1 – Times of isolation for respondents in childhood and adolescence

10This restricted family network of Île-de-France residents born between 1930 and 1950 comprised an average of four people during childhood, rose to more than six at ages 30-50, then gradually fell back to fewer than five people as the respondent reached the age of 70 (Golaz and Lelièvre, 2012). On average, over the whole life course, the respondent’s siblings and children were the most numerous members of the restricted family network.

11During childhood and adolescence, parents and siblings dominated, although in the second period spouses and children began to appear in some respondents’ lives (Figure 1).

Figure 1

Figure 1

Average restricted family network of young persons resident in Île-de-France and born between 1930 and 1950, during childhood and adolescence

12The respondent’s situation was determined by cross-linking data on the presence (co-residence or living nearby) of the people concerned each year. Reconstructing a contact circle in this way does not depend on direct description; rather, the reconstruction process is based on the cross-linking of precise factual data, allowing maximum comparability between respondents. This reveals periods of relative isolation during childhood or adolescence, when the respondents were in contact with one person in the immediate family at most, but were not necessarily living with them. The criterion applied thus depends on a de facto situation: being isolated for at least one year of childhood/adolescence. The case of Odette is an example of a life course that included a period of isolation (Figure 2).

Figure 2

Figure 2

Odette’s restricted family network during childhood and adolescence

13During her childhood, from the ages of 2 to 8, Odette did not live or have regular contact with anyone in her family. At that time, Odette lived in an orphanage, where she had been placed after her father went to prison; she was cut off from her mother and separated from her two brothers so she no longer had any contact with her family members. For Odette, this period was “difficult”, but was followed by some “very good years” when, at age 9, she entered a foster home where she was reunited with her brothers and where an unrelated person played a strong parental role. One of her brothers died young. In the meantime, she gained a vocational qualification in hairdressing and started working. At the end of her adolescence, she got married.

14Over the whole sample, 5.9% of individuals, on average, were affected by a period of relative isolation at each age (period of one year minimum). This proportion decreases to under 3% at ages 5-9. It is reasonable to assume that for a child (in this study, ages 0-12), having a contact circle limited to one person at most – who was not even necessarily co-resident – indicates difficult circumstances. In adolescence (in this study, ages 13-21), the proportion of years of isolation increases again. From age 15, there is a big difference between men (9%) and women (6%), in part because young women form unions earlier, which extends their contact circle, and because young men relatively more frequently leave to take up an apprenticeship, which separates them from their immediate family (Figure 3).

Figure 3

Figure 3

Proportion of Île-de-France residents born between 1930 and 1950 with small family circles during childhood and adolescence
Note: Respondents with small family circles in childhood and adolescence were those who reported being in contact with one person at most from their immediate family at these ages.

15Cases like Odette’s – where the close family circle shrank to the point of disappearing for several years of childhood – were not very frequent. These periods during which the family was totally absent represent only 17% of situations of childhood isolation. Most of the time, at least one person from the immediate family remained in contact with the respondent. In more than three quarters of such situations, the close family circle consisted of one parent (60%) or one parent figure (16%). By contrast, in only 40% of situations of adolescent isolation were the individuals in contact with a parent or parent figure; total isolation was more frequent in adolescence (27%) than in childhood (Table 2).

Table 2

Distribution (%) of periods of isolation by composition of close family circle

Period starting in Significance of differences between childhood and adolescence childhood adolescence Number of periods of isolation 258 282 Presence of or contact with one parent (father or mother) 160 135 *** another parent figure 116 115 *** a sibling 117 120 *** a spouse – 112 *** No restricted family network 117 127 ** Total 100 100

Distribution (%) of periods of isolation by composition of close family circle

Significance levels: * 5%; ** 1%; *** 0.1%.

2 – Contact circle outside the family during periods of isolation

16The Biographies et Entourage survey respondents were also asked to mention anyone who had played an important role – positive or negative – in their life, apart from people in their restricted family network. These were retrospective judgements, dated according to major life periods, including childhood and adolescence, and made with some 40 years’ hindsight. Over the respondents’ lifetimes, it was friends who dominated the social network outside the immediate family. When the presence of such people during childhood was mentioned, they were mainly more distant relatives (62%), who were generally female and older (Lelièvre et al., 2008). This effect diminished over the life course: in adolescence, relatives represented only 39% of the people mentioned outside the restricted family network.

17For respondents who had experienced a period of relative isolation from family during childhood or adolescence (i.e. contact with at most one member of their immediate family for more than one year), there is a lesser propensity to mention people they were close to during these periods. In fact, more than half (53%) of those who had experienced such periods during childhood did not mention any other important person in their contact circle at those times, versus only 48% of people who did not have this experience. In adolescence, the proportions were 55% and 49%, respectively. Thus, isolation from the family seems to go hand in hand with the memory of more general isolation, and the difference between perceptions of childhood and perceptions of adolescence is slight.

18Remaining in the area of perceptions, we will now look at the division of the respondents’ lives into major periods. We will assess the tone of these periods and analyse the comments made by respondents in the Biographies et Entourage survey to justify their opinions.

1 – Periods of childhood and adolescence perceived as “difficult”

21Respondents divided their life courses into between one and twelve periods. Overall, the dominant tone over the lifetime was one of good years (G = 33%). The most frequent “judgement trajectories” were G, GDG, DG and GD, with alternating periods of good years (G) and difficult years (D) (Laborde et al., 2007). Negative judgements represented slightly over a quarter of the periods discussed. However, the average length of periods varied according to the type of judgement. The most difficult periods tended to be shorter than the average.

22Some 15-16% of childhood years were judged to be difficult and only 5-6% very difficult (Table 3). There were more problem years during adolescence than during childhood, with about a quarter of years reported as difficult or very difficult. While judgements relating to childhood were very similar for both genders, differences were perceptible at adolescence. The proportion of difficult years was higher for adolescent girls (26.9%) than for adolescent boys (23.7%), and these differences are significant. The same trend is observed for years reported to be very difficult, which are more frequent for young women. If we compare the difficulties experienced in childhood and adolescence for each gender, women more often perceived their adolescence as difficult or very difficult. For men, these differences were smaller; their childhood seems to be the period during which they had the most “very difficult” times.

Table 3

Proportion of years of life judged as “difficult” or “very difficult”, by gender and life period

Number of years of life Proportion of difficult or very difficult years (%) Significance of differences Proportion of very difficult years (%) Significance of differences Between boys and girls Between childhood and adolescence Between boys and girls Between childhood and adolescence Childhood Girls (N=1,487) 19,331 21.6 ns girls *** boys *** 5.3 *** girls *** boys *** Boys (N=1,343) 17,459 21.1 6.0 Adolescence Girls (N=1,487) 10,409 26.9 ** 7.0 *** Boys (N=1,343) 09,401 23.7 5.5 Lifetime 20.2 5.3

Proportion of years of life judged as “difficult” or “very difficult”, by gender and life period

Significance levels: * 5%; ** 1%; *** 1%; ns: not significant.

2 – Word association of terms relating to isolation and difficulties experienced

23Going beyond a summary assessment of successive periods in respondents’ life histories, the questionnaire also recorded a description of each period (Questions Sy 1 and Sy 2, Box 1). This enabled us to conduct a textual analysis linking the words used by respondents in describing the periods of their childhood and adolescence to their judgement of these periods. For this, the text of the respondents’ comments was first lemmatized. [2] The lexis associated with discourse on childhood (or on adolescence) was then reduced by grouping words with the same root and eliminating connectives. An aggregated lexical table was then constructed, matching these words to the tone of the judgements and to the sequence of tones that characterized the period. Several qualifiers may follow one another through a period: for example, D then G – a difficult life course that improved – or G – a period of good years. We then used this contingency table to make a multidimensional descriptive analysis of the words obtained, by means of correspondence analysis (CA). [3]

A childhood dominated by family, school and war

24Table 4 (reproduced here only for the period of childhood) shows the words most frequently used to describe the period. Three main themes stand out in descriptions of childhood: family (notably parents), war and school. This period was mostly described as happy, as a carefree period, although some people found it difficult and reported problems. Despite some disparities, the comments on childhood show many similarities, at least in relation to these three main themes.

Table 4

List of words used for the period of childhood, in order of frequency

Words used Frequencies Happy + 438 Family + 355 Parent + 353 Good + 230 Father 201 Good 201 Difficult + 196 Problem + 180 War + 176 Mother 164 School 119 Study + 108 Carefree + 173

List of words used for the period of childhood, in order of frequency

Note: The + sign means that variant forms of this root were included when the number of uses was calculated.

25In addition, for each respondent’s childhood, the sequence of tones in the judgements was coded in the form of a 6-category variable (D, G, PF, DG, GD and Others). We then constructed the aggregated lexical table by matching each of these types of period to the vocabulary of the discourse on that same period, on which we performed a correspondence analysis. The words in the description of childhood provided by the respondents create a map onto which we can project the types of childhood period in order to study the distribution of vocabulary in given subpopulations (Garnier and Guérin, 2010).

26In Figure 4, childhood trajectories are mapped onto the specific discourse of childhood, and this provides an answer to the question: what is the vocabulary associated with a given life course in childhood? (Garnier and Guérin, 2010). While our preliminary textual analysis of the whole corpus – which by definition related to the whole lifetime – revealed a strong distinction between male and female trajectories (results not shown here), we did not, for childhood, distinguish individuals’ trajectories by gender, since for that period – and only for that one – the type of discourse adopted was not marked by gender differences. There are three distinct focal areas in Figure 4:

• a difficult childhood (north-east on the map);
• a good or problem-free childhood (west on the map);
• a childhood made up of contrasted periods (DG, GD or Others) (south-east on the map).

Figure 4

Figure 4

First factorial map derived from analysing correspondences on the lexical table of terms used to describe childhood, associated with life-course characteristics

27The strongest contribution to Axis 1 (51%) corresponds to difficult reported trajectories. Individuals who described their childhood as difficult were more likely to mention themes related to the material conditions of day-to-day life (“money”, “poverty”, “lack”) and to “work” than other respondents; they put more emphasis on the fact that this was “difficult” or “hard”; they used the adjectives “bad” and “unhappy”. In these difficult periods, family relationships were mentioned, as was their absence: “father”, “mother”, “death”.

28Then came trajectories reported as good and problem-free (37%). Those who remembered a problem-free, even good, childhood, talked about their “family” in the broad sense, about “education”, about feeling “carefree”, about “holidays” and “friends”. They tended to use the following adjectives: “caring”, “quiet”, “nice”, “happy”.

29Respondents who reported that their childhood had been more changeable (major contribution to Axis 2 – 46%), had improved (from difficult to good – DG) or, conversely, had deteriorated (GD) covered ambivalent themes. They talked about “school” and “boarding school”, “freedom”, “war” and “grandparents”. Thus, the terms reflect contradictory perceptions: some people felt that going to boarding school was synonymous with freedom, while for others this separation from home was a very bad experience; similarly, “grandparents” might reflect absence of parents or support from grandparents.

Emerging gender differences in adolescence

30For the period of adolescence, the frequency of the words used (table not shown here) again reveals the favoured position of the theme of family, with mention not only of “parents” but also of “brothers” and “sisters”. Siblings seem to become more important during this period. Setting aside “war” – still present here, as was “Algeria” – two threads can be distinguished: “college” and “work”. Some people mention “work”, “job” or “military service”, while others continue to talk about their education (“study”, “boarding school”). For others again, it is already a matter of “marriage”, “couple” and “children”. As in childhood, the adjectives “happy”, “hard”, and “nice” are found. Adolescence is also associated with the themes of “freedom”, “independence”, “discovery” and “holidays”.

31Comments made by respondents about adolescence appear to be strongly gendered. To take this into account, we performed an analysis that included not only the tone of the period, but also the respondent’s gender. This time we constructed a 12-category variable. On the map shown in Figure 5, the tonal variables are qualified: F_D means female adolescence reported as difficult, M_G, male adolescence judged to be good. Distribution of the adjectives in Figure 5 clearly highlights the gendered nature of the respondents’ life courses: all the adjectives on the women’s trajectories are to the left of the vertical axis and all those used by men are to the right (Figure 5). In terms of contribution to the axes, female and male trajectories are balanced on either side of Axis 2. The contribution to Axis 1 is greater for M_G (35%) than for F_D (22%); in contrast, F_G trajectories contribute 33% to Axis 2 and M_D, 19%.

Figure 5

Figure 5

First factorial map derived from the correspondence analysis on the lexical table of terms used to describe adolescence, associated with combined characteristics of periods and respondent’s gender

32From adolescence onwards, women in these birth cohorts have a greater tendency to focus on the home, while men focus on work.

33The men who had gone through a difficult period in adolescence covered the themes of “boarding school”, “war” and “work”. Women who described this period as difficult mentioned their “mother” and “illness”. For trajectories that improved (DG), perceptions and vocabularies seem to differ between men and women.

34For those who gave a more positive account of their adolescence (Figure 5, north on the map), regardless of gender, the themes of “meeting” and “discovery” appear. For young women only, the terms “marriage” and “family” were strongly associated when talking about events outside the home environment. Men in the same type of trajectory mentioned “college”, “friends” or feeling “carefree”.

35This textual analysis clearly shows that the contact circle – its absence as much as its presence – played a large part in descriptions of periods of childhood and adolescence. Nevertheless, its role also appears ambivalent. Parents and siblings were mentioned very frequently in both good and difficult periods, indicating that they were cited for their presence and their support as well as for their absence. Girls and boys – or rather, the women and men they have become – have different memories of their adolescence – the period when gender distinctions emerge.

1 – Presence of parent figures at difficult times

38To examine this first observation in more detail, we used logistic regression to analyse the effect of the characteristics of the respondent’s restricted family network on judgements of the periods of childhood and adolescence. In order to do this, each period identified by the respondents constitutes a statistical individual. The dependent variable was the fact of considering the period as “difficult” or “very difficult”. Respondent-related control variables (birth cohort/place of birth) and control variables relating to distinctive characteristics of the period (length) and of the contact circle (size and composition of restricted family network) were introduced. In all cases, period length and judgement of the period were correlated, with difficult periods being generally shorter than others. Periods beginning during childhood and completed in adulthood were not taken into account in this work, which leads to slight over-representation of difficult periods (the shortest) by comparison with periods that were judged positively (the longest). Nevertheless, we assume that this has a negligible influence on this first model.

39During childhood and adolescence (Table 6), birth cohort was significant only for boys. Overall, boys born in the 1940s experienced happier periods than boys born in the 1930s. This probably reflects the effect of the war and of the subsequent period of reconstruction. Men born in Île-de-France judged their childhood and adolescence more positively than those born in the provinces or abroad. Moving away from the place of birth showed a significant association with negative assessment of the period, although only for men. This suggests three hypotheses: either the episode of migration or the arrival in Île-de-France was a bad experience, or the migration came at the end of a difficult period, or men tended to retrospectively devalue the part of their childhood spent away from the Île-de-France region where they were living at the time of the survey.

40In contrast, parents’ occupations (results not shown here) had no significant effect on judgements of the period: regardless of their father’s or their mother’s occupational status during their childhood and adolescence, men and women assessed these periods in similar ways. Before age 21, as we have already seen, the immediate family mainly comprised parents, other parent figures and siblings. The presence or absence of siblings does not seem to play any role in judgements of the period (results not shown). People with a large number of siblings did not report their childhood and adolescence as having been more or less happy than those who were the only child in the family. Neither the average size of the immediate family nor the general trend in its size over the period had any significant effect on the respondent’s judgement (results not shown). We therefore constructed two variables describing the family network, one relating to the presence of parents and the other to the presence of parent figures. Women’s judgement of their childhood and adolescence was sensitive only to the presence or absence of parents in the network (periods spent with both parents were significantly happier). Being separated not only from parents but also from all other members of the immediate family also led to an increase in negative judgements. On the other hand, the presence of other parent figures did not play any significant role for girls. Among boys, we find the same influence of the presence of parents, but also a significant relationship with the presence of parent figures. The difficult periods experienced by men during childhood were associated with the presence of parent figures who had helped them to overcome their difficulties. This analysis highlights the importance of the presence of both parents, both in the contact circle and in the household, as well as the significant presence of other parent figures during periods judged to be more difficult than others, in particular for boys.

41Finally, regarding the contact circle outside the family network, 62% of respondents who judged at least one year of their childhood to be “very difficult” did not report any important person in their life during that period. This was the case for 53% of respondents who judged at least one year of childhood to be “difficult” and only 49% of those who did not report any difficult period in childhood. In adolescence, the trend was the same: 57% of respondents who had judged at least one year to be “difficult” or “very difficult” did not report any important person, versus 51% of those who reported no periods of difficulty.

42In order to take the analysis further, some additional characteristics of individual trajectories, such as death of a parent or place of residence, would have to be entered into the earlier logistic model. But because of the personal way in which different individuals divided up their life course trajectories, and because of the variable size of the periods that resulted, this cannot be done without averaging over the periods or making choices that would bias or impoverish the analysis considerably: for the death of a parent, the same category would have to include periods of one year centred on the parent’s death and much longer periods where the death was not necessarily the sole determinant of the overall tone. In the Biographies et Entourage survey, since data were mainly recorded by year, the year was the greatest common divisor of periods (periods which were difficult or not, periods in work, periods of residence, etc.). By analysing situations at the year level, we can evaluate even more fully the relative impact of factors that contribute to creating a bleaker tone for a given period.

2 – Family separation: a factor in judging certain years as “difficult”

43To model the probability of a year of childhood or adolescence being reported as “difficult” or “very difficult”, we first used a simple logistic model (Table 7, Model 3). However, because of the hierarchical structuring of the data – already highlighted for the preliminary analysis by period (Table 6) and appearing even more strongly at the level of years of life (21 years per individual, grouping childhood and adolescence, with characteristics of the year and characteristics of the individual) – a multilevel model becomes necessary (Table 7, Models 4, 5 and 6). The simple model matches characteristics at the “individual” level to the “year” level, so the estimation of their coefficients is incorrect. Yet what interested us here was to distinguish not so much what differentiates one individual from another (we are working on perceptions – and, what is more, retrospective ones – so we can expect large differences between individuals that are difficult to explain), but rather the elements within individual trajectories that influence the tone of one year in comparison to the tone of another year. The multilevel approach, by dissociating calculations at each “individual” and “year” level, enables us to handle this data structuring correctly and thus to answer our question.

Table 6

Relative probabilities (odds ratios) of experiencing a difficult period in childhood and adolescence, by characteristics of ego and of his/her restricted family network

Model 1 Model 2 Men Women Characteristics of ego Ego’s year of birth 1930-1935 Ref. Ref. 1936-1940 – 0.27 ns 0.12 ns 1941-1945 – 0.49 ** – 0.06 ns 1946-1950 – 0.42 * – 0.07 ns Place of birth Abroad or French overseas departments and territories Ref. Ref. Île-de-France – 0.50 ** – 0.01 ns Provinces – 0.35 * – 0.00 ns Characteristics of period Length of period Short Ref. Ref. Medium – 0.61 *** – 0.77 *** Long – 0.99 *** – 1.16 *** Characteristics of restricted family network Size of restricted family network over the period Down to 0 for at least one year Ref. Ref. Always greater than 0 – 0.09 ns – 0.67 * Number of parents present At least one parent, but not two for whole period Ref. Ref. No parent for at least one year – 0.20 ns – 0.32 ns 2 parents for whole period – 0.39 ** – 0.38 ** Other parent figures At least one other parent figure Ref. Ref. No other parent figure – 0.50 *** – 0.19 ns

Relative probabilities (odds ratios) of experiencing a difficult period in childhood and adolescence, by characteristics of ego and of his/her restricted family network

Interpretation: Men who mentioned at least one other parent figure described their childhood significantly more favourably than others did, all other things being equal (the odds ratio cannot be interpreted as a relative risk).
Significance levels: * 5%; ** 1%; *** 0.1%; ns: not significant.

44In Models 4 to 6, the variance between individuals is very strong and significant (11.2; 11.6; 10.9), as we had expected (Table 7). Taking the hierarchical data structure into account alters the relative weight of some of the explanatory variables in these models. In the logistic model, gender and cohort were significant; in the multilevel models, they were no longer so. Characteristics relating to parents’ occupations also became less significant. The relative balance of the explanatory variables nevertheless differs slightly by gender and, in the light of our preceding observations, different models for each gender yield results that are interesting to compare. The following comments therefore relate to the results of the multilevel models – those which best answer our questions.

Table 7

Effect of characteristics on reporting of difficult or very difficult years in childhood and adolescence

Model 3 Logistic Multilevel models Model 4 Model 5 Model 6 Overall Overall Girls Boys Coef. Coef. Coef. Coef. Constant – 1.45 *** – 2.20 *** – 1.46 * – 2.71 *** YEAR level Year of an isolation phase lasting at least two years Ref. Ref. Ref. Ref. Normal year – 0.53 *** – 0.47 ** – 0.46 ** – 0.52 *** Death of one parent in adjacent years [y-1, y, y+1] Ref. Ref. Ref. Ref. No death of close kin – 0.72 *** – 0.66 *** – 0.53 *** – 0.78 *** War (1939-1945) War year Ref. Ref. Ref. Ref. Other year – 0.21 *** – 0.65 *** – 0.59 *** – 0.73 *** Period of life Adolescence (ages 13-20) Ref. Ref. Ref. Ref. Childhood (ages 6-12) – 0.15 *** – 0.52 *** – 0.68 *** – 0.34 *** Early childhood (ages 0-5) – 0.36 *** – 1.10 *** – 1.41 *** – 0.75 *** Place of residence Île-de-France Ref. Ref. Ref. Ref. Provinces – 0.06 * – 0.07 ns 0.06 ns – 0.19 ns Abroad – 0.34 *** – 1.17 *** – 1.06 *** – 1.29 *** INDIVIDUAL level Gender Female Ref. Ref. Male – 0.11 *** – 0.24 ns Cohort 1930-35 Ref. Ref. Ref. Ref. 1936-40 – 0.04 ns – 0.04 ns 0.17 ns – 0.13 ns 1941-45 – 0.17 *** – 0.29 ns 0.05 ns – 0.69 ns 1946-50 – 0.07 * – 0.08 ns 0.07 ns – 0.24 ns Place of birth Abroad Ref. Ref. Ref. Ref. Provinces – 0.35 *** – 1.12 *** – 0.76 ** – 1.43 *** Île-de-France – 0.54 *** – 1.42 *** – 0.81 * – 2.00 *** Father’s occupation Economically inactive Ref. Ref. Ref. Ref. Farmer – 0.14 ** – 0.41 ns – 0.64 ns – 0.27 ns Self-employed – business owner – 0.35 *** – 0.68 * – 0.99 * – 0.35 ns Intermediate occupation – manager – 0.53 *** – 1.07 *** – 0.96 * – 1.24 ** Clerical worker – 0.55 *** – 1.09 *** – 1.11 ** – 1.06 * Manual worker – 0.25 *** – 0.60 * – 0.61 ns – 0.56 ns Mother’s occupation Manual worker Ref. Ref. Ref. Ref. Clerical worker – 0.35 *** – 0.70 * – 0.40 ns – 1.13 * Other employment – 0.42 *** – 0.72 * – 0.45 ns – 1.02 * No paid work – 0.30 *** – 0.56 * – 0.47 ns – 0.69 ns Variance of the constant 11.16 *** 11.56 *** 10.91 ***

Effect of characteristics on reporting of difficult or very difficult years in childhood and adolescence

Note: For each individual, 21 successive years are included in the models. Here, the significant non-zero variances of the constant confirm the validity of the multilevel estimate.
Significance levels: * 5%; ** 1%; *** 0.1%; ns: not significant.

45The factors that distinguish individuals are their parents’ occupations and their place of birth. Birth cohort (and gender, Model 4) do not introduce significant differences into the judgements given for childhood and adolescent years. For all cohorts combined, the individuals in our sample who were born abroad, had an inactive father and a mother in a manual occupation were those who gave the most sombre assessments of their childhood and adolescence. The most positive assessments came from individuals born in Île-de-France, with a father in a management, intermediate or clerical occupation and a mother who was working but not a manual worker.

46Within individual histories, the factors that distinguish one year from another are characteristics of the restricted family network (objectively isolated; a year close to the death of a parent), the historical context (Second World War), the period of life in question (adolescence) and finally the individual’s place of residence recorded for each year of his/her life.

47Place of birth and place of residence during the year under consideration have opposite effects: being born abroad is associated with difficult childhood and adolescent years, whereas living abroad is a positive factor. This allows us, for example, to distinguish people born abroad who lived in France during their childhood and adolescence (who combine more factors associated with reporting difficult years) from those living abroad. This result is particularly marked for boys. In addition, for both genders, the adolescent years are significantly more difficult than the childhood years, and early childhood (ages 0-5) is less difficult than the later childhood years (ages 6-12). War and the death of a parent are very negative factors, with a stronger influence on the tone of the life course for boys than for girls. Here we should recall that respondents were aged 15 at most in 1945. The boys lived through the war in similar conditions to the girls, but they were more marked by their experience.

48In all the models, there is a strong link between isolation from family and reported difficulty. Relative isolation is accompanied by a negative perception of that year. However, it is not the sole explanation for a given year having a negative tone, since other factors are involved, such as war, being uprooted, or social background. While this analysis confirms the existence of a clear link between being separated from members of one’s close family and a negative assessment of the period, it also puts this finding in perspective.