Single mothers’ contact frequency with family and non-family members

Existing studies on single mothers’ social contacts often examine small selective samples and are mostly cross-sectional. The lack of high-quality longitudinal survey data on this subject constrains the possibility to draw more generalizable conclusions. This paper exploits panel data to investigate whether transitions to single motherhood affect contact frequency. Fixed-effects models are used on the six waves of the German ‘Socio-economic Panel’ survey that contain social contact data (1990, 1995, 1998, 2003, 2008, 2013) for analyzing whether single motherhood is associated with changes in contact frequency with family and non-family members. Findings show that women transitioning to single motherhood maintain contact with family and non-family members. Single motherhood is unlikely to cause mothers’ structural isolation. However, the absence of a second parent in the household neither seems to be connected with an increase in contact frequency with others. Providing a more generalised account of single mothers’ social contacts over time than previous research, the present study does not find evidence for a disintegrative effect of single motherhood.


Introduction
Scholars and the general public commonly associate single motherhood with more individualised life courses, instable family bonds, economic insecurity and social disintegration (Beck-Gernsheim 2002;McLanahan & Sandefur 1994;Popenoe 1993). Due to their conflicting roles as both sole caregiver and worker, single mothers' difficult economic circumstances have been documented extensively (McLanahan & Percheski 2008;Nieuwenhuis & Maldonado 2018). More empirical research is still needed with regards to single mothers' social relationships. Indeed, previous research has been unable to clarify whether mothers' social contacts increase or decrease in frequency when they become single. However, studying single mothers' social contacts can be a useful case for evaluating the social implications of family change, especially in relation to social integration.
Social relationships are the bedrock of individual social and socio-psychological wellbeing (House, Landis & Umberson 1988), and can impact many areas of social life. For example, research has shown that a higher frequency of social interactions is associated with greater life satisfaction (Powdthavee 2008), with getting a job or maintaining employment (Brandt 2006), and with a lower risk of living in poverty (Gladow & Ray 1986). Previous research considering single mothers' social relationships has mostly focused on qualifying the structures of their social connections. This is crucial for understanding the range and quality of single mothers' social resources.
The main focus of that research was on exposing the heterogeneity in single parents' social relationships, rather than drawing generalizable conclusions (Krüger & Micus 1999). Most of the studies were conducted in the 1980s and 1990s and investigated small selective samples (Leslie & Grady 1985;McLanahan, Wedemeyer & Adelberg 1981;Niepel 1994;Schneider et al. 2001), others were also regionally specific (Keim 2018;Nestmann & Stiehler 1998). One more recent study did use a larger sample for analyzing social relationships of single mothers, but it remained largely descriptive (Fux 2011). The studies highlight differences in patterns of single parents' social relationships, suggesting that social isolation is one among many characteristics of these patterns.
The present study picks up on that observation, aiming at more general conclusions about the impact of single motherhood on the frequency of social contacts by including more recent cohorts of single mothers and by drawing on panel data with larger sample sizes. The focus is solely on mothers, because single fathers commonly face quite different sets of challenges. First, the study adds to the substantive debate on single mothers' position in society by considering social contacts as a dimension of their well-being. Second, the study contributes a novel empirical perspective by using large-scale panel survey data for investigating the associations between single motherhood and social contacts over time. This approach allows for more general conclusions to be drawn on the consequences of transitions to single motherhood than previous research. Third, the study contributes to the larger public and academic debates about whether social disintegration can be considered a consequence of family change. I address two research questions here: 1) Does the frequency of social contacts change after transitioning to single motherhood, and if so, how? 2) Are there differences in this association for contact frequency with family and non-family members? Panel regression analysis is used on data for about 1,200 women surveyed in the German Socio-Economic Panel (SOEP). Drawing on the six waves of the SOEP that include information on social contacts, I am able to capture a long-term window into these women's life courses.
Contrary to expectations about reduction or intensification, the findings show that there is, on average, little lasting change in contact frequency with family and non-family members among women who transition to single motherhood. Social contacts vary across one's life course generally, but there does not seem to be a 'social motherhood effect'. This adds to the findings of existing research, which map social relationships and highlight the strong variety in single mothers' social connectedness. Taken together, concerns about single motherhood being a driver of social disintegration are not warranted. Rather, since mothers maintain a similar degree of connectedness after transitioning to single motherhood, as compared to before, overall social integration is unlikely to be affected.

Social contacts over the life course
The life course perspective is a useful starting point for theorizing about the relationship between single motherhood and social contacts, not least because single motherhood is often a transitory family status. Social relationships, network sizes and frequency of contact with family and non-family members vary over the life course (Lang 2003). A person's overall social network size builds up early in life, stagnates in midlife and decreases in later life (ibid.). Furthermore, different types of social relationships matter to varying degrees over the life course (e.g Umberson, Crosnoe & Reczek 2010). For example, parent-child relationships are particularly important in early life stages, while romantic relationships are especially important in midlife. Life events, such as partnership formation, separation, birth of a child or death of a partner, affect social contacts (Kalmijn 2012;Wrzus et al. 2013). The transition to single motherhood can be defined as the result of such life events -where mother and dependent child or children end up living together with no other adult in the household. 1 Because the transition to single motherhood can occur at any life stage, the age of children will vary among single mothers as will single mothers' own age, and along with that their social and economic characteristics (Zagel & Hübgen 2018).

Social contacts
Previous research has focused on mapping the structures of social relationships during single motherhood, rather than on the consequences of becoming a single mother. It thus remains unclear whether there is a 'single motherhood effect' on social contacts over and 1 The household-based definition is the one that is most often used in quantitative research based on household survey data, but it is also common in qualitative studies. This definition bears fewer measurement problems as compared with others, for example those allowing multiple residences (but see Schier and Hubert 2015).

Structural isolation
Much of the previous research was driven by the concern that single motherhood might lead to social isolation. As a consequential transition in family life, single motherhood indeed exhibits several features that support this concern. There are at least two theoretical mechanisms by which single motherhood increases the level of family stress that can cause a reduction in frequency of personal social contacts: one is through time scarcity and the other is through emotional stress. First, the transition to single motherhood increases stress by reducing mothers' time availability. In contrast to mothers in couples, single mothers do not have the possibility to divide their responsibilities for childcare, work in the home and maintaining the household income with a partner (Nieuwenhuis & Maldonado 2018b). Even though childcare and household work remain far from being equally distributed among partners in couples today (Craig & Mullan 2011), time is particularly scarce for single mothers (Mattingly & Bianchi 2003;Mattingly & Sayer 2006). Social contacts, however, require time investment, which is why we could expect single mothers to have fewer contacts. Second, the transition to single motherhood is likely to increase mothers' psycho-emotional stress. Conflict during separation can be one source of such stress. Accordingly, single mothers often have lower mental health outcomes than partnered mothers, especially if they have experienced a divorce (Afifi, Cox & Enns 2006). As a possible consequence, single mothers may pull back from social relationships (Krüger & Micus 1999;Schneider et al. 2001). Moreover, friends and family members of the separating partner may be involved in conflicts during separation (Sprecher et al. 2006). Social discrimination is another reason for potential emotional stress among single mothers that may cause a reduction of social contacts. Despite the ongoing liberalisation of family norms, single motherhood is often still not socially accepted as a full equivalent to the two-parent family. Partnership constellations and family status types that diverge from the traditional two-parent family are associated with social stigma (Valiquette-Tessier, Vandette & Gosselin 2016), and single mothers commonly face negative stereotypes in their daily lives (Bock 2000;Lauster & Easterbrook 2011). Social discrimination could hence reduce contacts both as the result of mothers' withdrawal or through avoidance by others. In line with this expectation, some studies find that divorced or widowed women have fewer social contacts compared to those in couples (Hurlbert & Acock 1990;Milardo 1987;Sprecher et al. 2006;Wrzus et al. 2013). However, this line of research has limited informative value for the question of single mothers' connectedness, because it includes both mothers and women without children. Some empirical evidence for fewer social contacts of single mothers does exist. For example, Cairney et al. (2003) find that single mothers report, on average, lower frequency of contact with friends and family than partnered mothers.

Connectedness
In contrast to these arguments, the transition to single motherhood might also increase social contacts. Considering research showing that people withdraw from social relationships when becoming a couple (Johnson & Leslie 1982), the reverse behaviour may be true for women transitioning to single motherhood. In particular, mothers may well find strategies to compensate for the absence of a co-resident partner by increasing contact frequency in other types of social relationships (Kalmijn 2012). Emotional and structural support can be found in the interaction with other people, and may compensate for what a partner could cover otherwise (Niepel 1994). From this perspective, single mothers are likely to increase their social connectedness to fulfil emotional and structural support needs (Krüger & Micus 1999). Some evidence exists undergirding the expectation that single mothers' overall frequency of social contacts might increase. For example, Fux (2011) finds that single parents tend to have extensive networks and higher contact frequency with non-family members in order to compensate for the absence of contact with a cohabiting partner. This compensation strategy seems to be an adjustment process developing over the course of the single motherhood episode. Although Fux's study does look at single mothers' social contacts over time using retrospective data, it does not exploit their explanatory potential with longitudinal methods.
Picking up on the second characteristic of social contacts, who is interacting with whom, it seems quite likely that single motherhood does not affect all types of social relationships in the same way. Previous research shows, for example, that most of the fluctuation in social relationships over the life course is due to changes in more peripheral contacts (Lang 2003). Likewise, life events such as childbirth or separation affect the frequency of contacts with friends more strongly than with family members (Wrzus et al. 2013). Transition to first-time parenthood, especially, often reduces the number of friends (Kalmijn 2012), for example due to adaptation to the new parental role or potential estrangement from childless friends. Arguably, close relationships are less affected by life events because the commitment to the relationship is stronger. Family bonds are also often characterized by a particular kinship loyalty. For example, grandparents can be an important source of childcare support for single mothers (Wheelcock & Jones 2002), especially for young single mothers, who may even be living with their parents (see Bryson, Casper & Census 1999;Mutchler & Baker 2009). Accordingly, Leslie and Grady (1985) observe that after divorce, mothers' networks contain more kin than before. Nestmann and Stiehler (1998), on the other hand, point out that single mothers' networks are dominated by female friends. Against the backdrop of this inconclusive evidence, the following analysis will address the question of whether there are differences in the association between single motherhood and contact frequency for family and non-family members.

The case of Germany
Germany provides a particular country setting for analysing family structures and social well-being. Even after 30 years since German reunification, the country hosts two rather different contexts of family life. Family structures differ in that, in former East Germany, single motherhood is more widespread, fewer women remain childless and a greater share of mothers has more than three children than in former West Germany (Goldstein et al. 2010;Statistisches Bundesamt 2010). Single parents made up 28 per cent of all families in Eastern Germany in 2015, compared to 19 per cent in former West Germany (Bundesministerium für Familie, Senioren, Frauen und Jugend 2017). This difference is partly explained by the higher rates of cohabitation and extramarital births, and lower age at first birth in the East. In the West, more traditional family norms prevail (Huinink, Kreyenfeld & Trappe 2012). As in other countries, single motherhood is associated with socioeconomic disadvantages. Single mothers in Germany are at a particularly high risk of poverty (Statistisches Bundesamt 2010). A large proportion of them depend on statutory benefits, and the share is higher for single mothers in the former East than in the West (Achatz et al. 2013).
Less is known about single mothers' social well-being in East and West. What we do know is that norms around social relationships differ between the two German contexts. Historically, family-orientation was high, and extra-familial networks comparatively smaller in East Germany as compared to former West Germany (Uhlendorff 2004). Even today these differences in family norms appear to prevail (Bernardi, Keim & Lippe 2007). There is no straightforward theoretical reason, however, as to why single mothers' social connectedness can still be expected to differ between the former East and West German contexts. In order to account for any possible consequences of the different socio-cultural settings, the following analyses have been adjusted for mothers' residential context, but without further investigation of empirical differences.

Data
Longitudinal investigations of single motherhood put high demands on the data (Ott, Hancioglu & Hartmann 2011). Small sample sizes are a common problem. For the observed individuals, the information gained from quantitative surveys are a valuable approximation for the actual dynamics in partnership trajectories. Analyses of social contacts across family types is even more difficult, because such data are rarely collected at all, and if so it is mostly cross-sectional. I use data from the German Socio-Economic Panel (SOEP, v30) (Goebel et al. 2018) because it provides sufficient sample sizes for longitudi-nal analyses of single motherhood dynamics and social contacts. Data from the first wave of the Gender and Generations Survey (GGS) would be an alternative source, as it covers aspects of the subject matter. The German Family Panel (pairfam) (Huinink et al. 2011) could be a further data source, which includes rich longitudinal information on families' social networks. However, social network data in pairfam are limited to two waves, and there are very few observations of transitions to single motherhood among them. Like in the SOEP, the retrospective biographical data available in pairfam cannot be used, because they do not contain information on social contacts. Robustness checks performed with pairfam based on the prospective data nevertheless did support the substantive findings of the present study. 2 The SOEP is a representative longitudinal survey of demographic and socio-economic characteristics of individuals in Germany. The annual panel started in 1984 in West Germany. It was extended to East Germany in 1990 and includes several refresher probability samples. Social contacts were surveyed in six waves in the SOEP (1990SOEP ( , 1995SOEP ( , 1998SOEP ( , 2003SOEP ( , 2008SOEP ( , 2013, and I use data from all these waves, hereafter referred to as 'social contact waves'. The five-year collection intervals of social contacts data in the SOEP imposes a complex data structure on the analytical sample, because transitions in family life also occur between the 'social contact waves' (see Figure 1). The following analyses hence include indicators further characterizing the family trajectories between these waves. Although the time gaps may seem long in the panel data logic, new social relationships take time to build up and old ones do not just stop from one day to another. So, while not all dynamics occurring in that time window are captured, we benefit from a view into longer-term developments of peoples' social contacts.
The sample was restricted to mothers age 18-59, who were observed living with a child under the age of 18 in the same household in at least one wave during the observation period. Only mothers born after 1945 were retained in the sample to avoid comparisons with individuals raised in wartime. Keeping only those women who were surveyed about their contact frequency with family and non-family members, and considering missing values on all covariates, 5,982 women make up the sample. Due to how fixed effects models operate, the multivariate analyses are based on 1,199 persons experiencing change in contact frequency with family and 1,383 persons experiencing change in contact frequency with non-family members. Among these women, 606 transitions into single motherhood are observed. 2 The pairfam social networks module includes name-generating questions. To delineate personal networks, respondents are asked to state a) with whom they do certain activities, b) with whom they share feelings and thoughts, and c) who helps with information and advice. For the robustness analysis, I used the sum of all persons mentioned by the respondents as a dependent variable as also done by Rözer et al. (2015). Results can be obtained from the author upon request.

Dependent variables
I construct two binary dependent variables: a) contacts with family (including family members and relatives); and b) contacts with non-family members (including neighbours, friends, and acquaintances). In the SOEP, social contacts were measured as part of a battery of items on activities. Respondents were asked: 'Please indicate how often you do the following activities…', giving their response on a five-point ordinal scale: i) never, ii) less than once per month, iii) at least once per month, iv) at least once per week, v) daily. I use the two following items on contact frequency: 'Mutual visits with neighbours, friends or acquaintances' and 'Mutual visits with family members and relatives'. And I collapse the scales into 0 = "less than weekly" and 1 = "weekly or more", assuming that visiting each other weekly or daily is a good proxy for frequent contacts. With this, I argue that people visiting with family or non-family members at least weekly are at a lower risk of structural isolation than those visiting less than weekly. The result of the analyses, however, also holds when changing the threshold for the dichotomisation so that the middle category is included in the '1'. Collapsing the data into a binary variable allows for a less complex presentation of the results at the cost of a loss of information. This means that some nuances in contact patterns go missing, but the binary approach is in line with the aim of testing the connectedness vs. isolation hypotheses. It should be stressed that frequent social contacts do not necessarily indicate that the person receives social support (Schwarzer, Hahn & Schröder 1994). Studying social support is not the focus of this research and would require different data.

Independent variables
The main independent variable single motherhood takes on the value 1 for mothers living with a child aged under 18 who do not have a co-resident partner, and is 0 otherwise. Single motherhood episodes are not considered, if both their start and end occur between two 'social contact waves' (see Example 1 in Figure 1). Single motherhood starting between two 'social contact waves' and lasting through the later wave(s) is counted as a transition (see Example 2 in Figure 1). In the analysis, this strategy gives more weight to the longer episodes of single motherhood in the dataset. This could bias the results, if potential negative effects of single motherhood transitions on the frequency of contact level off over time. Results from additional analyses (not reported) performed on a sample with only short single motherhood episodes showed a small negative effect on contact frequency with family members (but not with non-family members). However, the point estimate does not differ on a statistically significant level from the one obtained in the model on the full sample.

Moderators
Whether someone is repartnered is indicated by a variable that takes on the value 1 for all waves in which someone previously observed as single mother is now living with a partner in the household and is 0 otherwise (including when she goes back to single motherhood 3 ). Women appear as repartnered in the sample even when they repartner between 'social contact waves' and stay repartnered beyond the later wave (Example 2, Figure 1).
The indicator duration of single motherhood counts the survey waves starting with the first observation of single motherhood and remains 0 for all mothers who are never observed as single mothers. The variable is based on information for all survey waves, and can hence take values between 1 and 29 -although the maximum observed is 21 waves. When included in the model, the interpretation of the single motherhood coefficient changes. The single motherhood coefficient then refers to the effect of the initial transition into single motherhood, while the duration coefficient in the model refers to the effect of any additional year in single motherhood. Notes: Data from SOEP (1990,1995,1998,2003,2008,2013), own calculations.

Control variables
The regression models control for socio-economic status (educational attainment level, employment status, household income), the number of children, the age of the youngest child in the household, the age of the mother, and whether the mother lives in East (= 1) or West Germany (= 0). In the fixed-effects framework, the residence variable will produce a coefficient based on the within-person variation. That means, the effects should be interpreted as effects for changes between the East and West German regions. Because this is not of substantial interest to this paper, these coefficients will not be interpreted. Descriptive information on the sample is given in Table 1. The table indicates that in 56 per cent of the person-years, mothers have weekly or more contacts with family members, and a little less often at 51 per cent with non-family members. In 11 per cent of the person-years mothers are single, and they remain so for an average of 5.84 waves. Mothers are on average 35 years old, have one or two children and most are observed with children aged seven and older.

Method
I apply fixed effects logistic regression analysis, exploiting person-level variation in partnership status and contact frequency over time (Allison 2009). 4 Fixed effects regression disregards between-person variation, which means that time-constant characteristics of the individuals are not considered in the models. This is sometimes seen as a particularly efficient estimation strategy for approaching unbiased explanatory estimates. Although the literature highlights that fixed effects regression models may be used for making causal arguments (Brüderl 2010), I avoid strictly causal statements not least due to the complex data structure of the sample in this study. As mentioned above, the gaps between two 'social contact waves' are three to five years (3.8 years on average) in the analytical sample. Nonetheless, using fixed effects is a suitable strategy for answering the research question on the effects of transitioning to single motherhood. 5 While the main analysis uses the binary variable for single motherhood as the main independent variable, complementary analyses (not shown) are done by using the duration of single motherhood instead. This strategy serves as a robustness check for the analyses based on the widely spaced panel data.

Findings
The results of the fixed effects logistic regression models are presented in Table 2. Four models are shown, two for contacts with family members and two for contacts with nonfamily members. Model 1 contains the single motherhood indicator, the repartnering moderator and all controls; and Model 2 additionally adjusts for the duration of single motherhood respectively. The coefficients are reported as odds ratios. That means the effect for single motherhood in the 'Family' models can be interpreted as that, for women who transition to single motherhood, the odds of visiting with family members weekly or more are .86 times lower than for continuously partnered mothers. The effect is not statistically significant. We can hence conclude that, based on the information we gain from our data, single motherhood does not seem to have an effect on contact frequency with family members. The same holds for contact frequency with non-family members. The odds of visiting at least weekly with non-family members are .95 times lower for women in single motherhood compared to those who do not transition, but this effect is not statistically significant either. Including the repartnering indicator, which one might imagine could moderate this relationship, does not change the coefficients for single motherhood in the two models (steps not shown). The effect of repartnering is positive for contact frequency with family members, in that the odds of frequent contact are higher for repartnered mothers, but again the effect is not statistically significant. However, single mothers who repartner have .6 times lower odds of visiting with non-family members weekly or more often. That means they are 40 per cent less likely to have frequent contacts. The results from Models 2, where the duration of single motherhood is added, further support these findings. The estimate for single motherhood changes little and the odds for frequent contact with family members for each additional year of single motherhood are barely different from one (OR .98), neither effect being statistically significant. Likewise, there is no effect on the odds of frequent contacts with non-family members, neither of single motherhood nor of its duration. Considering the control variables, which adjust for some of the differences among single mothers, adds to this picture. Results show slightly different patterns for contact frequency with family and non-family members. Contact with family members seems to be overall more frequent at earlier life stages for the younger age groups of children and mothers. For contact with non-family there is no such clear age-grading (maternal age is no significant predictor); although, mothers with older children do have higher odds of at least weekly visits. There are no clear patterns in the effects of the variables indicating mothers' socio-economic position, but the odds for frequent contact with non-family members are reduced for mothers in households with higher incomes. The coefficient for mothers' residence in this fixed effects model is based on the information of individuals who change their location from the former West to East Germany. The analysis suggests that this increases the odds of contact frequency, but not on a statistically significant level. As with the other controls, including the variable allows us to interpret the other effects net of mothers' residence in the east or west. 6 These findings overall do not support the social isolation hypothesis, nor do they indicate that single motherhood generally leads to greater social connectedness. Mothers seem to find ways to maintain a relatively stable degree of social connectedness when they transition to single motherhood. What makes single mothers vulnerable to social isolation seems to be counterbalanced by what makes them stay connected. Surely, the intimacy of a certain contact or the persons behind the contacts may have changed with the transition (Leslie & Grady 1985;Nestmann & Stiehler 1998). But we do not find indications for shifts between family and non-family contacts when measured in these categories. Overall, the findings of this paper provide tentative evidence for viewing single motherhood as an event with little power over the broader quantity of women's social contacts. While this leaves open several avenues for future research, it is also one step further towards understanding the social implications of single motherhood.  SOEP (1990SOEP ( , 1995SOEP ( , 1998SOEP ( , 2003SOEP ( , 2008SOEP ( , 2013, own calculations. * p<0.05; ** p<0.01; *** p<0.001

Conclusion
Single motherhood is a phenomenon experienced by a growing number of women in affluent societies at some point in their lives. This has alarmed many commentators as to the potential social consequences of this aspect of family change. Previous studies have pointed to the disadvantageous circumstances of single mothers compared to coupled mothers in many areas of social life. They have greater difficulties to reconcile work and family life and fewer economic resources available. Despite a great wealth of research in this area, single mothers' social contacts have received much less attention in the literature. It is however important to understand how single motherhood affects social contacts, because social connectedness generally supports psycho-emotional well-being and hence may be a resource for single mothers. Further, discussions of how family change could affect society routinely include concerns about social integration. With theories allowing for both possibilities, it remains an empirical question whether single motherhood reduces or increases social contact frequency.
Existing empirical research has put a strong focus on mapping different patterns of social relationships among single mothers, rather than aiming to generalize on implications for social connectedness. The cross-sectional approach and the small selective samples underlying previous research have produced a good understanding of the qualities of single mothers' social relationships. Common ground in that research is that single mothers' social networks are diverse and vary in scope and depth. The present study con-tributes to the literature in two ways. First, the study takes into account the consideration of social contacts as a dimension of single mothers' well-being, thereby adding to the substantive debate on their position in society. Second, the use of large-scale panel survey data allows us to draw more generalizable conclusions regarding the consequences of transitions to single motherhood than previous research does. Hence, the results provide some ground to contrast ideological claims about single motherhood being a driver of social disintegration.
The results are sobering in that they suggest there is no negative nor positive effect of the transition to single motherhood on the frequency of contact. Furthermore, this finding is consistent for contact frequency with family as well as with non-family members. Certainly, we cannot conclude from the results that mothers' social networks are unaffected by separation or divorce. There may still be qualitative changes in the shape of the social network or shifts between different types of social relationships. What I do find, however, is that the transition to single motherhood does not lead to structural isolation per se. This finding can be seen in opposition to concerns about social disintegration being a consequence of family change. Findings suggest that reduction and increases in social contacts with family and non-family members appear to occur at different stages over the course of one's life, but single motherhood does not have an additional effect. The results fit with the findings from small-scale cross-sectional studies, which demonstrate heterogeneity in single mothers' social relationships but do not allow for generalization. Single mothers are a heterogeneous social group whose members are integrated in various social network contexts. As such and in tandem with the present findings, it would be misleading to assume a disintegrative role of changes in family structures that involve single motherhood.
Despite the study's value for the debate around single motherhood, it also points to issues that we do not know and will not find out based on the data that is available to date. Information on single mothers' social relationships over time are scarce in large scale surveys. In the case of the present study, the complex structure of the available data on social contacts restrains its explanatory power. Especially the long intervals between the panel observations call for caution in drawing causal conclusions. In summary, the study is both a reminder of single mothers' capabilities to stay socially connected as well as a sign of the need for richer data collection in order to analyse family dynamics and social relationships over time.