 Spouse urban resident before mar. – 0.357 ** 2.022 ** – 0.420 * 1.551 * Spouse’s socio-occ. group – 0.303 0.097 – 1.063 ** 0.489 ** Socio-occ. group after third birth – 0.730 ** 0.325 ** – 0.730 ** 0.205 * Tenant or housed by employer after 3rd birth – 0.769 ** 0.758 ** – 1.424 ** 1.290 ** Maximum log-likelihood – 1720.76 (father working in agriculture) – 1677.13 – 1091.15 (foreign or naturalized) – 1080.13

Birth of the third child amongst women of non-metropolitan origin, modified by migration to a metropolitan area ; migration to the metropolitan areas modified by the birth of the third child

^* Result significant at 10 % level.
^** Result significant at 5 % level. β₁ = Principal effect ; β₀ = Modification ; β₂ = Interaction.

Table 5

Birth of the third child amongst women of metropolitan origin, modified by migration to a non-metropolitan area ; migration to the non-metropolitan areas modified by the birth of the third child

Variables taken separately Variables taken simultaneously β1 β0 β2 β1 β0 β2 Variables Birth of the third child modified by migration to a non-metropolitan area Migration to non-met. area 0.569 ** 0.254 Educational level – 0.345 ** 0.546 ** 0.097 – 0.171 * – 0.021 Eldest child – 0.173 0.554 0.049 – – No. of siblings 0.119 ** 0.646 ** – 0.031 0.075 ** – 0.014 Parents resident in met. area 0.163 0.542 ** – 0.033 – – Foreign or naturalized 0.597 ** 0.563 ** – 0.077 Father in agriculture – 0.500 ** 0.394 0.201 – 0.228 0.209 First socio-occupational group – 0.209 ** 0.363 0.097 – 0.113 * 0.110 Tenant at the time of move 0.540 ** 0.044 – Socio-occ. group on moving 0.432 ** 0.061 – Maximum log-likelihood – 826.48 (academic qualifications) – 821.98 Variables Migration to non-metropolitan areas modified by birth of the third child Birth of third child 0.145 – 0.110 Educational level 0.216 ** 0.328 – 0.117 0.253 ** – 0.250 Eldest child 0.158 0.004 0.343 No. of siblings 0.038 0.563 * – 0.146 * 0.069 ** – 0.188 ** Parents resident in met. area – 0.870 ** – 0.215 0.623 – 0.890 ** 0.533 Foreign or naturalized – 0.168 0.157 – 0.014 – 0.246 0.402 Father in agriculture – 0.008 – 0.399 0.626 0.182 0.535 First socio-occupational group 0.036 – 0.389 0.261 Spouse’s education – 0.198 0.314 Spouse born abroad 0.159 – 0.125 0.611 Spouse born in metropolitan area 0.492 * – 0.656 * Spouse urban resident before mar. 0.955 * – 0.905 * Spouse’s socio-occ. group 0.205 – 0.053 Socio-occ. group after third birth 0.248 – 0.060 Tenant or housed by employer after 3rd birth – 0.492 0.803 * Maximum log-likelihood – 880.20 (parents resident in met. area) – 823.06

Birth of the third child amongst women of metropolitan origin, modified by migration to a non-metropolitan area ; migration to the non-metropolitan areas modified by the birth of the third child

^* Result significant at 10 % level.
^** Result significant at 5 % level. β₁ = Principal effect ; β₀ = Modification ; β₂ = Interaction.

56We proceed to consider the effects of the birth of a third child on mobility.

57The influence of a woman’s family origins will be different for women of different origins. Daughters of men engaged in agriculture of non-metropolitan origin will be attracted to the city, whereas women of metropolitan origin will not be affected. Likewise, if a woman lived in the city as a child, the city will be more attractive if she lived in a non-metropolitan area at her 14^th birthday, and she will be less willing to leave the city if she was living there at that stage of her life. This unwillingness is also found amongst women whose husbands were born in the city or who lived there before marriage. A woman of foreign origin or one with a foreign-born husband will also be attracted to metropolitan areas.

58Migration in either direction is more likely the larger the number of sibs, but this effect disappears after the birth of the third child. An increase in educational level raises the likelihood of migration in either direction.

59All these results largely confirm our observations of the interactions between nuptiality and mobility. They explain the effect of the third birth on mobility, which was shown in the nonparametric analysis. Tables 4 and 5 illustrate this effect when the different significant variables are taken into account simultaneously.