Introduction

Research on immigrant families often has suggested that the process of immigration can lead to changes in family hierarchy (Portes 1997; Baptiste 1990), and to a distancing (Hwang 2006) or even alienation (Fillmore 2000) between adolescents and their parents. One assumption in the research on such family processes is that adolescents and parents drift away from one another because the adolescents adjust more easily to the new society (Cheung et al. 2011; Birman 2006; Telzer 2011) and start to assume familial roles usually reserved for adults (Titzmann 2012; Morales and Hanson 2005); this results in parents and adolescents becoming estranged. In other words, it is often implicitly assumed that the intergenerational communication in immigrant families is less functional than in families without an immigrant background. However, the actual degree of adolescent-parent agreement in parent–child communication has rarely been investigated. This study fills this research gap by studying mother–adolescent agreement with regard to adolescents’ disclosure, which is characterized by adolescents’ sharing information with their parents. Specifically, we were interested in whether mothers and adolescents in native families are more similar in their perceptions of the adolescents’ disclosure about their whereabouts and activities compared to those from immigrant families. The second question aimed to investigate which variables best predict interdyadic differences in mother–adolescent agreement concerning children’s disclosure: normative age-related variables or, among immigrants, variables related to their advances in sociocultural adjustment. The third research question focused on the associations between mother–adolescent agreement and psychosocial outcomes.

Mother–Adolescent Agreement on Child Disclosure

In recent decades, research has shown that parental knowledge concerning their children’s whereabouts is an essential asset in avoiding negative developmental outcomes, because parents with greater knowledge can prevent their offspring from associating with deviant peers, can reinforce positive and punish deviant behavior, and may provide the specific support their child needs (Kerr and Stattin 2000; Dishion and McMahon 1998). For many decades, researchers assumed that parental knowledge was based on parents’ controlling their children and thus the term “monitoring” was used to describe parental knowledge. Recent evidence suggests, however, that parents in fact acquire their knowledge through their children’s disclosure (Kerr and Stattin 2000; Kerr et al. 2010). High levels of children’s disclosure represent a form of parent–child communication characterized by children voluntarily sharing information with their parents, and is assumed to be the result of adolescents’ “trust in their parents—whether they feel that their parents are willing to listen to them, are responsive, and would not ridicule or punish if they confided in them” (Stattin and Kerr 2000, p. 1083). The level of adolescents’ disclosure was found to differ between various domains. Adolescents disclose more about school-related topics and less about peers or personal issues, and they disclose more information to mothers than to fathers (Smetana et al. 2006). We used probably the most renowned conceptualization of children’s disclosure—that developed by Kerr and Stattin (2000). This measure covers mother–adolescent agreement in the domains of school and leisure activities. Using this measure in longitudinal research revealed a decrease in children’s disclosure over a 2 years period, most likely because adolescents are granted more freedom and autonomy from parents (Kerr et al. 2010). In sum, parental knowledge is mainly gained from their offspring’s disclosure, but it remains unclear whether mothers and adolescents would agree on the amount of information adolescents disclose to their parents.

A direct transfer of information from adolescents to mothers is not necessarily the norm. Mothers may think that their children are sharing all relevant information, but the children may in fact keep crucial facts to themselves. Similarly, children may disclose their whereabouts and activities, but their mothers may be inattentive. This raises the question of how similar mothers’ and adolescents’ reports of children’s disclosures actually are. A look at parent–adolescent correlations suggests substantial variability between children’s and parents’ views on children’s disclosure. In Kerr and Stattin’s (2000) study, there was only about 20 % shared variance between adolescents’ and parents’ reports of children’s disclosure (r = .45). In other studies, the correlation was even lower: r = .23 or 5 % shared variance (De Los Reyes et al. 2008) and r = .21 or 4 % shared variance (Laird et al. 2010). A potential explanation for the difference between children’s disclosure reported by mothers and adolescents is related to the nature of the samples. Kerr and Stattin (2000) sampled adolescents in a small town in Sweden, whereas De Los Reyes et al. (2008) as well as Laird et al. (2010) included a substantial proportion of minority adolescents in their US samples. Variability in parent–adolescent agreement concerning children’s disclosure can therefore be seen as an issue for further research.

The overall aim of this study was to investigate interfamilial variation in mother–adolescent agreement concerning the perception or reporting of adolescents’ disclosure. High mother–adolescent agreement concerning adolescents’ disclosure (independent of the actual degree of child disclosure) represents an index of how well mothers can judge whether or not their child does disclose where they were or what they were doing. As long as mothers and adolescents are on the same page (either high or low in the disclosure ratings), we consider the communication processes balanced. Problematic patterns emerge when the appraisals differ substantially. Adolescents might start lying about their whereabouts and activities or mothers might become over controlling and may perceive the child as becoming more uncommunicative. Children’s disclosure is not a typical outcome in acculturation research, but it can be assumed to be affected by immigration because research has shown that the parent–child relationship can change with the transition to a new country (Oznobishin and Kurman 2009; Portes 1997; Titzmann 2012).

Our research focused on mother–adolescent dyads because immigrant as well as nonimmigrant mothers were found to spend more time with adolescents than fathers and they often have deeper insights into the lives of their offspring (Dubas and Gerris 2002; Updegraff et al. 2009). Furthermore, in a recent study, adolescents reported on maternal and paternal warmth over a 5-year period (Rodríguez et al. 2013). The results indicated that paternal warmth was reported to decrease, whereas maternal warmth remained stable over time. These changes were independent of the immigrant status of the families. This can be seen as an indication that the mother–child relatedness is less affected through the adolescent years than the father–child relatedness and that mothers, therefore, may be better at judging whether or not their children are disclosing information.

Ethnic Group Differences in Mother–Adolescent Agreement

Our first aim was to predict group-level differences in child-disclosure agreement between immigrant and native mother–adolescent dyads. Various arguments suggest that mother–adolescent agreement is greater among native families than immigrant families. First of all, Berry et al. (2002) assume that individual psychological outcomes of human development are the result of enculturation and socialization derived from various social partners, particularly from parents, peers, and relatives. Natives usually receive congruent information concerning norms of conduct and values from these different sources, because most often all these contacts have grown up in the same cultural environment. Immigrant adolescents, however, are often confronted with different values and behavioral norms because their parents mostly received their cultural imprint in the heritage culture, whereas teachers and peers are embedded in the host culture (Oppedal 2006). Adolescent immigrants thus have to deal with two different working models, which might increase the likelihood of intergenerational dissimilarities. Whereas immigrant mothers may have the feeling that their children do tell them what happens in school, the adolescents may be hesitant about doing so because their behavior at school, for example, may not conform completely to the mothers’ values. A second argument relates to additional differences between mothers and adolescents due to different paces of adaptation (Cheung et al. 2011). Usually, adolescents adapt more rapidly than their parents to a new cultural environment, which results in an acculturative gap (Birman 2006; Telzer 2011) and in children taking over family responsibilities, such as translating documents for their parents (language brokering) (Morales and Hanson 2005; Weisskirch 2005), or assuming adult roles (parentification) (Titzmann 2012). This differential adaptation is assumed to bear the risk of intergenerational alienation (Fillmore 2000) or acculturative family distancing (Hwang 2006; Hwang et al. 2010). There is no comparable risk in native families. These arguments point to the fact that agreement between mothers and their adolescent children can be expected to be higher in native mother–adolescent dyads than in immigrant dyads.

Interdyadic Differences in Mother–Adolescent Agreement

Although group-level differences are important in gaining an initial understanding of relationships in immigrant and native families, such differences can only provide limited insights into the mechanisms of the family dynamics responsible for interdyadic differences in mother–adolescent agreement. First and most importantly, child-disclosure agreement between mothers and adolescents needs to be seen against the backdrop of normative development in adolescence, independent of immigrant or native background. In particular, growing adolescent autonomy may have an effect on mother–adolescent agreement with regard to children’s disclosure. In general, high levels of adolescent autonomy can be expected to reduce mother–adolescent agreement because adolescents gain independence and distance themselves from their parents; this has been found, for example, to relate to greater generational differences in the perception of family climate (Seiffge-Krenke 1997). In addition, achieving autonomy includes adopting less socially acceptable behaviors, such as smoking or drinking, which can cause parental concern because such behaviors may harm their offspring and typically elude their supervision (Titzmann and Silbereisen 2012). It can be assumed that it is more difficult for parents to judge their adolescents’ disclosures when those adolescents gain greater autonomy because parents may not realize what their children are or are not doing with friends or in their spare time. Furthermore, it may be difficult for parents to recognize when changes in their children’s disclosure start to occur.

These arguments apply to native and immigrant families alike. For immigrants, however, additional parent–child dynamics have to be considered. We have already indicated that immigrant adolescents seem to adjust to a new context more quickly than their parents (Cheung et al. 2011), which results in an acculturation gap (Birman 2006; Telzer 2011) and in adolescents’ taking over responsibilities. These responsibilities are often related to language brokering activities (Morales and Hanson 2005; Weisskirch 2005), which means that adolescents translate documents, bills, and so forth for their parents (Weisskirch 2005), a behavior that is exhibited by approximately 90 % of immigrant adolescents, but with substantial interindividual variation (Jones and Trickett 2005; Morales and Hanson 2005). Adolescents who take on such translation responsibilities more frequently than others can be expected to distance themselves more from their parents because they become aware of the cultural differences and the different rates in adaptation between themselves and their parents, and may also realize that their parents need help rather than being providers of it (Hwang 2006; Hwang et al. 2010). It is likely that this distancing impedes their parents’ judgment of how much adolescents actually reveal to them. Therefore, we expected mother–adolescent agreement to be lower in families in which adolescents often serve as language brokers.

Studying both language brokering and adolescents’ levels of autonomy allows exploration of the interplay between normative development (autonomy) and acculturation (language brokering). In particular, comparisons of associations between an immigrant and a native sample will allow us to find out whether adolescent-mother agreement in immigrant families is based on the same processes as it is in native families (similar associations between autonomy and agreement without acculturation-related effects), whether it is primarily the result of acculturation-related processes (strong effects of language brokering and small effects of autonomy), or whether both processes reinforce each other or cancel each other out (interaction effects of autonomy and language brokering). Although interactions between acculturation and normative development have received increasing attention recently (Michel et al. 2012; Fuligni 2001), empirical results are still quite scarce. We expected that both processes would interact and that they would either reinforce each other or cancel each other out. On the one hand, a reinforcement could be expected based on coercive family interactions (Granic and Patterson 2006) in which family processes can escalate over time. In our example, language brokering and autonomy not only may have a cumulative effect but also may potentiate each other’s effects, resulting in accelerated disagreement between parents and adolescents. On the other hand, one of these processes (high levels in language brokering or autonomy) may be sufficient to create a family dynamic in which mothers are not able to recognize adequately how much their children actually tell them. In this situation, both processes would cancel each other out.

Although the prediction of interdyadic differences focused on normative development (autonomy) and acculturation (language brokering), we also needed to consider some additional control variables. These were primarily demographic variables: age, gender, parental and adolescent education, and a history of family disruption (e.g., divorce). Age seems especially relevant because it relates to substantial changes in intergenerational family relationships, with adolescents becoming more distanced from their parents as part of the individuation processes (Paikoff and Brooks-Gunn 1991). Consequently, older adolescents in both native and immigrant families probably agree less with their mothers on disclosure than younger adolescents do. In addition, female adolescents may be more similar to their mothers in their ratings of disclosure than males are because mothers and daughters have comparable gender roles, and this tendency would lead to more parental knowledge in same-sex dyads (Crouter et al. 1999). Regarding education (parental or adolescent), one could argue that better-educated individuals may have better capabilities in perspective taking so that dissimilarities within dyads are less likely (Cutting and Dunn 1999). Furthermore, better educated parents generally adhere to more egalitarian child-rearing practices and probably also pay more attention to what their children do (Franzoi and Davis 1985). In addition, a history of divorce or widowhood can result in a closer mother–child relationship (Arditti 1999), which may result in higher levels of adolescent-mother agreement; this may be even more likely among immigrants due to the stress the migration process imposes on the family (Remennick 2004). These variables, therefore, were controlled in all analyses.

Developmental Outcomes of Mother–Adolescent Agreement

The final research question concerns the relevance of mother–adolescent agreement on children’s disclosure for the development and psychosocial functioning of adolescents and their families. In general, agreement in social relationships relates to dyadic harmony and subsequently to a better mutual understanding as indicated by better mutual ratings of social skills and higher levels of reported need fulfillment (Gavin and Furman 1996). Thus, higher levels of mother–adolescent agreement on children’s disclosure, reflecting dyadic agreement on communication processes in terms of sharing private information, may be expected to result in better psychosocial and family functioning. In a similar vein, the theory of acculturative family distancing, developed from an immigration-related perspective, assumes that the distancing between adolescents and their mothers is related to the risk of mental illness and dysfunction through increases in family conflict (Hwang 2006; Hwang et al. 2010). Hwang assumes that such a dynamic is particularly likely in the adolescent years (Hwang 2006; Hwang et al. 2010). Empirically, the association between mother–adolescent agreements and developmental outcomes has been shown. For example, mother–adolescent discrepancy concerning monitoring-relevant behaviors (including children’s disclosure) was found to predict adolescent delinquency (De Los Reyes et al. 2010). Both of these approaches suggest that higher levels of agreement are associated with better psychosocial and family functioning. In other words, it was expected that adolescents would report lower levels of conflict and depressive symptoms when their agreement with their mothers was higher. This was expected because high mother–adolescent agreement can be assumed to reduce familial distress, support mutual understanding, and hinder erosion in the mother–adolescent bond.

Hypotheses

Although agreement may be higher between parents and children than in other social relationships (partnerships and friendships), we expected substantial interdyadic differences in mother–adolescent agreement. More specifically, we expected that mother–adolescent agreement on children’s disclosure would be higher in native than in immigrant families (Hypothesis 1) because native adolescents can be assumed to experience value congruence at home and in other spheres of life, whereas immigrant adolescents are often exposed to heritage culture values at home and host culture values in school. In addition, the different rates of parental and filial cultural adaptation were assumed to result in acculturative family distancing, which may erode mother–adolescent agreement in immigrant families.

When considering interdyadic differences within the groups, it was expected that normative development and acculturation-related processes must be considered simultaneously to predict mother–adolescent agreement in reported children’s disclosure. It was expected that the autonomy of the adolescent would be related negatively to mother–adolescent agreement in both the native and the immigrant groups (Hypothesis 2). In addition, language brokering, an indicator of adolescents’ advanced sociocultural adjustment, should be related to lower levels of mother–adolescent agreement in the immigrant group (Hypothesis 3). Furthermore, we aimed to test the interplay between normative development (autonomy) and acculturation (language brokering) and an interaction effect was expected. However, the direction of this interaction cannot easily be derived from the current literature. For this reason, two competing hypotheses were tested. Autonomy and language brokering effects may either reinforce each other, resulting in more pronounced associations between autonomy and child-disclosure agreement under conditions of high levels in language brokering (Hypothesis 4a), or cancel each other out, resulting in depressed associations between autonomy and child-disclosure agreement under conditions of high levels in language brokering (Hypothesis 4b).

The final research question addressed the outcomes of mother–adolescent agreement for the adolescent and the family. We assumed that higher levels of child-disclosure agreement would support a mutual understanding between mothers and adolescents, ensuring that parents are aware of whether or not they receive information from their offspring and enabling them to act accordingly. Thus, we expected child-disclosure agreement to relate to lower levels of conflict (Hypothesis 5a) and lower levels of depressive symptoms (Hypothesis 5b) among adolescents in both groups.

Methods

Sample

Recruitment of the sample took place in 2010/2011 in collaboration with registry offices in three West-German cities in North Rhine-Westphalia. This sampling design limits the generalizability of our results to those areas in Germany which host a substantial share of immigrants. These cities provided a list of citizens based on the country of origin (states of the former Soviet Union or Germany) and age (between 10 and 18 years old). From this list, participants were randomly selected and were invited, together with their mothers, to participate in the study. The participation of mothers and adolescents was independent and both were offered €10 (approximately US $14) in cash or a voucher of similar value that they could spend in a variety of shops in the participating cities. Mothers and adolescents completed the questionnaires anonymously and separately, and were matched by a unique code that only the mothers and adolescents could provide (based on letters and numbers of names and birth dates). Altogether, 665 immigrant families from the former Soviet Union and 510 native German families were invited to participate; of these, 196 immigrant and 203 native German family dyads responded and participated. Unfortunately, we have no data to compare participating and nonparticipating families. It is known, however, that in German survey research a response rate of 50 % is rarely ever achieved (Mohler et al. 2003), and this rate can be expected to be even lower in immigrant samples due to higher mobility and possibly also due to greater reservations. For the study presented here, 11 immigrant and six native German dyads with single mothers were excluded because the relationship between single mothers and their children may differ substantially from other families. Thus, the final sample consisted of 185 immigrant and 197 native German dyads. All participating immigrant adolescents were first generation immigrants (born in a country of the former Soviet Union) with an average of 9.7 years of residence in Germany.

The demographics for native and immigrant families are presented in Table 1. Both groups were highly comparable on nearly all the variables assessed, including adolescents’ gender distribution and parental education. Differences between the two groups were only found for age and school track. The immigrant adolescents were approximately 1 year older than the native German adolescents, probably because the number of immigrants from the former Soviet Union entering Germany decreased from more than 90,000 in 2002 to about 3,300 in 2009 (Bundesverwaltungsamt 2010), so that the share of older adolescents from the former Soviet Union in the population is much larger than that of younger immigrants from these countries. Adolescents in both the native and the immigrant group came from all school tracks and—in line with earlier research (Baumert and Schümer 2002)—immigrant adolescents were overrepresented in lower school tracks.

Table 1 Means, standard deviations, and proportions of the variables
Full size table

Measures

The questionnaires for immigrants (both mothers and adolescents) were presented in both languages (German and Russian) to enable participants to complete the questionnaire in the language in which they had the highest competence. The comparability of the Russian and German versions was ensured by a translation-back-translation method in which two bilingual, native-Russian speakers independently translated the German version first into Russian and then back into German. Any deviations between the forms were discussed and changed so that the meaning of the items was the same in both languages.

Child Disclosure

This variable assessed mothers’ and adolescents’ perceptions of how much the adolescents report about their lives and was measured using a scale from Kerr and Stattin (2000). It comprised five items, including the following examples: Do you talk at home about how you are doing in different subjects in school? If you are out at night, when you come home, do you tell your parents what you have done that evening? Do you hide a lot from your parents about what you do during nights and weekends? The same items were given to mothers, but slightly reformulated. For example: Does your child hide a lot from you about what he/she does during nights and weekends? Mothers and adolescents rated each item on a 6-point Likert scale between 1 (No, never) and 6 (Yes, all the time). The scale’s internal consistency was sufficient at between α = .62 and α = .75, depending on informant and ethnic group.

Conflicts

Parent–child conflicts were assessed by considering mother–adolescent arguments in 10 domains that are typical of parent–child disagreements (school-related topics, bedtime, time on the phone, watching television, internet use, time on the computer, tidiness of bedroom, pocket money, type of clothes, and friends with whom the child spends spare time); these are situations in which mothers have reported experiencing at least occasional disagreements with their adolescents (Adams and Laursen 2001; Laursen 1993). For this study, the sum across the domains was used and this varied from 0 (no arguments in any of these domains) to 10 (arguments in all 10 domains).

Depressive Symptoms

Depressive symptoms were assessed through nine items adapted from the Child Behavior Checklist (Achenbach 1991). Adolescents were rated on a 6-point Likert scale from 1 (does not apply) to 6 (does apply) in terms of the extent to which certain statements applied to them. The statements included: I feel lonely, I am fearful/anxious, I am unhappy/sad/depressed, and I worry a lot. The alpha consistency of this scale was α = .88 for immigrants and α = .91 for native Germans.

Autonomy

This scale assessed emotional autonomy based on a well-tested instrument (Schmitz and Baer 2001; Steinberg and Silverberg 1986). However, due to a recent critique and reanalysis of this scale (Beyers et al. 2005), we confined ourselves to items referring to the positive subcomponent of emotional autonomy (deidealization and nondependency) and did not use items referring to detachment (ignorance and alienation). Five items were used, including: It’s better for kids to go to their best friend than to their parents for advice on some things, and My parents know everything there is to know about me (recoded). Adolescents rated the extent to which they agreed with each statement on a 6-point scale ranging from 1 (completely disagree) to 6 (completely agree). Internal consistency was α = .75 for immigrants and α = .72 for native Germans.

Language Brokering

Adolescents used a 6-point scale to report how often they had translated documents from a given list for their parents in the previous 3 months (Weisskirch and Alva 2002). The list comprised 11 documents including notes or letters from school, medical forms or bills, job applications, insurance forms, and bank statements. The scale ranged between 1 (never) to 6 (very often). The alpha consistency of this scale was α = .89.

The scales applied in both groups were (1) tested for factorial structure using two-group confirmatory factor analyses (CFAs), and (2) tested for the agreement of factor loadings between the native and immigrant groups. These analyses were conducted in two steps because the conflict scale used dichotomous indicators, which requires adapted estimation and modeling techniques (WLSMV estimator for dichotomous indicators vs. MLR estimator for continuous indicators). The two-group CFA for the conflict scale showed a good fit to the data: χ 2 (68, n = 379) = 123.37, p < .01; CFI = .97, RMSEA = .07; 90 % CIRMSEA [.05; .08]. Fixing the factor loadings did not lead to a worse model fit: Δχ 2 (9, n = 379) = 11.38, p = .25. Thus, the factor loadings can be considered the same between the two groups. Next, the other scales (maternal and adolescent disclosure, autonomy, and depressive symptoms) were entered into a two-group CFA together. All scales were modeled as correlated latent factors based on their items as manifest indicators. The correlations between reverse-coded items were correlated in order to cover method variance. The proposed factorial structure showed a sufficient fit with the data: χ 2 (462, n = 381) = 807.07, p < .01; CFI = .90, RMSEA = .06; 90 % CIRMSEA [.06; .07]. Restricting the factor loading to be equal between groups did not affect the model fit: TRd (df = 20) = 26.38, p = .15. It should be noted that TRd is the test statistic of the scaled χ2-difference test (Satorra and Bentler 2001). Thus, the factor loadings can be considered group invariant.

Analyses

The main purpose of this study was to investigate the role of mother–adolescent agreement regarding adolescent disclosure. For that purpose, profile agreement within each mother–adolescent dyad was measured by applying intraclass correlation (one-way random) (McGraw and Wong 1996). This type of intraclass correlation, known from interrater agreement estimations, has the unique advantage over Pearson-based intraclass correlation that differences in elevation, shape, and scatter/variation of the profiles feed into a single indicator of within-dyad agreement (Furr 2010). Thus, a profile agreement measure was calculated for each dyad. As the distribution of these intraclass correlations is not normal, the values were Fisher Z transformed in order to serve as variables in all subsequent analyses. The reported descriptive agreement measures are retransformed Fisher Z correlations. Here, the Pearson correlation metric has been reestablished. Thus, high positive values indicate high mother–adolescent agreement, while correlations close to zero point to independent disclosure ratings and thus low agreement.

All research questions were analyzed using the statistical package AMOS 20.0 (Arbuckle 2011). For the first research question, the mean-level differences were tested by using equality constraints for the intercept in child-disclosure agreement, and we tested whether constraining the means to be equal across groups significantly worsened the model fit (χ 2 difference test between the unconstrained and the constrained model). The second research question aimed to predict mother–adolescent agreement on child disclosure using multivariate regressions. The analysis consisted of three single steps. In the first step, control variables (age, gender, parental education, academic track, and family history) were entered as direct predictors of the outcome. In the second step, autonomy was added as predictor, and in the third step we added migration-specific predictors (length of residence and language brokering), as well as the interaction between language brokering and autonomy.

The third research question focused on the outcomes of mother–adolescent agreement and thus the agreement measure regarding disclosure was used as a predictor of conflicts and depressive symptoms in a regression analysis. In a first step, the control variables were entered (see above) together with autonomy. In the second step, the agreement measure was added as the predictor.

Results

The first hypothesis expected higher mother–adolescent agreement in the native sample compared to the immigrant sample and was tested by using equality constraints in a two-group model using AMOS (mean values in the intraclass correlation coefficient (ICC) were set to be equal in the native German and the immigrant samples and change in model fit was tested). This test revealed that the mother–adolescent agreement concerning child disclosure was indeed significantly lower in immigrant dyads: Δχ 2 (1, N = 382) = 7.82, p < .05. The agreement was r icc  = .67 for native Germans and r icc=.51 for immigrants. Including the covariates (age, gender, parental education, school track, and family history of disruption) did not change this result substantially and after controlling for these variables, native German dyads reported an agreement of r icc  = .67, whereas immigrants reported an agreement of r icc  = .54, which were still significantly different: Δχ 2 (1, N = 382) = 4.28, p < .05. Thus, Hypothesis 1 was supported by the data. Although the ICC takes into account various parameters for estimating mother–adolescent agreement, one deviation between mothers and adolescents can be seen in Table 1. Whereas the native mothers and immigrant mothers do not differ in their perceptions of how much their adolescents tell them, we found significant differences between native and immigrant adolescents, suggesting that the immigrant adolescents actually disclose less information to their mothers.

Our second research question related to the prediction of mother–adolescent agreement in reported child disclosure. To answer this question, various regression analyses were conducted using AMOS. As all these models were fully saturated models, no fit statistics are reported. The bivariate intercorrelations of the variables in the model are shown in Table 2 and the coefficients of the multivariate regressions in Table 3. We conducted two steps for the native German group and three steps for the immigrant group. In the first step, the control variables were entered. In the second step, autonomy was added to the prediction, and in the third step we added the immigration-specific predictors for the immigrant group. Step 1 and Step 2 are the same for native and immigrant dyads and allow the comparison of effects of immigration-unspecific predictors, whereas the focus of Step 3 is on the question of what the immigration process can add in explaining mother–adolescent agreement. The results of Step 1 showed some association between the control variables and mother–adolescent agreement. More specifically, higher age was associated with lower levels of adolescent–mother agreement in both groups and, not surprisingly, being female was related to higher levels of mother–adolescent agreement (even though the effect just missed significance in the immigrant sample).

Table 2 Bivariate correlations of variables
Full size table
Table 3 Unstandardized regression coefficients (standard error) of a structural equation model predicting mother–adolescent agreement (ICC)
Full size table

In Step 2, autonomy was added as predictor to the model to test Hypothesis 2, which assumed a negative association between autonomy and mother–adolescent agreement. As expected, higher levels of autonomy were associated with lower levels of mother–adolescent agreement on disclosure in both samples, thus supporting Hypothesis 2. The immigration-specific variables (Step 3 for the immigrant group) added significantly to the prediction of mother–adolescent agreement. More specifically, higher levels of language brokering related to lower levels of mother–adolescent agreement on child disclosure, thus supporting Hypothesis 3. Finally, the coefficients for the interaction terms of autonomy and language brokering also reached significance. This interaction is depicted in Fig. 1. The graph shows that the association between autonomy and mother–adolescent agreement is significantly stronger for adolescents who rarely serve as language brokers in their families compared to the association found for adolescents who frequently take on this responsibility. In addition, the graph shows that it does not matter whether an adolescent serves as a broker or not in terms of mother–adolescent agreement as long as the level of autonomy is high. These results indicate that both processes seem to cancel each other out rather than reinforcing each other. Thus, Hypothesis 4b was supported by the data, whereas Hypothesis 4a has to be rejected. The association between agreement and autonomy was barely affected by adding the immigration-related variables to the prediction.

Fig. 1
figure 1

Interaction of language brokering and autonomy in the prediction of mother–adolescent agreement on child disclosure (Fisher Z-values)

Full size image

The third research question focused on the outcomes of mother–adolescent agreement in reports on child disclosure. We expected mother–adolescent child-disclosure agreement to be associated with better mother–child relationships as indicated by lower levels of conflict and depressive symptoms. However, the results revealed a more complex picture than that formulated in the hypotheses (see Table 4). Regarding parent–adolescent conflicts, there were very few predictions based on the control variables. In the native group, students attending the higher school track and girls reported fewer conflicts with their parents, and among immigrants, higher levels of autonomy were associated with more conflicts (Step 1). Adding mother–adolescent agreement on disclosure to the model (Step 2) revealed that this measure of mother–adolescent agreement was one of the strongest predictors in the model for both samples. As expected, higher levels of mother–adolescent agreement on child disclosure predicted fewer conflicts in both samples, thus supporting Hypothesis 5a.

Table 4 Unstandardized regression coefficients of the prediction of conflicts and depressive symptoms
Full size table

In the prediction of depressive symptoms, some significant associations with the control variables were found. In the group of native Germans, adolescents from disrupted families reported higher levels of depressive symptoms. In the immigrant group, a positive association between age and depressive symptoms was found, as well as a higher level of depressive symptoms among female participants compared to male participants. In addition, only among immigrants was a positive association between autonomy and depressive symptoms found. Adding the agreement measure did not change the patterns of associations and also did not predict depressive symptoms in either group. Hypothesis 5b was therefore not supported.

Discussion

Mother–adolescent relationships are frequently discussed in immigration literature and various authors assume that intergenerational distancing in immigrant families is one source of long-term maladjustment and lower levels of well-being (e.g., Fillmore 2000; Hwang 2006). However, actual mother–adolescent agreement and its predictors and outcomes have rarely been assessed directly. Furthermore, comparisons with development in nonimmigrant native groups have been missing. To fill this gap, our study focused on immigrants’ and natives’ mother–adolescent agreement on children’s disclosures. Children’s disclosures can be seen as the result of mutual communication processes between parents and children that are driven by both these sources (De Los Reyes et al. 2008; Kerr and Stattin 2000).This understanding goes beyond existing research with immigrants, which often assumes that mother–child agreement is the result of transmission from parents to children (e.g., Schönpflug 2009). The premise behind this study was that, if the adaptation of adolescent immigrants to a new cultural setting does indeed undermine parent–child relationships, we should see lower mother–adolescent agreement on children’s disclosure compared to that found among natives, an indication that parent–child communication in immigrant families is to some extent less optimal compared to that in families without acculturation-related challenges.

Ethnic Group and Interdyadic Differences in Mother–Adolescent Agreement

Mothers in both of our samples reported higher levels of children’s disclosures than their children did, a result that has also been found in other studies (De Los Reyes et al. 2010). The mean-level generational difference reflects that mothers tend to overestimate the amount of information they receive from their adolescents. However, our data show that immigrant mother–adolescent dyads disagree to a greater extent on the adolescent’s disclosure compared to native mother–adolescent dyads. These findings support the assumption that the communication between mothers and their adolescents is somewhat different in immigrant families; in particular, immigrant mothers may have greater difficulty in estimating how much their children actually tell them. One explanation for this mean-level difference in agreement can be derived from the finding that language brokering predicted mother–adolescent agreement in immigrant families over and above the normative processes of autonomy development, which was of significance in both samples. The explanation for this additional acculturation effect can be found in the theory of acculturative family distancing (Hwang 2006; Hwang et al. 2010). Language brokering relates to the channeling by immigrant adolescents of their parents through the challenges of a new society by helping them negotiate the new language, a behavior that is likely to enhance the adolescents’ self-sufficiency and independence. However, one needs to bear in mind that the role of language brokering differs among adolescents. Among those who reported high levels of autonomy, the level of language brokering is of little relevance for the level of mother–adolescent-agreement (Fig. 1).

The interaction between language brokering and autonomy shows that normative development and acculturation can be competing processes for select outcomes. Research focusing solely on normative developmental aspects may result in an oversimplification of the dynamics in immigrant families, as would a focus on acculturation-related variables alone. Furthermore, the interaction was significant despite rather low levels of language brokering in our sample. This finding suggests that even rare language brokering behaviors can affect the autonomy-related changes in the relationship between mothers and adolescents. It is important to note that the low German competence of mothers is one of the main reasons for adolescents’ being family language brokers (Schulz et al. 2013). Maternal language training may therefore be one way to avoid the occurrence of such a family distancing. Taken together, our results demonstrate that lower mother–adolescent agreement in immigrant families can be derived from two sources—either language brokering or autonomy. For adolescents with a high level in one or other predictor, the other process seems to have little relevance. Of course, this conclusion is solely based on our cross-sectional results and needs corroboration with longitudinal data, which were not available in this project.

The interaction may, however, also shed light on age-related findings, suggesting that language brokering is associated more strongly with higher levels of internalizing problems among early adolescents as compared to late adolescents (Schulz et al. 2013), and that younger children especially show greater role confusion than older children when they broker (Puig 2002). Our results extend these findings by suggesting that language brokering is particularly related to lower mother–adolescent agreement when the autonomy is low: that is, at younger ages. This means in turn that parents of younger adolescent language brokers possess less information about their children’s whereabouts and actions than would be normal given the age of the adolescents. The result may be that the corrective parental actions and support necessary in these early years are not implemented and that there may be a greater likelihood of internalizing problems. Among late adolescents, language brokering may not be associated with internalizing problems, because they—due to their advanced autonomy levels—already show lower levels of mother–adolescent agreement (independent of their level of language brokering) and may also need parental guidance to a lesser extent than early adolescents. These speculations certainly require more research and empirical support.

Mother–Adolescent Agreement and Psychosocial Outcomes

With regard to the developmental outcomes of mother–adolescent agreement, our data first of all show that higher levels of agreement relate to lower levels of family conflict in both samples. In both native and immigrant families differences in the estimation of how much adolescents tell their parents are reflected in lower levels of family understanding and harmony. This is exactly what we expected on the basis of the theory of acculturative family distancing (Hwang 2006; Hwang et al. 2010). Along with the results predicting mother–adolescent agreement, our research suggests that the effects of lower levels of mother–adolescent agreement on family conflicts are rather similar across ethnic groups.

Initially, we also expected that lower levels of disclosure agreement between mothers and adolescents would relate to higher levels of depressive symptoms. This hypothesis was not supported in the multivariate analyses, although a small significant negative association was found in the bivariate correlations for the immigrants (Table 2). Various reasons can be considered for the missing link between disclosure agreement and depressive symptoms. Perhaps the most relevant factor could be that internalizing problems, such as a depressed mood, depends on many more variables. Particularly in adolescence, peer environment and personal characteristics need to be considered (Nguyen et al. 2011; Buck and Dix 2012). These factors may be an even more important influence than mother–adolescent agreement on adolescents’ disclosure. In addition, one of the main protective factors against depressive symptoms is social support (Cohen and Wills 1985), which may be provided by other sources and not necessarily by parents. Such external support may buffer against a less than optimal family communication as indicated by low levels of mother–adolescent agreement on disclosure. This is an area that certainly deserves more attention in future research.

Among the other results found, two stand out. First, a family history of divorce or widowhood was associated with more depressive symptoms in native adolescents, whereas no such association was found in the immigrant sample. We cannot say why such a difference was found in our study. It may be that other sources of (acculturation-related) stress are more prevalent in immigrant families compared to the family history of widowhood and divorce. It may also be that this history leads back to life in the former Soviet Union and has little impact on life circumstances in Germany, or it may be that immigrant-specific buffers (e.g., a strong ethnic community) help in coping with these past experiences. Future research should address such differences more closely, as they may reveal the specific mechanisms by which immigrant and native families deal with adversities.

A second noticeable difference refers to the association between autonomy and depressive symptoms, which was significant and positive in the immigrant group but not in the native group. Based on Steinberg and Silverberg (1986), becoming more autonomous is a normal developmental task in adolescence and its accomplishment should be adaptive rather than maladaptive. In our study, autonomy does indeed seem to represent normal development in both samples. Age and autonomy were correlated substantially in both samples (Table 2) and the negative association between age and mother–adolescent agreement, which was also found in both samples, disappeared after autonomy was taken into account (Table 3). This pattern of effects suggests that both groups become more autonomous with age and that this process is associated with lower mother–adolescent agreement. Among immigrants, however, the age-related increase in autonomy may be an indication of adaptation to the western individualistic culture represented by peers and school. In the often heritage-oriented collectivist family context, such as among immigrants from the former Soviet Union, being autonomous in adolescence might not meet the cultural expectations of the family (Kagitcibasi 2005; Love and Buriel 2007), and thus a cultural misfit might lead to heightened levels of depressive symptoms (Friedman et al. 2010). However, this assumption naturally requires more research.

Limitations

When interpreting the results, some limitations have to be borne in mind. First, all analyses were cross-sectional. Therefore, no causal inferences can be made and longitudinal studies are necessary to corroborate our interpretation of the result patterns. In this regard, investigating trajectories of mother–adolescent dyads in particular over time would expand the scientific opportunities tremendously. Such a design would allow researchers to test whether a decrease in language brokering relates to an increase in mother–adolescent agreement (in other words, whether the mother–adolescent relationship can change back), or whether lower levels of agreement remain low despite the fact that the potential source is eliminated. In a similar vein, important variables that could account for the reported associations, such as the maternal warmth of the mother–adolescent relationship, have to be investigated in future research in terms of their roles in the processes suggested here (Rodríguez et al. 2013).

Second, the sample was restricted to a specific group of immigrants: namely, those born in the former Soviet Union. We cannot be sure whether we would find similar results in a group of immigrants with a different cultural background. For example, adolescent task burdens were not associated with health hazards in Turkish immigrant families, whereas they were in Greek and Italian families (Nauck and Niephaus 2006), probably due to a higher family cohesion in Turkish families. For our research, this means that the effects of language brokering, as one type of task burden, may differ depending on the ethnic or cultural group being studied. However, our assumptions and interpretations were not strongly related to a particular ethnic or cultural group. In order to strengthen our interpretation, the results should therefore be replicated with different immigrant samples that differ in immigration history or religious background. Of particular interest would be the question of whether the effects are less pronounced in groups with higher levels of “filial piety” or “familism,” since such groups place a very high value on family and duties for family.

Third, we only focused on a very specific area of disclosure in adolescents’ lives—school and spare time—so the results may differ for other contexts, as adolescents disclose different issues depending on the domain (Smetana et al. 2006). Moreover, we only focused on two specific outcomes: conflicts and depressive symptoms. In order to draw broader conclusions concerning the relevance of parent–adolescent agreement in developmental processes, a wider range of outcomes should be considered, because low mother–adolescent agreement may show impairment in one area (e.g., intrafamilial conflicts as studied here), but adolescents may profit from it in another (e.g., through better integration into their peer group in the dominant host culture). Opening the complexity of the study so that it includes different types of disclosure as well as more outcomes may therefore be illuminative.

Finally, studying dyads is innovative, but we had to restrict ourselves to mother–adolescent dyads. It would be highly informative to compare the result patterns regarding mother–adolescent agreement with those for father–adolescent agreement, because processes of trust and communication may differ between mothers and fathers (Smetana et al. 2006). We might expect similar associations for fathers as for mothers, despite the fact that the absolute level of disclosure may be lower in father–adolescent dyads. Some associations, however, might be more pronounced for fathers, because their involvement with their children is less scripted than mothers’ (Crouter et al. 1999). These assumptions remain to be tested, however, because paternal warmth was found to decrease over time, whereas maternal warmth remained stable (Rodríguez et al. 2013). Another possible avenue of research would be to test whether parent–adolescent agreement is higher in same-sex parent–child dyads than in opposite-sex dyads because of similar gender roles. These considerations show that the complexity of family interactions provides much more scope for research than can be encompassed in a single study.

Conclusion

Our study provides new insights about the communication of mothers and adolescents in immigrant and native dyads. Quite remarkable in this regard are the similarities found. In both groups, mothers perceive that their children tell them more than the adolescents actually do. In both groups, autonomy is related to lower levels of mother–adolescent agreement with regard to adolescents’ disclosure. In both groups, higher mother–adolescent agreement on child disclosure is related to lower levels of conflict. Differences between groups are found primarily in the average degree of mother–adolescent agreement and in the fact that language brokering is an additional factor in the prediction of lower mother–adolescent agreement in immigrant families. Thus, family communication in immigrant families seems to be rather similar to that in native families. Beyond such family dynamics, our study also provides information concerning the source—mother reports versus child reports—when studying adolescent behaviors. Our data suggest that mothers’ perceptions of adolescents’ levels of disclosure are not always in line with the reports of their adolescents. Higher levels of agreement can be found when adolescents have lower levels of autonomy (which is likely to be at younger ages). With increasing autonomy or—among immigrants—when adolescents act as language brokers, mothers’ reports on their adolescents’ disclosures differ more from those of their adolescents. It thus seems that our study can contribute substantially to the debate on family dynamics in immigrant and native families. It does, however, also point to future directions for study, such as research on the differences in variable associations between the two groups studied, the inclusion of other family members, and longitudinal studies that could address changes over time. All these approaches may help in developing a much more differentiated picture of the dynamics in immigrant families, which is vital as the family is certainly one of the most influential contexts for the long-term adjustment of adolescents.