Linking Maternal Problematic Smartphone Use to Maternal Hostility and Warmth Toward Children: A Daily Diary Study of Chinese Families

ABSTRACT

Although the implications of parental smartphone use for parenting have been discussed among researchers and practitioners, prior studies have mostly focused on the between‐person associations of problematic smartphone use with negative parenting behaviors. Therefore, using 5 days of daily diary data, the present study examined whether day‐to‐day fluctuation in maternal problematic smartphone use was associated with day‐to‐day fluctuation in maternal hostility as well as warmth toward children, on the same day and on the following day. Data were collected from mothers of 74 fourth graders (mean age = 9.20 years; age range = 8–10 years; 69% were girls) studying in five elementary schools in Hong Kong, China. On each of five consecutive days, mothers filled in an online questionnaire in the evening on their own smartphone use and their parenting behaviors during the day. Random intercept multilevel linear models within the Bayesian framework revealed that on days when mothers used smartphones more problematically (compared to their cross‐day averages), they showed more hostility and less warmth toward their children (compared to their cross‐day averages) on the same day, even controlling for fluctuation in parenting on the previous day. However, no evidence of lagged‐day associations was found. Theoretically, findings pointed to the potential role of parents' problematic smartphone use in understanding daily variation in both negative and positive parenting behaviors. Practically, findings underscored the possible utility of using community education and intervention programs to help parents monitor their smartphone use and foster positive parenting.

1. Introduction

Equipped with high‐speed Internet and innovative applications, smartphones have become indispensable to nearly every facet of modern life. In fact, as of 2025, there were 5.8 billion smartphone users worldwide, representing a global penetration rate of 70% (Kemp 2026). Importantly, smartphones are designed to adapt to individuals' needs: As a highly personalized technology, smartphones can offer entertainment, information, and resources that fit the unique psychological, cognitive, and social needs of each user (Ratan et al. 2021; Yang et al. 2020). Not surprisingly, smartphone use continues to surge, with emerging research documenting patterns of problematic use (Ratan et al. 2021; Wacks and Weinstein 2021).

Problematic smartphone use refers to excessive engagement with smartphones, typically accompanied by impaired daily functioning (e.g., interference with planned work), emotional dysregulation (e.g., impatience or distress when not being able to use smartphones), uncontrollable urges to use smartphones (e.g., repeated checking even without notification), and a preference for smartphone use over other meaningful activities, such as seeking personal fulfillment, walking in nature, and talking to someone in person (Busch and McCarthy 2021; Ratan et al. 2021). Given smartphones' ability to deliver extensive, instant, and personalized gratification, problematic smartphone use is prevalent. Based on 20 studies involving 12,462 participants from 10 countries, a meta‐analytic review (Humood et al. 2021) found that more than 70% of participants experienced moderate to severe addiction‐like symptoms, such as nomophobia and dependency when disconnected from their devices.

As parents are primary socializers of children, parents' use of smartphones has emerged as a key area of investigation (Frackowiak, Ochs, et al. 2024; Frackowiak, Oleszkowicz, and Łockiewicz 2024). Indeed, surveys indicated that, on average, parents in Japan, Thailand, and the U.S. spend about 2.2 h, 3.2 h, and 3.9 h on smartphones per day, respectively (Chaibal and Chaiyakul 2022; Lu et al. 2017; Yuan et al. 2019). Furthermore, studies with Chinese and Korean parents found that about 10%–23% of participating mothers were at risk for addictive smartphone use (Kim et al. 2021; Lam and Chung 2024). Compared to adolescents, children spend substantially more time with their parents (Lam and McHale 2015). Therefore, it is important to examine the potential implications of parents' use of smartphones for their parenting behaviors during childhood (Beamish et al. 2019).

As postulated by the Couple and Family Technology Framework (Hertlein and Blumer 2014; Wolfers et al. 2020), technology use may both facilitate and disrupt parenting. While sensible use of smartphones can help parents better manage time, organize activities, and identify educational and recreational resources for children, problematic use may reduce parental involvement, distract parents from in‐person interactions with children, and leave parents feeling less patient and more irritable at home. Noteworthily, smartphones are distinguished from other technological devices not only by their extensive functionality, instant responsiveness, and personalized design, but also by their extreme portability (George et al. 2023). Unlike televisions or computers, which are typically stationary, smartphones can be easily accessed at any time, any place. In fact, compared to other forms of media and in‐person conversations with others, smartphones are particularly adept at distracting parents' attention from their children (Lemish et al. 2020; Vanden Abeele et al. 2020). Oftentimes, such distractions result in delayed, inconsistent, and even harsh responses from parents (Solecki 2022). In a now classical qualitative study, Radesky et al. (2014) observed that, in fast‐food restaurants in the U.S., many parents were so absorbed in their smartphones during mealtime that they often ignored their children's benign attempts to gain attention, which then often escalated into misbehaviors.

In line with these views, parental problematic smartphone use has been linked to suboptimal parenting. For example, cross‐sectional studies have linked parental problematic smartphone use to greater parental hostility and more parent–child conflict and (albeit to a lesser extent) lower parental sensitivity and coparenting quality (McDaniel et al. 2018; Shao et al. 2024; Wolfers et al. 2020; Zhu et al. 2025). Moreover, longitudinal studies have linked parental problematic smartphone use to lower parental monitoring and supervision, poorer parent–child attachment, and more parent–child conflict over time (Matthes et al. 2021; Xie et al. 2025).

Though important, existing literature has focused on the between‐person associations of problematic smartphone use with negative parenting behaviors (Beamish et al. 2019; Braune‐Krickau et al. 2021), that is, whether parents who exhibit higher levels of problematic smartphone use than other parents tend to exhibit more negative parenting behaviors than other parents. In contrast, within‐person associations focus on whether parents may exhibit more negative parenting behaviors (compared to their usual selves) when they have more problematic use of smartphones (compared to their usual selves). As parents are compared to their own selves, within‐person associations are not confounded by time‐invariant factors that remain stable over the study period, such as child gender and personality, parental work status and education backgrounds, and family incomes and household conditions (Curran and Bauer 2011).

Very few studies have differentiated between‐person differences (i.e., stability) from within‐person variation (i.e., fluctuation) in smartphone use (Beierle et al. 2020). One exception was Marciano et al.'s (2022) study, which used 45 days of daily diary data from 82 adolescents to test the within‐person associations of general smartphone use with individual well‐being. Another exception was Rodman et al.'s (2024) study, which used 12 months of monthly observation data from 26 adolescents to test the within‐person associations of general smartphone use with positive affect. The within‐person association of parents' smartphone use and their parenting behaviors remains underexplored, however.

1.1. The Present Study

To recap, prior research has focused on the between‐person associations of problematic smartphone use with negative parenting behaviors (McDaniel et al. 2018; Shao et al. 2024; Wolfers et al. 2020; Zhu et al. 2025). Expanding on this work, the present study used 5 days of daily diary data collected from the mothers of 74 fourth graders from Hong Kong, China, to examine both same‐day and lagged‐day within‐person associations of maternal problematic smartphone use with maternal hostility and warmth toward children. Same‐day and lagged‐day within‐person associations are conceptually distinct: same‐day associations capture concurrent co‐fluctuation or short‐term daily coupling between problematic smartphone use and parenting, whereas lagged‐day associations capture more enduring spillover or carry‐over processes from 1 day to the following day (Bolger et al. 2003).

As mothers only needed to record what had happened during the day, a daily diary approach might reduce mothers' memory bias in reporting (Lischetzke and Könen 2024). Moreover, as children tend to respond spontaneously to their mothers' use of smartphones (Stockdale et al. 2020), a daily diary approach might better capture the interplay of maternal problematic smartphone use and maternal hostility and warmth as it unfolds in daily life. We opted to collect data on 5 consecutive days, as Wilcockson et al. (2018) recommended that a 5‐day approach can adequately reflect the typical smartphone use pattern of an individual. We focused on maternal hostility and warmth, as they may constitute two parenting dimensions that are particularly vulnerable to technology‐related interference. Moreover, smartphone‐related unavailability, distraction, and irritability may be especially likely to reduce warm involvement and increase hostile responses among parents (Shao et al. 2024; Solecki 2022; Stockdale et al. 2020). Furthermore, increasing research with Chinese families shows that negative and positive aspects of parenting may explain unique variance in child adjustment (Chen et al. 2022; Lam et al. 2018; Lam and Chung 2023).

We focused on mothers of fourth‐grade students, as preteens aged 8–10 years are sufficiently mature to perceive parental problematic smartphone use as a sign of distraction or neglect (Frackowiak, Ochs, et al. 2024; Frackowiak, Oleszkowicz, and Łockiewicz 2024) but not nonchalant enough to disregard parental attention—as many adolescents do when they increasingly shift their orientations toward their peers (Lam and McHale 2015). Additionally, mothers are more involved in childrearing than other family members (Yaffe 2023), and problematic smartphone use has been rising among middle‐aged adults (Kim et al. 2024). Therefore, the implications of maternal problematic smartphone use may be particularly pronounced for preteens (Beamish et al. 2019). Meanwhile, in Hong Kong, fifth and sixth graders are busy preparing for high‐stakes internal assessments that may determine their high school placements (Education Bureau 2024). Therefore, focusing on mothers of fourth graders in a daily diary study was more feasible in the Hong Kong context.

Based on prior research, we hypothesized that daily fluctuation in maternal problematic smartphone use would be associated positively with daily fluctuation in maternal hostility and negatively with daily fluctuation in maternal warmth, on the same day and on the following day. Although within‐person associations account for time‐invariant factors, they may still be affected by time‐varying factors (Curran and Bauer 2011). For example, when parents show more negative or positive parenting behaviors than usual (Wang et al. 2024), these behaviors may persist into the following day. Therefore, we controlled for the autocorrelations of maternal hostility and warmth on the previous day to rule out temporal persistence as an alternative explanation.

2. Method

2.1. Participants and Procedures

Participants were the mothers of 74 Chinese children studying in five elementary schools in Hong Kong, China. Using publicly available contact information, we sent invitation letters to all elementary schools in Hong Kong and made follow‐up phone calls (in random order) to explain our research goals and procedures. As shown in simulation studies (Maas and Hox 2005; McNeish 2014), when the highest‐level sample size exceeds 10, five lower‐level units can yield precise estimates of within‐person variance. Meanwhile, at the highest level of multilevel models, at least 50 clusters are necessary to detect medium effect sizes with robust estimation of between‐person variance. Therefore, assuming that at least 15 mothers per school would participate in our daily dairy study, we stopped recruiting once five elementary schools agreed to participate and returned written informed consent. Through these five participating schools, we sent invitation letters to all families with fourth graders (typically 8–9 year‐olds). Eventually, 90 mothers provided written informed consent and agreed to participate in the 5‐day daily diary study. On each of five consecutive days, mothers received a text reminder in the evening to fill in an online questionnaire on their own smartphone use and their parenting behaviors during the day. Only 74 mothers reported at least 1 day of data (i.e., one observation). The total number of observations based on these 74 mothers was 287, and the compliance rate was 78%. As a token of appreciation, each mother received a supermarket coupon of HK$100 (about US$13). The research procedures were approved by The Education University of Hong Kong's Human Research Ethics Committee.

Mothers varied in age (2% aged 30 years or younger, 43% between 31 and 40 years, 55% between 41 and 50 years). Mothers also varied in education (35% completed secondary school only, 24% completed higher diploma or associate degree programs only, and 41% completed college degree programs or above). According to the Census and Statistics Department (2024), 39% of all women aged between 30 and 49 years in Hong Kong had completed college degree programs. Therefore, mothers in our sample appeared to be better educated than their same‐age peers in the population, possibly reflecting that better‐educated individuals are more willing to participate in studies with more complex designs (Lischetzke and Könen 2024). The mean age of children was 9.20 years (range = 8–10 years, SD = 0.46), and 69% of them were girls.

2.2. Measures

2.2.1. Problematic Smartphone Use

Problematic smartphone use was measured using the 10‐item Smartphone Addiction Scale (Kwon et al. 2013). On each of five consecutive days, mothers used a 4‐point scale ranging from 0 (strongly disagree) to 3 (strongly agree) to rate their smartphone use during the day. Sample items were “I miss planned work due to smartphone use” and “I feel impatient and fretful when I am not holding my smartphone.” Ratings were averaged per day, with higher scores indicating higher levels of problematic smartphone use. The reliability and validity of this measure had been evidenced among Chinese mothers (Lam and Chung 2024). Compared to the more commonly reported Cronbach's alphas, McDonald's omegas—which examine the reliability of measures at both the within‐ and between‐person levels—are more suitable for multilevel models (Geldhof et al. 2014). In the present study, McDonald's omegas were 0.77 and 0.90 at the within‐ and between‐person levels, respectively.

2.2.2. Hostility and Warmth Toward Children

Hostility and warmth toward children were measured, respectively, using the 3‐item Aversive Parent Behavior subscale and the 3‐item Positive Parent Involvement subscale from the Youth Everyday Social Interaction and Mood Scale (Repetti 1996). On each of 5 consecutive days, mothers used a 4‐point scale ranging from 0 (never) to 3 (all the time) to rate their parenting behaviors during the day. Sample items were “I get mad at my child today” (hostility) and “I have fun with my child today” (warmth). Ratings were averaged separately per day for each subscale, with higher scores indicating higher levels of hostility and warmth. The reliability and validity of this measure had been evidenced among Chinese parents (Luo et al. 2025). In the present study, McDonald's omegas for hostility were 0.72 and 0.89 at the within‐ and between‐person levels, respectively. For warmth, McDonald's omegas were 0.71 and 0.93 at the within‐ and between‐person levels, respectively.

2.3. Data Analyses

We conducted data analyses in two steps. First, using SPSS 28.0, we computed the person‐specific means of problematic smartphone use, hostility, and warmth by averaging the daily scores for each mother. We then computed the descriptive statistics of these person‐specific data.

Second, using Mplus 8.4, we ran random intercept multilevel linear models within a Bayesian framework (Asparouhov and Muthén 2016; Hamaker et al. 2015) to examine whether maternal problematic smartphone use was linked to maternal hostility and warmth toward children, at the within‐ and between‐person levels. To reduce the risk of Type I errors due to multiple testing, we included both hostility and warmth as simultaneous outcome variables in the same model. Intraclass correlation coefficients (ICCs) indicated that significant variance existed at the person level for problematic smartphone use (ICC = 0.39), parenting hostility (ICC = 0.72), and parenting warmth (ICC = 0.47). Moreover, significant variance existed at the time level for problematic smartphone use (ICC = 0.61), parenting hostility (ICC = 0.29), and parenting warmth (ICC = 0.52). Therefore, to account for the clustered structure of the data (i.e., daily observations correlated within each mother), we fitted two‐level models, with Level 1 being Time (i.e., day) and Level 2 being Person (i.e., mother). To differentiate between the Level 1 (within‐person) variation and Level 2 (between‐person) differences of problematic smartphone use (Asparouhov and Muthén 2016), we estimated it as both a latent person‐mean centered variable (i.e., centered around the cross‐day average of each person) and a latent grand‐mean centered variable (i.e., centered around the overall sample mean).

We fitted same‐day and lagged‐day multilevel models. In the same‐day model, the within‐person variable captured whether the daily fluctuation in problematic smartphone use (from the cross‐day average) at Time t was associated with the daily fluctuation in parenting behaviors (from the cross‐day average) at Time t. In contrast, in the lagged‐day model, the within‐person variable captured whether the daily fluctuation in problematic smartphone use at Time t was associated with the daily fluctuation in parenting behaviors at Time t + 1. However, in both same‐day and lagged‐day models, parenting behaviors at Time t were regressed on parenting behaviors at Time t − 1 to rule out temporal persistence as an alternative explanation. Moreover, for both models, at Level 2, we tested the main effect of between‐person problematic smartphone use on the random intercept of parenting behaviors. With person‐mean centering, the random intercept at Level 2 reflected each mother's average parenting behaviors on days when their smartphone use was at their own person‐specific average. Therefore, the between‐person effect captured whether mothers' averages of problematic smartphone use (across Times) were associated with their averages of parenting behaviors (across Times).

Tests for normality of the outcome variables indicated that warmth was normally distributed (Kolmogorov–Smirnov statistic = 0.06, n.s.), but not hostility (Kolmogorov–Smirnov statistic = 0.17, p < 0.001). Therefore, we estimated the parameters using a Bayesian estimator with non‐informative priors and a Markov Chain Monte Carlo (MCMC) algorithm, which yield robust estimations even under violations of normality assumptions (Kruschke 2010; Van de Schoot et al. 2014). To allow for individual variability around the fixed effects, we estimated three random effects at Level 1, including random intercepts of parenting behaviors (which captured between‐person differences in average parenting behaviors), random autoregressive effects (which captured between‐person differences in the temporal persistence of parenting behaviors), and random slopes of problematic smartphone use (which captured between‐person differences in the magnitude of relationship between daily problematic smartphone use and parenting behaviors). We tested only the effect of average problematic smartphone use on the random intercept of parenting to differentiate within‐person from between‐person associations. We did not test its effects on the random autoregressive effects or random slopes, as explaining between‐person differences in these parameters was not a focus of the present study. Figures 1 and 2 show the path diagrams for the same‐day and lagged‐day models, respectively. The model equations are presented in Supporting Information.

FIGURE 1.

Path model for the same‐day multilevel model. Level 1 (Time‐level) setup. Level 2 (Person‐level) setup. The subscript t = Time; the subscript i = Person. Residual variances are not shown for clarity.

FIGURE 2.

Path model for the lagged‐day multilevel model. Level 1 (Time‐level) setup. Level 2 (Person‐level) setup. The subscript t = Time; the subscript i = Person. Residual variances are not shown for clarity.

3. Results

Table 1 presents the means, standard deviations, and ranges of the person‐specific averages of variables as well as the within‐ and between‐person correlations derived from the multilevel model. Based on the means of the person‐specific averages, our sample appeared to be relatively well‐adjusted, with problematic smartphone use and hostility on the lower ends and warmth on the higher end of the measures. Based on the correlations, at the within‐person level, problematic smartphone use was associated positively with hostility (r = 0.23, p < 0.001) and negatively with warmth (r = −0.21, p < 0.01). At the between‐person level, however, problematic smartphone use was not correlated with hostility (r = 0.21, n.s.) or warmth (r = −0.21, n.s.).

TABLE 1.

Means, standard deviations, and ranges of person‐specific variables and within‐ and between‐person correlations among variables.

	Ms (SDs)	Min	Max	Correlations
				(1)	(2)	(3)
(1) Problematic smartphone use	0.63 (0.39)	0.02	2.11	—	0.23***	−0.21**
(2) Hostility toward children	0.29 (0.26)	0.00	1.00	0.21	—	−0.26***
(3) Warmth toward children	2.07 (0.45)	1.00	3.00	−0.21	−0.44**	—

Note: Within‐ and between‐person correlations are presented above and below the diagonal, respectively.

Abbreviations: Ms, means; SDs, standard deviations.

^** p < 0.01.

^*** p < 0.001.

3.1. Same‐Day Multilevel Model

Table 2 presents the unstandardized fixed and random effects, along with their corresponding 95% credible intervals (CIs), of the same‐day multilevel model. Results indicated that the fixed effect of problematic smartphone use at Level 1 was 0.24 (95% CI = [0.06, 0.39], not including a zero) for hostility, suggesting that fluctuation in maternal problematic smartphone use from the cross‐day average was positively associated with fluctuation in hostility from the cross‐day average, even when fluctuation in hostility on the previous day was controlled for. In other words, on days when mothers used smartphones more problematically (than usual), mothers showed more hostility toward their children (than usual) on the same day. At Level 2, the fixed effect of maternal problematic smartphone use was 0.11 (95% CI = [−0.06, 0.30], including a zero), suggesting that mothers who generally used smartphones more problematically (than do the rest of the sample) did not generally show more hostility toward their children (than do the rest of the sample). Overall, the model explained 10% of the within‐person variance in hostility, indicating a modest effect size (Ferguson 2016).

TABLE 2.

Unstandardized effects and 95% credible intervals of same‐day multilevel model with full sample (N = 74).

Parameters		Hostility toward children			Warmth toward children
		Posterior median	95% CI: lower	95% CI: upper	Posterior median	95% CI: lower	95% CI: upper
Fixed effects (means)
Problematic smartphone use means	${\gamma }_{30}$	0.67	0.59	0.76
Parenting intercepts	${\gamma }_{00}$	0.20	0.06	0.33	2.16	1.91	2.44
Autoregressive effects
Parenting (t − 1) → Parenting (t)	${\gamma }_{10}$	0.07	−0.07	0.22	0.02	−0.13	0.16
Main effects
Problematic smartphone use (t) (within‐person)	${\gamma }_{20}$	0.24	0.06	0.39	−0.27	−0.53	−0.04
Problematic smartphone use (t) (between‐person)	${\gamma }_{01}$	0.11	−0.06	0.30	−0.13	−0.48	0.22
Random effects (variances)
Problematic smartphone use means	$u_{3i}$	0.12	0.08	0.18
Parenting intercepts	$u_{0i}$	0.04	0.02	0.06	0.17	0.11	0.26
Autoregressive effects
Parenting (t − 1) → Parenting (t)	$u_{1i}$	0.01	0.01	0.10	0.08	0.02	0.20
Main effect
Problematic smartphone use (t) (within‐person)	$u_{2i}$	0.03	0.01	0.19	0.15	0.01	0.44

Abbreviations: CIs, 95% credible intervals; γs, unstandardized effects.

Focusing on warmth, the fixed effect at Level 1 was −0.27 (95% CI = [−0.53, −0.04], not including a zero), suggesting that on days when mothers used their smartphones more problematically (than usual), they showed less warmth toward their children (than usual) on the same day. However, the fixed effect at Level 2 was −0.13 (95% CI = [−0.48, 0.22], including a zero), suggesting that mothers who generally used smartphones more problematically (than do the rest of the sample) did not generally show less warmth toward their children (than do the rest of the sample). Overall, the model explained 19% of the within‐person variance in warmth, indicating a modest‐to‐moderate effect size (Ferguson 2016).

3.2. Lagged‐Day Multilevel Model

Table 3 presents the unstandardized fixed and random effects of the lagged‐day multilevel model. The fixed effects of problematic smartphone use at Level 1 were −0.04 (95% CI = [−0.29, 0.20], including a zero) for hostility and −0.09 (95% CI = [−0.31, 0.16], including a zero) for warmth, suggesting that on days when mothers used their smartphones more problematically (than usual), they did not show more hostility or less warmth toward children on the following day. The fixed effects at Level 2 were 0.14 (95% CI = [−0.05, 0.35], including a zero) for hostility and −0.18 (95% CI = [−0.54, 0.17], including a zero) for warmth, suggesting that mothers who generally used smartphones more problematically (than do the rest of the sample) did not generally show more parenting hostility or less warmth (than do the rest of the sample).

TABLE 3.

Unstandardized effects and 95% credible intervals of lagged‐day multilevel model with full sample (N = 74).

Parameters		Hostility toward children			Warmth toward children
		Posterior median	95% CI: lower	95% CI: upper	Posterior median	95% CI: lower	95% CI: upper
Fixed effects (means)
Problematic smartphone use (t − 1) means	${\gamma }_{30}$	0.69	0.60	0.77
Parenting intercepts	${\gamma }_{00}$	0.18	0.02	0.33	2.20	1.93	2.47
Autoregressive effects
Parenting (t − 1) → Parenting (t)	${\gamma }_{10}$	0.07	−0.08	0.23	0.03	−0.10	0.16
Main effects
Problematic smartphone use (t − 1) (within‐person)	${\gamma }_{20}$	−0.04	−0.29	0.20	−0.09	−0.31	0.16
Problematic smartphone use (t − 1) (between‐person)	${\gamma }_{01}$	0.14	−0.05	0.35	−0.18	−0.54	0.17
Random effects (variances)
Problematic smartphone use (t − 1) means	$u_{3i}$	0.11	0.07	0.17
Parenting intercepts	$u_{0i}$	0.04	0.02	0.06	0.17	0.11	0.28
Autoregressive effects
Parenting (t − 1) → Parenting (t)	$u_{1i}$	0.02	0.01	0.11	0.07	0.02	0.18
Main effect
Problematic smartphone use (t − 1) (within‐person)	$u_{2i}$	0.15	0.01	0.41	0.07	0.02	0.18

Abbreviations: CIs, 95% credible intervals; γs, unstandardized effects.

3.3. Sensitivity Analyses

To ensure that our findings were not due to our use of a particular sample or a particular analytic method, we conducted sensitivity analyses. First, among the 74 participating mothers, seven provided only a single day of data. As participants with a single observation may not meaningfully contribute to the estimation of autoregressive effects or within‐person variance, we excluded these seven participants and refitted the same‐day and lagged‐day multilevel models using the reduced sample (N = 67). The results were substantively the same as those based on the full sample (N = 74), as presented as Supporting Information. To maximize our statistical power (for the between‐person associations) and the generalizability of our findings, however, we opted to present the results based on the full sample of 74 mothers as the main analyses.

Second, to examine the potential impact of the violation of assumptions of normality, we re‐estimated both the same‐day and lagged‐day models using the full sample with two alternative estimators: (a) standard maximum likelihood estimator, which assumes normality, and (b) robust maximum likelihood estimator, which is robust to mild violations of normality assumptions (Finney and DiStefano 2013). The substantive findings were consistent across the original (i.e., a Bayesian estimator with non‐informative priors and a MCMC algorithm) versus the two alternative estimation methods, supporting the stability of our findings. Therefore, we continued to report findings based on our original estimation method.

4. Discussion

This study examined the same‐day and lagged‐day within‐person associations of maternal problematic smartphone use with maternal hostility and warmth toward children. A number of strengths of our study were noteworthy: First, by using a daily diary approach, we reduced memory bias inherent in self‐report measures and more precisely captured the day‐to‐day relationship between problematic smartphone use and parenting behaviors (Lischetzke and Könen 2024). Second, by comparing mothers to their usual selves, we controlled for factors that remained stable across the 5‐day period of data collection, ruling out these time‐invariant factors as alternative explanations (Curran and Bauer 2011). Additionally, by controlling for the autocorrelation of maternal hostility and warmth as time‐varying covariates, we ruled out the temporal persistence of parenting behaviors as an alternative explanation (Wang et al. 2024). Finally, by focusing on both maternal hostility and warmth as outcome variables, we highlighted the multifaceted role of problematic smartphone use in understanding both negative and positive aspects of parenting, which may uniquely contribute to Chinese children's adjustment (Chen et al. 2022; Lam et al. 2018; Lam and Chung 2023).

As problematic smartphone use is a negative construct, prior research has primarily focused on its links with negative family processes and negative indicators of individual adjustment (Lemish et al. 2020; Vanden Abeele et al. 2020). Our findings extended this work by demonstrating that problematic smartphone use was associated with not only the presence of deficiency (e.g., hostility) but also the absence of strength (e.g., warmth) during parent–child interactions on the same day. Given these associations, a logical next step is to extend our study (which did not measure any mediators) by exploring the underlying mechanism. Future researchers may, for example, examine how problematic smartphone use may deplete parents' cognitive and emotional resources, which in turn may impair parents' abilities to manage anger and provide responsive parenting to their children (Draheim et al. 2022).

4.1. Associations of Problematic Smartphone Use With Parenting Behaviors

Consistent with our hypotheses, our results indicated that, on days when mothers used their smartphones more problematically than usual, they showed more hostility and less warmth toward their children than usual on the same day. These results were consistent with the Couple and Family Technology Framework (Hertlein and Blumer 2014; Wolfers et al. 2020), which posits that problematic smartphone use may contribute to less‐than‐optimal or hostile responses among parents toward their children.

Although prior studies have used within‐person designs to examine the association between smartphone use and individual adjustment (Marciano et al. 2022; Rodman et al. 2024), scarce research has examined the within‐person associations of parents' problematic smartphone use with their parenting behaviors. Importantly, daily diary methods represent only one kind of within‐person designs. Experience sampling (i.e., participants randomly prompted to provide data within and across days), weekly diaries, monthly diaries, and annual data spanning numerous years can also allow for within‐person analyses. However, compared to these within‐person designs, our end‐of‐day daily diary assessment might be more “balanced” in that it preempted response burden imposed by experience sampling and reduced retrospective bias inherent in longer‐term longitudinal designs. Given the statistical advantages of within‐person associations (Curran and Bauer 2011), future investigators may consider using more within‐person designs to examine parents' problematic smartphone use. One compelling research question, for example, concerns which time frame (e.g., momentary, daily, weekly, monthly, or yearly) may best capture the within‐person associations of parents' problematic smartphone use with their well‐being and their family relationships.

Inconsistent with our hypotheses, our study did not detect lagged‐day associations of problematic smartphone use with parenting behaviors. One possible explanation was that (as noted above) a different time frame is needed to capture the longer‐term implications of problematic smartphone use for parent–child interactions. Additional research is needed to examine whether day‐to‐day smartphone use may, over weeks, months, or years, gradually develop into a more stable pattern that leaves a detectable mark on family routines. Another possible explanation was that, by linking problematic smartphone use to parenting behaviors (at Times 1–4) on the following day (at Times 2–5), the lagged‐day model was based on truncated data and thus fewer observations. This analytical specification further reduced the effective sample size, which might have limited our statistical power to detect meaningful lagged‐day associations. Future studies should re‐examine the lagged association between problematic smartphone use and parenting behaviors with more assessment occasions at the Time level.

Although between‐person associations were not a focus of our study, maternal problematic smartphone use was not linked to maternal hostility or warmth toward children at the between‐person level. One plausible reason for the null findings was that parents in our sample were requested to report their parenting behaviors during the day (e.g., “I get mad at my child today,” “I have fun with my child today”) and an aggregation of these day‐specific reports (i.e., cross‐day averages) might not adequately capture the overarching “parenting styles.” The between‐person association of parents' smartphone use and their parenting practices may be more readily detected by studies using more global measures of parenting (Matthes et al. 2021; McDaniel et al. 2018; Shao et al. 2024; Wolfers et al. 2020; Xie et al. 2025; Zhu et al. 2025). Another plausible reason was that our sample size was relatively small at the Person level (N = 74), which might have limited our ability to detect between‐person associations. Indeed, even bivariate correlations indicated that problematic smartphone use was not significantly associated with parenting behaviors at the between‐person level. To better capture both short‐term fluctuation (i.e., within‐person process) and long‐term trait‐like features (i.e., between‐person process) of problematic smartphone use, further investigators may use multi‐timescale measures with larger samples to retest the associations of problematic smartphone with parenting practices.

Our findings should not be interpreted to mean that smartphone use is inherently problematic. Indeed, as acknowledged by the Couple and Family Technology Framework (Hertlein and Blumer 2014; Wolfers et al. 2020), a more balanced use of smartphones that supports other meaning activities may actually facilitate parenting by promoting emotional connections. However, our focus on problematic smartphone use, which involves uncontrollable impulses to use smartphones (Busch and McCarthy 2021; Ratan et al. 2021), also implies that we focused on the potential negative impact of smartphone use. Instead of invariably positioning smartphone use as problematic, future researchers may explore how its implications may be moderated by usage intensity, underlying motives, or specific application categories.

4.2. Practical Implications

Considering the significant within‐person associations of maternal problematic smartphone use with maternal hostility and warmth toward children, community education may be launched to inform parents of the nature of problematic smartphone use and its potential negative implications. Intervention programs may also be designed to assist parents in monitoring their own use of smartphones and maintaining a balance between virtual interactions with devices versus in‐person interactions with people. Effective strategies may include increasing parents' awareness of their smartphone use frequency and duration, helping parents set action plans to cope with automatic impulses to use smartphones, and encouraging parents to fully engage in present‐moment experiences with their children.

4.3. Limitations and Conclusions

Our study had several limitations. First, due to time and resource constraints, we were only able to recruit mothers of children from five elementary schools in Hong Kong. Additionally, mothers in our sample were relatively well‐adjusted and better educated than their same‐age peers in the Hong Kong population. Furthermore, our sample consisted exclusively of Chinese families living in Hong Kong, a megacity with distinct cultural, technological, and relational contexts. Therefore, our findings may or may not be generalizable to other cultural contexts or to other parts of China, particularly rural villages and smaller towns. Future investigators should retest our hypotheses using representative samples of families across different cultural contexts and from diverse regions of China.

Second, we assessed problematic smartphone use with self‐reported questionnaires, a subjective measure of smartphone use. However, evidence indicates that subjective (e.g., questionnaire ratings) versus objective (e.g., screen time from smartphone records) measures may tap onto different aspects of problematic smartphone use, with the former reflecting perceived difficulty in refraining from using smartphones in general and the latter quantifying time spent on various types of applications (Marciano et al. 2022). Assessing problematic use on specific smartphone applications may be particularly important, as different types of smartphone use (e.g., social versus non‐social, obligatory versus voluntary) may be linked to family dynamics in different ways (Rodman et al. 2024). For example, a mother addressing a work emergency using her smartphone may convey different emotional cues and elicit distinct child responses compared to a mother aimlessly scrolling online. Therefore, combining both subjective and objective measures may best operationalize problematic smartphone use. Additional studies should use both measures to assess application‐specific problematic use to examine their unique contributions to parenting.

Third, with the squared association index being > 4% (Ferguson 2016), our model explained a practically significant amount of variance in parental hostility and warmth. However, the effect sizes documented were modest to moderate. In other words, problematic smartphone use is likely only one of different factors that may shape momentary parenting behaviors. In fact, prior theory and research (Hertlein and Blumer 2014; McDaniel et al. 2018; Qiao et al. 2025; Sebre et al. 2024) have indicated that parenting may be shaped by the interplay of parental (e.g., technology use), child (e.g., behavioral responses), and environmental (e.g., reactions of others in the surroundings) factors. A critical avenue for future investigation involves using a comprehensive approach to simultaneously consider the roles of parents, children, and contexts in understanding momentary parenting behaviors.

Fourth, we only tested the effect of average problematic smartphone use on the random intercept of parenting behaviors at the Person Level, as our relatively small sample size did not allow for a comprehensive examination of potential between‐person moderators. Considering that between‐person variables, such as personality and emotion regulation patterns, may be implicated in understanding the implications of problematic smartphone use (Yeshua and Berger 2026), future research should use larger samples to explore the potential roles of between‐person variables in explaining the association of problematic smartphone use with parenting behaviors. Relatedly, although a 5‐day design can adequately capture an individual's typical smartphone use patterns (Wilcockson et al. 2018), statistical power is a linear function of the number of observations. Therefore, five observations per person might have posed constraints on the estimation of other within‐person dynamics, including random slopes of problematic smartphone use, autoregressive effects, and lagged‐day associations. Future research should use more repeated assessment occasions to fully capture different within‐person processes.

Finally, we focused on mothers, as they continue to be the primary caregivers of children and adolescents (Yaffe 2023). However, as fathers are becoming increasingly involved in childrearing (Cabrera et al. 2018; Lam and McHale 2015), additional work should be directed at collecting dyadic data from both mothers and fathers to examine how problematic smartphone use may affect their own and their partners' parenting behaviors. Moreover, as noted above, parenting behaviors may also be affected by child characteristics (Sebre et al. 2024). Considering that many elementary school students now have their own smartphones and that maternal problematic smartphone use was associated with parental hostility and warmth at a within‐person level, future investigators should explore if child problematic smartphone use may have similar patterns of associations with parental hostility and warmth.

Despite these limitations, our study offered significant theoretical and practical implications. Theoretically, our findings pointed to the role of parental problematic smartphone use in understanding day‐to‐day fluctuations in negative and positive aspects of parenting. Practically, our findings underscored the possible utility of using community education and intervention programs to help parents monitor their smartphone use and foster positive parenting.

Funding

This study was supported by a grant by Research Grants Council, University Grants Committee, Hong Kong, China, to Chun Bun Lam (GRF 18620022).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Unstandardized effects and 95% credible intervals of same‐day multilevel model with reduced sample (N = 67).

Table S2: Unstandardized effects and 95% credible intervals of lagged‐day multilevel model with reduced sample (N = 67).

Data Availability Statement

The data and analysis codes used in this study are not publicly available but are available from the corresponding author upon reasonable request.