This study utilized data from the Smart Toddlers study, a study conducted in Switzerland on parental smartphone use, parent–child interactions, and child development in the toddler years (aged 14–36 months). Smart Toddlers is the follow-up study to the Smart Start study (Wade-Bohleber et al., 2024); some parents in that study participated in the Smart Toddlers study as well. Other parents were recruited through health institutes, child care facilities, social media channels, and a magazine advertisement in the German-speaking part of Switzerland. Inclusion criteria were: participation with a first-born child, resident of Switzerland, no chronic diseases, use of digital media, and sufficient German or English language skills to fill out the questionnaires, which were available in German and English. For this study we used Smart Toddlers study data from time point T1 (14 months postnatal), in which 261 parents had participated. Of these 261 parents, 110 couples and 23 mothers with a first-born child had participated. Of the 119 couples, one was a same-sex couple. Together with this couple, one mother was defined as a co-mother. One couple participated with her twins, whereby we used the data only from the older twin in the study. Table 1 shows characteristics of the sample.

Prior to the start of the study, all participants were informed about the study aim and procedures and signed a participation consent form. Study participants received no financial compensation for their participation.

When their child was 14 months old, parents received a link by email to the online questionnaire.

Statistical analyses were conducted using SPSS Version 29.0.1 and RStudio Version 2023.09.1. First, we calculated the descriptive statistics of the variables of interest and then computed Pearson correlation coefficients among the scale mean scores. To reduce the complexity of the further analysis, the number of items for the three constructs parental problematic smartphone use, difficult child temperament, and parenting stress was reduced through item parceling. Item parceling has been used more and more frequently in recent years (see Bandalos, 2002; Hapsari and Widhiarso, 2023; Little et al., 2002; Hau and Marsh, 2004; Marsh et al., 1998). Initially, confirmatory factor analyses (CFA) were conducted to verify the theoretically assumed measurement models for the constructs of parental problematic smartphone use, difficult child temperament, and parenting stress prior to item parceling. For the unidimensional constructs parental problematic smartphone use and difficult child temperament, a CFA with a single-factor model was conducted for each. For the multidimensional construct parenting stress, a seven-factor model was specified, with items assigned to seven latent dimensions (PS1 to PS7) according to their theoretical allocation. The results of the CFA (see Appendix Table A1) indicate that an optimal model fit was not achieved for all constructs. Specifically, the root mean square error of approximation (RMSEA) for the constructs parental problematic smartphone use and difficult child temperament exceeds the recommended thresholds (Hu and Bentler, 1999). However, according to Kenny et al. (2014), this can occur in models with a low number of degrees of freedom. All items show significant and predominantly acceptable factor loadings (mostly > 0.50), indicating satisfactory conceptual coherence of the respective constructs. Given the planned use of item parcels for the subsequent analyses, all items, including those with low factor loadings, were retained due to their theoretical relevance. Based on this decision, and adapted from Little et al. (2002), the items for the unidimensional constructs parental problematic smartphone use and difficult child temperament were divided into three parcels using the balancing approach of item scale correlation. The items were divided such that the item with the highest and the item with the lowest correlation formed a parcel, the second highest with the second lowest a parcel, and the remaining three items another parcel. For the construct difficult child temperament, the item with the highest, the lowest, and the third-highest item-scale correlation formed a parcel, and so on. For the multidimensional construct parenting stress, we used domain representative parceling: Here four parcels were created with one item each from each dimension (Little et al., 2002). Cronbach’s alpha of the parcels for parental problematic smartphone use, difficult child temperament, and parenting stress was α = 0.83, α = 0.82, and α = 0.92, respectively. To ensure internal consistency, the parcel discriminations were considered. These ranged from 0.65 to 0.72 for difficult child temperament, from 0.78 to 0.84 for parenting stress, and from 0.69 to 0.70 for parental problematic smartphone use, indicating good scale reliability (Bortz and Döring, 2006). We then conducted a moderation mediation analysis using latent variables, based on structural equation modeling (SEM) using lavaan (Rosseel, 2012) with the maximum likelihood robust (MLR) estimator, which provides robust standard errors. To further account for the nested data structure, we applied cluster-robust standard errors (CRSE; Cameron and Miller, 2015), which correct for within-cluster dependencies in standard error estimation.

The moderated mediation analysis was conducted with parental problematic smartphone use as the outcome variable, difficult child temperament as the predictor variable, and parenting stress as the mediator. Parental gender was modeled as the moderator of the association between parenting stress and parental problematic smartphone use. The latent interaction variable was modeled using double mean centering (Lin et al., 2010). As a covariate we had wanted to include parental education as a binary variable (0 = no university degree, 1 = university degree) to measure family socioeconomic status (SES), as research has shown it to be the most consistent and widespread component of SES (Sirin, 2005). However, as parental education did not significantly correlate with any other variable (see Table 2), we did not include it in the model. As model fit indices, we report the relationship between the chi-square (χ²) and degrees of freedom (df), the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). CFI values over 0.95, RMSEA values below 0.06, and SRMR values less than 0.08 indicated a good model fit (Hu and Bentler, 1999).

Table 2 presents bivariate correlations among the variables in this study, Table 3 the descriptive statistics for the variables.

No significant mean differences were found between mothers and fathers for the variables problematic smartphone use, difficult child temperament, and parenting stress. For parenting stress, 46% of mothers and 37.5% of fathers had a high level according to the test manual (t-value ≥ 60), which can lead to difficulties in coping with parenting tasks (Tröster, 2011).

The model was found to fit the data well. For information on the model fit indices (scaled) and the standardized results, see Figure 2 and Appendix Table A2.

Difficult child temperament was significantly positively associated with parenting stress (ß = 0.42, p < 0.001). The more difficult the child’s temperament was rated, the higher the reported parenting stress was. In turn, parenting stress was significantly positively associated with parental problematic smartphone use (ß = 0.42, p < 0.001). The greater the extent to which parents reported experiencing parenting stress, the higher the level of parental problematic smartphone use was. Parent gender did not moderate the association between parenting stress and parental problematic smartphone use (ß = −0.03, p = 0.607). There was no direct association between difficult child temperament and parental problematic smartphone use (ß = 0.01, p = 0.956). The indirect association between difficult child temperament and parental problematic smartphone use, mediated by parenting stress, was statistically significant (β = 0.18, p = 0.001). This means that parenting stress acted as a mediator. The total association between difficult child temperament and parental problematic smartphone use was significant (β = 0.18, p = 0.023) as well. When examining the indirect association in a model differentiated by the gender subgroup [χ²(64) = 79.07; CFI = 0.99; RMSEA = 0.043; SRMR = 0.040], we observed a significant indirect association for both mothers (β = 0.18, p = 0.016) and fathers (β = 0.17, p = 0.039). However, the difference between genders was not significant (p = 0.892).

The more difficult a child’s temperament was rated by their parents, the more parenting stress the parents reported. Consistent with prior research demonstrating the association between difficult child temperament and parenting stress (e.g., Chang et al., 2004; McBride et al., 2004; Molfese et al., 2010; Östberg and Hagekull, 2000; Solmeyer and Feinberg, 2011), parents reported increased levels of parenting stress when they perceived their child as irritable and difficult to soothe. Shin et al. (2021) demonstrated that higher negative affectivity and lower effortful control in toddlers were associated with increased parenting stress. Parenting stress can arise among other things due to feeling overwhelmed in parenting, having a feeling of low self-efficacy in the parent role, and experiencing personal restrictions imposed by parenthood (Deater-Deckard, 2004; Diener and Swedin, 2020). The behavior of a 14-month-old child with a temperament perceived as difficult could intensify these stressors. For instance, Porter and Hsu (2003) found that mothers who perceived their infants as having a difficult temperament reported lower maternal efficacy.

In addition, parenting stress can restrict parental mentalization. A central assumption of the mentalization approach is that the ability to mentalize decreases when stress increases (Luyten and Fonagy, 2015). Parental mentalization refers to the capacity of parents to think about their own and their child’s behaviors in the context of mental states (needs, desires, feelings, etc.) (Fonagy et al., 2015). If parents are not able to mentalize, they do not see difficult behaviors as caused by their child’s temperament; instead, they may feel that the behavior is directed against themselves, which in turn can lead to feeling low self-efficacy as a parent and a higher stress level. This could have a negative effect on the child’s behavior and result in a vicious circle.

The goodness of fit model by Thomas and Chess (1977) could also be relevant. The model, which describes the fit between parents’ behavior and the child’s temperament, describes how parents manage and respond to their child’s temperament. High goodness of fit means that there is a strong match between the child’s temperament and parents’ expectations, requirements, and behaviors, so that parents can deal better with the challenges of a child’s difficult temperament. This can reduce the stress that parents experience. Low goodness of fit means that the parents find the child’s needs and behaviors very difficult, which can lead to a higher level of stress. The goodness of fit may therefore also play a role in the experience of parenting stress.

Parents of 14-month-old children who reported a higher level of parenting stress also scored high on problematic smartphone use. This is consistent with Chiu (2014), Gökçearslan et al. (2018), and Tu et al. (2023), who also found positive associations between stress and problematic smartphone use independent of parenting. A possible explanation of the association could be that parents with higher stress levels use their smartphones as a coping strategy, for example for social support (Radesky et al., 2016), to reduce feelings of isolation (Coyne et al., 2022), or to take some time for themselves (Wolfers, 2021). Coping is defined as the cognitive and behavioral efforts a person employs to manage internal and external stressful situations (Lazarus and Folkman, 1984). Lazarus and Folkman’s (1984) transactional model of stress and coping focuses on three coping strategies: perception-based coping strategies, problem-based coping strategies, and emotion-based coping strategies. In coping with parenting stress by using a smartphone, the question arises as to what people are doing on their smartphones. Two qualitative studies (Torres et al., 2021; Wolfers, 2021) found that parents primarily use their smartphones for information, self-distraction, and social support. Acquiring information, such as tips on parenting when children’s behavior is challenging, can be assigned to problem-oriented coping. Self-distraction and social support are more emotion-focused coping strategies. However, smartphone-based coping strategies have been found to be less effective in reducing stress than other strategies (Wolfers et al., 2023).

Parent gender did not moderate the association between parenting stress and problematic smartphone use. This means that the effect of parenting stress on problematic smartphone use does not differ between mothers and fathers. Both mothers and fathers may turn to their smartphones as a coping mechanism when experiencing increased parenting stress. Due to social change, fathers – for example in Germany [Bundesministerium für Familie, Senioren, Frauen und Jugend (BMFSFJ), 2021] – are increasingly involved in child care, and in our sample, almost 50% of fathers worked part-time. This means that fathers as well as mothers have to develop coping strategies for parenting stress. We can also assume that the couple relationship influences smartphone behavior and could lead to similar coping strategies in mothers and fathers. McDaniel and Radesky (2018b) found that there is a positive correlation between problematic digital technology use in mothers and in fathers.

Our finding that parent gender did not have a moderating effect only partially agrees with Tu et al.’s (2023) results: They found that among high school students, gender had no moderating effect on stress and information acquisition addiction but did have a moderating effect on stress and social media addiction (significantly higher in females) and game addiction (significantly higher in males). The different operationalizations of problematic smartphone use as well as the age of the students could play a role here.

The total association indicated that difficult child temperament has a significant overall impact on parental problematic smartphone use. However, this significant association was fully mediated by parenting stress. These findings are consistent with McDaniel and Radesky (2018a), who also identified parenting stress as a mediator between children’s behavioral problems and parental smartphone use. Parenting stress can arise from the daily challenge of handling the many tasks and responsibilities in child care as well as other obligations (Diener and Swedin, 2020). Parenting made more challenging due to difficult child temperament could result in parents being less able to satisfy their own needs for competence, autonomy, and relatedness, which in turn increases parenting stress. Parents possibly turn to their smartphone as an escape from this stress and to satisfy their needs at least in the short term, which in addition to other factors can lead to a higher level of problematic smartphone use. As previously mentioned, this study did not find a direct association between difficult child temperament and parental problematic smartphone use. This does not agree with the findings of the Wade-Bohleber et al. (2024) study, in which a partial sample of our study sample also took part. Wade-Bohleber et al. (2024) found a positive association between difficult child temperament and parental problematic smartphone use in the first months of a child’s life. They assumed that parents of a child with a difficult temperament experience less autonomy and social connectedness and that smartphone use has a compensatory function (Wade-Bohleber et al., 2024). According to Ryan and Deci’s (2000) self-determination theory, the need for competence, autonomy, and relatedness are basic human psychological needs. As the children in the Wade-Bohleber et al. (2024) study were considerably younger than the children in our study, this could indicate that over time, the parents had found other ways to satisfy those needs. It could also be that they found no other strategies and that their unsatisfied needs led to more parenting stress, for example feeling more restricted personally. That would explain our finding that parenting stress mediates the association between difficult child temperament and parental problematic smartphone use. In sum, the results show that parenting stress, which can vary depending on the child’s temperament, is a decisive determinant in problematic smartphone use of parents of toddlers.

To our best knowledge, this is the first study to investigate whether parenting stress acts as a mediator between child temperament and parental problematic smartphone use and taking potential gender moderation into account. Paternal data were also considered, which are often neglected in studies on this topic. Another strength of this study is that the questionnaires were provided in both German and English, making it possible for immigrants to participate in the study and increasing the diversity in our sample. In summary, this study represents a unique contribution to our understanding of the determinants of parental problematic smartphone use in early childhood.

We must also mention some limitations of the study. One limitation is the high education level of the study sample: A high percentage of mothers (69%) and fathers (64.7%) had university degrees. However, for the year 2022, statistics collected by the Federal Statistical Office [Bundesamt für Statistik (BfS), 2023] indicate that only 35.5% of women and 34.3% of men aged 35–44 years had a university degree. A higher education level is therefore overrepresented in our sample. In addition, a tendency toward self-selection in the sample cannot be excluded. For example, the parents who decided to participate in the study may have had an interest in and curiosity about the topic of the study as well as a certain awareness regarding smartphone use.

Another limitation is that although cross-sectional studies can identify correlations between variables, they cannot clearly determine the direction of causality. For example, it cannot be excluded that the indirect association between difficult child temperament and parental problematic smartphone use may be bidirectional. This would mean that a higher level of problematic smartphone use leads to more parenting stress, which in turn leads to more negative behavior in the child, such as more negative affect. This behavior could in turn be perceived by parents as difficult child temperament.

In addition, parental problematic smartphone use, difficult child temperament, and parenting stress were measured by parents’ self-report. In general, self-report surveys have the disadvantage that the results may possibly be affected by participants giving socially desirable responses (Paulhus, 2017). We did not measure social desirability bias in this study.

Further, the results of the confirmatory factor analyses indicate that an optimal model fit was not achieved for some constructs. These suboptimal model fits could impair the validity of the results, as they suggest that the measurement models may not fully capture the underlying theoretical constructs. This could affect the reliability and accuracy of the measurements, thereby limiting the interpretability and generalizability of the findings.

This cross-sectional study found that parenting stress fully mediates the positive association between difficult child temperament and problematic smartphone use in parents of 14-month-old children. The findings highlight the importance of considering parenting stress as a central mechanism in understanding the link between child temperament and parents’ smartphone use behaviors.

The results underline that—in both research and practice—parental problematic smartphone use (and smartphone use in general) should not be viewed in isolation. Instead, it should be considered an integrated component of a comprehensive and complex system and analyzed in relation to parents’ broader context. This means that for a comprehensive understanding of parental smartphone use and the development of effective interventions, smartphone use should be examined in the context of life circumstances, stress factors, and stress coping strategies. Furthermore, the increasing digitalization of everyday life and the associated social pressure to always be available should not be overlooked.

These interventions should not primarily convey to parents that they should reduce their smartphone use but should focus instead on how they can manage their stress effectively. Approaches may include promoting mindfulness, facilitating peer support among parents, organizing parent groups, or providing targeted counseling services. Studies have shown, for example, that mothers were able to reduce their parenting stress through a Mindful Parenting Training (Fernandes et al., 2022). The overarching goal is to foster good enough parenting (Winnicott, 1953), thereby supporting both healthy child development and parents’ well-being.

Future studies should employ a bidirectional longitudinal design to clarify the direction of causality. This would help determine whether parental smartphone use also influences children’s behavior in the context of temperament. As both fathers and mothers exhibit problematic smartphone use behaviors in connection with parenting stress, it is important to include both fathers and mothers in future research. Furthermore, targeted guidelines and counseling services could be developed to provide parents with strategies for managing challenging situations in early childhood and effective methods for coping with stress. Adequate stress management is particularly relevant in light of the fact that smartphone use as a coping strategy is less effective in reducing stress than other strategies (Wolfers et al., 2023). Finally, the concept of problematic smartphone use is widely used but lacks a consistent definition. Future research should aim to establish clear and standardized definitions to enhance comparability across studies.