Original Paper
Abstract
Background: Contrary to popular concerns about the harmful effects of media use on mental health, research on this relationship is ambiguous, stalling advances in theory, interventions, and policy. Scientific explorations of the relationship between media and mental health have mostly been found null or have small associations, with the results often blamed on the use of cross-sectional study designs or imprecise measures of media use and mental health.
Objective: This exploratory empirical demonstration aims to answer whether mental health effects are associated with media use experiences by (1) redirecting research investments to granular and intensive longitudinal recordings of digital experiences to build models of media use and mental health for single individuals over the course of 1 year, (2) using new metrics of fragmented media use to propose explanations of mental health effects that will advance person-specific theorizing in media psychology, and (3) identifying combinations of media behaviors and mental health symptoms that may be more useful for studying media effects than single measures of dosage and affect or assessments of clinical symptoms related to specific disorders.
Methods: The activity on individuals’ smartphone screens was recorded every 5 seconds when devices were in use over 1 year, resulting in a dataset of 6,744,013 screenshots and 123 fortnightly surveys from 5 adult participants. Each participant contributed between 0.8 and 2.7 million screens. Six media use metrics were derived from smartphone metadata. Fortnightly surveys captured symptoms of depression, attention-deficit/hyperactivity disorder, state anxiety, and positive affect. Idiographic filter models (p-technique canonical correlation analyses) were applied to explore person-specific associations.
Results: Canonical correlations revealed substantial person-specific associations between media use and mental health, ranging from r=0.82 (P=.008) to r=0.92 (P=.03). The specific combinations of media use metrics and mental health dimensions were different for each person, reflecting significant individual variability. For instance, the media use canonical variate for 1 participant was characterized by higher loadings for app-switching, which, in combination with other behaviors, correlated strongly with a mental health variate emphasizing anxiety symptoms. For another, prolonged screen time, alongside other media use behaviors, contributed to a mental health variate weighted more heavily toward depression symptoms. These within-person correlations are among the strongest reported in this literature.
Conclusions: Results suggest that the relationships between media use and mental health are highly individualized, with implications for the development of personalized models and precision smartphone-informed interventions in mental health. We discuss how our approach can be extended generally, while still emphasizing the importance of idiographic approaches. This study highlights the potential for granular, longitudinal data to reveal person-specific patterns that can inform theory development, personalized screening, diagnosis, and interventions in mental health.
doi:10.2196/59875
Keywords
Introduction
Over the last decade, researchers have conducted extensive studies involving hundreds of thousands of participants to understand the relationship between media use and mental health. Despite significant investments, findings remain ambiguous, often reporting null or small associations [-]. This lack of clarity undermines progress in theory development, public policy, and clinical interventions. Researchers have proposed solutions, including granular tracking of media use [-], longitudinal study designs [,], the inclusion of at-risk populations [-], and person-specific analytics to account for individual differences [,].
A central challenge in this research is the problem of idiosyncrasy—the substantial variability in how individuals use media and how it affects their mental health [,]. In psychology, the opportunity to reclaim idiosyncratic variance has been defined as one of identifying idiographic filters for nomothetic interests []. The core argument is that the conventional practice of averaging out idiosyncrasy is neither theoretically satisfactory nor practically useful. We explore if and how an idiographic filter approach might enhance this research about media use and mental health. While researchers often seek universal rules to inform public policies or clinical interventions, such generalizations obscure the reality that individuals’ media experiences are inherently personal and shaped by unique combinations of content and context. Aggregate analyses, which average out these differences, are insufficient and often misleading. Similarly, mental health symptom clusters frequently cut across traditional diagnostic boundaries, further emphasizing the need for individualized approaches. Addressing this challenge requires an idiographic perspective, which preserves individual variability and analyzes person-specific patterns rather than treating them as noise. By adopting an idiographic approach, research can uncover meaningful insights into the unique configurations of media use and mental health for each person.
To address this challenge, we used p-technique canonical correlation analysis (pCCA) [], a method designed to analyze within-person relationships over time. This approach creates idiographic filters [] that highlight how core constructs, such as media use and mental health metrics, manifest differently for each participant. Unlike traditional methods that assume consistency across individuals, pCCA uncovers the specific patterns and dynamics unique to each person based on the common set of variables. For example, media use heavily characterized by frequent app-switching may correlate with symptoms of hyperactivity for 1 individual, while prolonged social screen time may have significant weight in describing the changes in symptoms of depression for another. By prioritizing individual-level insights, this method provides a pathway to more personalized interventions ( [-,,-,-]).
This study uses granular longitudinal data collected through the Human Screenome Project [,], which recorded smartphone activity every 5 seconds for 1 year, resulting in over 6.7 million screenshots and 123 surveys across 5 participants. Mental health metrics, including symptoms of depression [], attention-deficit/hyperactivity disorder (ADHD) [], state anxiety [], and positive affect were assessed fortnightly. Using pCCA, we explored person-specific associations between media use and mental health to address the following questions.
- Can granular longitudinal recordings reveal associations between media use and mental health that differ from the null or small effects reported in aggregated studies?
- Can new metrics, such as fragmentation of media use, explain person-specific mental health effects and advance media psychology theory?
- Can combinations of mental health symptoms provide better insights into media effects than single measures of dosage or individual clinical symptoms?
By focusing on individual experiences and variability over time, this study offers a novel approach to understanding the relationship between media use and mental health, emphasizing the potential of person-specific data to inform theory development and precision interventions.
Methods
Procedure and Participants
This study was conducted in the United States, leveraging a longitudinal design to assess smartphone use and its relationships with mental health. Data were collected via a custom smartphone app and biweekly participant surveys, enabling the integration of behavioral data with mental health measures. Participants were recruited via the Qualtrics research panel as part of a larger research project on assessing digital experiences with smartphone data []. Eligibility criteria included residing in the United States, being the sole user of an Android phone, and providing informed consent and Health Insurance Portability and Accountability Act (HIPAA) authorization.
Participants installed a custom study app on their smartphones, enabling an unobtrusive collection of smartphone screens. The app required no active user management and operated transparently, with a status bar icon indicating its activity. Screenshots, along with operating system metadata like phone battery state and foreground app were captured and recorded automatically every 5 seconds when the screen was active.
Our study selected a focused sample that satisfied the minimum requirements for robust within-person analysis and allowed for a demonstration of our methodology, prioritizing long time series from a small number of participants over larger aggregate datasets so that we might uncover detailed person-specific insights and enable future scalability through computational methods [,,]. We conducted a detailed analysis of 5 participants, randomly chosen from a group of 36 individuals that met several eligibility requirements. First, they had contributed at least 800,000 screenshots, reflecting an average of at least 3 hours of daily smartphone use over the year. This measure was chosen to capture use data that is close to the national average [-]. The goal was to ensure that participants had sufficient smartphone use to demonstrate relationships with mental health. Second, the selected participants had completed at least 20 biweekly surveys, creating a sufficient longitudinal record for each participant that exceeded the number of variables in the pCCA []. Finally, the selected participants had reported at least 1 biweekly period where their mental health symptoms reached clinically significant thresholds [,,,].
Ethical Considerations
This study adhered to national regulations and institutional policies, including the Declaration of Helsinki (2013), and was approved by Stanford University’s institutional review board under protocol (#56430). All procedures were reviewed and classified as minimal risk.
Participants provided electronic informed consent and HIPAA authorization before enrollment. The consent process included detailed information about the study’s goals, procedures, and privacy protections, and participants confirmed their consent using an e-signature interface. To ensure privacy and confidentiality, all data were encrypted during transfer and securely stored on Stanford’s high-risk personal identifiable information and protected health information–compliant servers. Screenshots and metadata were deidentified prior to analysis, with access to raw screenshots restricted to a small subset of researchers under strict institutional protocols. Only aggregated, deidentified data were analyzed and reported to prevent the identification of individual participants. Participants were compensated US $15 for each fortnightly survey and data submission, with potential earnings of up to US $390 for completing the 1-year study.
Data were collected unobtrusively using a custom Screenomics app that captured screenshots every 5 seconds when the device was active. The app included an icon on the status bar to ensure transparency. Although nearly all screenshots were processed through automated methods, a small subset could be viewed by researchers in specific analyses. To mitigate risks, research staff were trained as mandated reporters under Stanford University and California guidelines. Any concerning content signaling potential legal, medical, or social risks was escalated to the Principal Investigators for appropriate action. Protocols were in place to refer participants to professionals or services as needed, ensuring their privacy while addressing potential risks to their well-being.
No images or supplementary materials in this manuscript include identifiable participant information. All visual data were anonymized to ensure privacy, and no consent was required for image inclusion as all images were deidentified.
All research staff completed mandatory HIPAA and human subjects training. Data security measures included strong passwords, restricted database access, and frequent reminders of privacy policies. These procedures ensured the highest level of data integrity and participant confidentiality.
Measures
Mental Health Measures
Participants completed fortnightly surveys over a year to assess symptoms of depression, and ADHD, in addition to state anxiety and positive affect. This extended time frame allowed the study of fluctuations across weeks or months [-].
Depression symptoms were measured using the 10-item Center for Epidemiologic Studies Depression scale []. Participants rated the frequency of symptoms (eg, “I felt I could not shake off the blues”) on a 4-point Likert scale. Scores range from 0 to 60, with scores above 16 indicating clinical risk []. These scores were used to calculate the number of fortnight participants who were at clinical risk for depression (
Table 1). The CES-D scale is in the public domain.
State anxiety was assessed with the 20-item State Anxiety subscale of the State-Trait Anxiety Inventory []. Items (eg, “I am tense; I am worried”) were rated on a 4-point Likert scale. Scores range from 20 to 80, with scores over 40 indicating moderate to high anxiety []. This threshold was used to calculate fortnights at clinical risk (Table 1). A license was obtained from Mind Garden, Inc.
Positive affect was measured using a single item: “Looking back over the last two weeks, I felt happy,” rated on a 100-point slider scale from “strongly disagree” to “strongly agree.”
Hyperactivity and inattention symptoms were measured fortnightly using the 18-item Adult ADHD Self-Report Scale [,]. Participants rated hyperactivity and inattention on a 5-point Likert scale. Subscale scores range from 0 to 36. A score of 9 or more on the first 6 items (4 inattention and 2 hyperactivity) indicates clinical risk for adult ADHD []. These thresholds were used to calculate risk fortnights (Table 1). A license was obtained from NYU TOV Licensing, New York University.
| Participant A (n=26) | Participant B (n=25) | Participant C (n=20) | Participant D (n=25) | Participant E (n=27) | |
| Total screens | 824,111 | 1,191,958 | 897,514 | 1,106,164 | 2,724,266 |
| Age (year) | 62 | 22 | 37 | 20 | 24 |
| Sex | Female | Female | Female | Female | Female |
| Race | White | White | Multiracial Hispanic | White | Black |
| Region | Midwest | Midwest | Northeast | Pacific | Northeast |
| Residence type | Rural | Suburban | Suburban | Suburban | Urban |
| Highest educational level | 2-year college | High school or GEDb | Some college | Some college | Some College |
| Marital status | Divorced or separated | Single or never married | Married | Single or never married | Married |
| Income level (US $) | 15,000-24,999 | 14,999 or less | 25,000-29,999 | 75,000-99,999 | 14,999 or less |
aThe table presents the demographic characteristics of participants in the study, as collected through an initial onboarding survey. The survey was administered before the commencement of the study to gather pertinent information regarding participant backgrounds. “n” is the count of fortnightly surveys completed by each participant throughout the study year, while “Total screens” is the cumulative number of screenshots recorded from each participant’s smartphone during the study year.
bGED: General Educational Development.
Media Use Metrics
Six metrics for smartphone media use were derived from the screenshots and operating system metadata recorded continuously at 5-second intervals over the study year. Metadata enabled us to map app names and categories using Google Play Market classifications [,]. All metrics were averaged daily within biweekly periods, defined as the 14 days preceding each survey, except session duration calculated as the average screen-on interval. Missing data were addressed under the “missing at random” assumption.
Screen time was measured as the daily average number of screenshots captured during screen-on periods. The data collection app captured screenshots every 5 seconds when the screen was on. For each fortnight, the total screenshots were divided by the number of active data collection days.
Social screen time was measured as the daily average of screenshots recorded while using social apps. Social apps were those classified as “social” by Google Play Market categorization [,]. Social apps included common social media apps (eg, TikTok and Instagram), certain messaging apps (eg, Telegram and Text Free), and social network or dating apps (eg, Tagged and Grindr). Similar to the screen time measure, we calculated this metric by dividing the total number of social app screenshots by the number of active data collection days.
The number of sessions of smartphone use indicated average daily sessions, defined as intervals between screen on or off events. For each fortnight, we calculated the daily average number of sessions by dividing the total number of sessions by the number of days the participant actively used their phone. A higher number of sessions (eg, ten 30-second sessions) relative to the duration of each session (eg, one 5-minute session) indicates a more fragmented digital experience.
Session duration was calculated as the time interval between the initiation of a session and its termination. For each fortnight, we summed the duration of all sessions that occurred within that period and then computed the mean. This provided an average session duration metric for the specified time frame.
The number of unique apps was calculated as the daily average of distinct apps used, calculated by dividing the count of unique apps within a period by the number of active days.
The number of app switches indicated the daily average number of times individuals switched between apps and was calculated as a measure of fragmentation. For each biweekly period, we counted how many times individuals moved from app to app (ie, from Facebook to Zoom=1 switch) and divided that number by the number of active smartphone data collection days. Please see [-,,-,-] for more discussion on theories and measures examining media use and mental health over time and the origins of our measures.
Data Analysis
Canonical correlation analysis (CCA) is a classical technique used to obtain the maximal correlation between objects that are represented by 2 sets of variables []. Adopting the Nesselroade et al idiographic filter approach [], we use pCCA to model each individual’s multivariate repeated measures data. The observed correlation between an individual’s media use and mental health is not static; rather, it depends on the chosen coordinate system, that is, how observed variables are projected onto latent constructs of media use and mental health. Unlike traditional factorial analyses that assume a uniform variable projection across individuals, the idiographic filter approach allows person-specific rotations of the coordinate system.
pCCA does this by locating the specific rotation where the linear correlation between the latent media use and mental health canonical variates is mutually maximized. The size and significance of the maximum correlation (ρ) are used to answer our first research question: Does the analysis of single individuals offer different conclusions than the small or null effects reported in the across-individual analyses dominating the literature? The configurations of the basis vectors x and y are interpreted to address our second and third questions: Can we identify combinations of media use metrics and mental health metrics that might advance theories about media effects?
Given Ax (27 occasions × 6 media use variables) and Ay (27 occasions × 5 mental health variables) for an individual, the data are projected onto canonical variates, zx = Axx and zy = Ayy, the correlation of which is given by the following equation.
where x and y are chosen to maximize ρ, achieved via singular value decomposition [,]. We use permutation tests to ensure observed correlations are not simply artifacts of random chance. Specifically, for each participant, we perform 1000 permutations by randomly shuffling the values within the columns of Ax or Ay, disrupting any temporal ordering between media use and mental health. This process generates a null distribution for ρ, representing the range of chance correlations expected in the absence of a systematic relationship. Comparing the observed ρ to this null distribution provides an exact P value based on the empirical data, thereby avoiding reliance on theoretical distributions that could inflate P values. Histograms of these permutations are presented in . All pCCA and permutation test analyses were done using the cca package [] in R (R Foundation for Statistical Computing).
Results
The following results include an analysis of a full year of media use and mental health data (6,744,013 screenshots and 123 fortnightly surveys) provided by 5 individuals. The background of each participant and descriptive statistics about media use and mental health are in Tables 1-3.
| Participants | Depression symptoms (0-60) | State anxiety (20-80) | ADHDb symptoms (0-72) | Positive affect (0-100), mean (SD) | |||||||||||
| Mean (SD) | Clinical risk periods, n | Mean (SD) | Clinical risk periods, n | Inattention (0-36), mean (SD) | Hyperactivity (0-36), mean (SD) | Clinical risk periods, n | |||||||||
| A | 9.23 (5.31) | 1 | 44.23 (4.43) | 22 | 1.88 (3.24) | 2.04 (3.32) | 0 | 53.85 (17.54) | |||||||
| B | 24.80 (5.07) | 24 | 51.84 (10.15) | 21 | 12.72 (1.99) | 8.72 (1.79) | 13 | 62.12 (19.81) | |||||||
| C | 12.40 (4.77) | 4 | 45.05 (14.25) | 13 | 10.10 (2.25) | 5.70 (0.80) | 3 | 65.35 (5.43) | |||||||
| D | 12.80 (6.10) | 7 | 45.84 (8.76) | 17 | 5.36 (2.27) | 2.76 (1.39) | 0 | 57.00 (15.21) | |||||||
| E | 14.78 (7.06) | 4 | 55.00 (8.98) | 27 | 4.52 (4.37) | 4.34 (4.36) | 0 | 16.85 (27.98) | |||||||
aThe table shows the means and SD for each mental health measure, as well as the clinical risk associated with symptomatology exceeding certain thresholds. Clinical risk periods are calculated as the number of fortnightly surveys in which participants’ responses exceeded the clinical threshold for each measure.
bADHD: attention-deficit/hyperactivity disorder.
| Participant | Screen time (hours per day), mean (SD) | Social screen time (hours per day), mean (SD) | Number of sessions (per fortnight), mean (SD) | Session duration (minutes per day), mean (SD) | Number of unique apps (per day), mean (SD) | Number of app switches (per day), mean (SD) |
| A | 2.99 (0.51) | 0.41 (0.17) | 57.84 (8.12) | 3.10 (0.49) | 4.08 (0.39) | 210.27 (32.79) |
| B | 4.15 (1.10) | 0.37 (0.12) | 32.76 (3.18) | 7.59 (1.86) | 3.64 (0.37 | 206.87 (18.96) |
| C | 4.18 (0.95) | 0.15 (0.23) | 57.81 (8.89) | 4.33 (0.85) | 7.29 (2.50) | 324.88 (52.73) |
| D | 4.18 (1.05) | 0 (0) | 90.90 (25.28) | 2.87 (0.89) | 4.54 (0.61) | 273.22 (54.38) |
| E | 9.55 (2.03) | 0 (0) | 9.45 (2.10) | 62.62 (14.75) | 1.50 (0.59) | 39.51 (14.48) |
The main objective of this study was to assess whether analysis of media use metrics derived from super-intensive observational data passively collected from the smartphones individuals use in their daily lives could unearth relationships between media use and mental health within individuals that have previously been difficult to observe using analyses aggregated across individuals. Related to that objective, we first report the within-person associations between the media use and mental health latent variables that were obtained separately for each participant using pCCA.
The main results of this study are reported in 
Figure 1, which shows correlations and plots of the fortnightly media use and mental health variables across 1 year for each participant. The observed p-technique canonical correlations are large, ranging from 0.78 to 0.92. These values provide a compelling answer to our primary research question: for 4 of the 5 participants in this study, fluctuations in canonical media use variate were significantly and tightly coupled with fluctuations in a canonical mental health variate, more so than for any previous associations observed in this literature [,].
The composition of media use and mental health variates was unique for each participant, reflecting person-specific relationships and highlighting the heterogeneity in how media use impacts mental health. While individual metrics (eg, app-switching, screen time) varied considerably over time, the canonical variates showed consistent and strong covariation for four of the 5 participants. These findings underscore the use of pCCA in identifying tightly coupled, within-person associations that are otherwise obscured in aggregated analyses, paving the way for more individualized theoretical insights and interventions.
The figures below highlight the unique person-specific relationships between media use metrics and mental health measures. These findings demonstrate the variability in how media use correlates with mental health for each participant, emphasizing the value of individualized approaches.
For participant A (
Figure 2), the media use behavior most related to greater mental health symptoms over time was primarily shorter sessions (–0.93), and secondarily a higher number of sessions (0.51) and shorter screen time (–0.48). This suggests that the fluctuations in media use most related to mental health for this person are quick-checking behaviors on the smartphone. In complement, the configuration of mental health symptoms and positive affect that was most related to fluctuations in media use was characterized by lower state anxiety (–0.85) and lower levels of ADHD symptoms for both inattention (–0.59) and hyperactivity (–0.52), and lower positive affect (–0.71). Generally, for participant A, quick checking on the smartphone was associated with lower positive affect, and lower state anxiety and ADHD symptoms.
Figure 3 shows that for participant B, the media use behaviors most related to greater mental health symptoms over time were characterized by longer sessions (0.66) and secondarily by more screen time (0.53) and little app switching (–0.55). We characterize this combination as focused smartphone use. The mental health symptoms most related to media use were characterized by depression symptoms (0.61) and lower positive affect (–0.53). It is also notable that this participant’s mental health changes, especially true for depression symptoms, had a stronger association with media use changes than for state anxiety. The media use canonical variate for participant B was driven by fluctuations in total screen time rather than social app use, specifically.
For participant C (
Figure 4), the media use canonical variate shows the strongest loadings for the number of social screen time (distinct from total screen time) at 0.79. Secondarily, there is a high loading for the use of unique apps (0.53). The mental health canonical variate shows a strong loading associated with depression symptoms (0.75). Additionally, there is a noteworthy negative loading in relation to hyperactivity symptoms (–0.60). The much larger loading for social screen time in comparison to total screen time shows the value of separating these uses when examining relationships with mental health. This distinction is often part of discussions about media effects (ie, more worry about intensive social app use and depression than about the effects of all media use). As for many media effects, however, the value of the distinction is person-specific.
Participant D (
Figure 5) did not use any social apps during the entire year (ie, there are no screens from any of the hundreds of social apps in the Google Play store). Their media use canonical variate is characterized by fewer app switches (–0.77), extended session durations (0.71), and a reduced number of sessions overall (–0.57). This pattern indicates longer, focused interactions. The mental health canonical variate was characterized primarily by reduced positive affect (–0.58), with no large loadings associated with any of the mental health symptoms. For participant D, longer and more focused smartphone sessions had a more singular definition (diminished positive affect). However, the number and duration of the sessions were still highly correlated with smartphone use features (r=0.82; P=.008).
There was no statistically significant association between the media and mental health canonical variates for Participant E (
Figure 6).
Discussion
Principal Findings
The results highlight the person-specific associations between media use and mental health. Canonical correlations revealed substantial covariation (0.82 to 0.92) over time for 4 out of 5 individuals. Simultaneously, as hypothesized, the specific combinations of media dosage metrics and dimensions of mental health that defined this covariation were different for each person ( [-,,-,-]. The approach supports the idea of a common association between media use and mental health while accommodating idiosyncrasy in how that association manifests for individuals. The couplings between media use and mental health for each person offer insights into the complexity of this relationship, directly addressing the research questions posed in the introduction.
In summary, our conclusions are the following.
- As research investments shift toward intensive, granular longitudinal data about individuals’ digital experiences, this study demonstrates the potential to construct distinct, person-specific models of media use and mental health, revealing insights that transcend traditional aggregated analyses.
- The study’s innovative approach, combining various dosage metrics of fragmented media use, advances person-specific theorizing in media psychology, offering potential explanations for the impact on mental health and emphasizing the significance of tailored approaches.
- By identifying individualized combinations of mental health symptoms, the study underscores the necessity of considering nuanced factors beyond single measures of dosage and affect or clinical symptoms associated with specific disorders.
Causality and Theory
The substantial p-technique canonical correlations observed in this study reflect within-person associations between media use and mental health but do not establish causality. Media use metrics may act as markers responding to mental health changes, antecedents influencing mental health, or reflections of confounding variables. A causal explanation likely involves complex mechanisms interacting over different time domains and individual histories, which this analysis does not aim to disentangle.
Past research has explored single mechanisms over varying time frames, such as the introduction of Facebook on campuses (months/years) [], ecological momentary assessment studies linking self-reported media use with mental health (days/weeks) [], and cohort analyses over decades []. Additionally, momentary exposures, like a single unflattering social media post, might amplify depression symptoms, particularly during periods of heightened vulnerability. Interactions between short- and long-term domains—for example, a negative social media post during an extended depressive episode—could further complicate causal pathways.
Combining mechanisms into a single causal conclusion for media use and mental health is beyond the scope of this project and likely any single study, given the heterogeneity observed across individuals. The participants in this study suggest that no universal explanation applies across diverse media behaviors, content, and time domains. Instead, multiple mechanisms, each tailored to specific contexts, are likely needed. Controlled experiments targeting distinct media experiences [,] and idiographic approaches like ours can help identify mechanisms for further exploration.
The within-person associations in this project emphasize the need for new theories addressing the accumulation of similar media experiences over time, regardless of specific content. Such mechanisms echo the concept of resonance in social reality theory [,], where outcomes arise from the alignment of digital and real-life environments. For instance, increased screen use during a depressive episode may reinforce social isolation, creating a feedback loop that intensifies mental health symptoms []. These findings highlight the importance of reexamining how repetitive media experiences influence mental health over extended periods.
Interventions
The p-technique canonical correlations observed in this project (Figure 1) suggest that results obtained for a participant over 1 year may generalize to future months. Acute changes in media use could serve as markers for triggering mental health interventions without requiring frequent or intrusive direct assessments. Interventions could integrate patient and clinician access to data through dashboards alerting users to mental health risks and offering tailored recommendations. For instance, systems might limit social app usage for some individuals or promote it for others based on identified risks. Person-specific thresholds for intervention provide a more accurate approach than generalized thresholds derived from population averages [].
The key to these interventions is precision timing and targeted actions that connect continuous monitoring of screen activity with information designed to improve mental health. Screen-based systems might rival the diagnostic accuracy of traditional methods for some disorders. Studies estimate, for example, that up to 60% of people with depression are never diagnosed [,], and even those connected to care often face delays in appointments, diagnosis, and treatment initiation []. Screen summaries might accelerate care by offering behavioral insights prior to intake or as indicators of treatment effectiveness.
Determining the clinical use of idiographic tracking will require research to establish optimal baseline collection periods, which may vary by individual. While longer data collection can enhance accuracy, even shorter periods could yield valuable insights. For example, smartphone monitoring during the weeks between initial clinician contact and the first appointment, or in critical postdischarge periods, could inform preliminary evaluations and adaptive interventions. These monitoring systems could feed data into clinician dashboards and support the implementation of just-in-time adaptive interventions []. Importantly, such tools are intended to augment rather than replace direct patient-clinician interactions, which require robust support resources.
Limitations of the Study
This longitudinal study, conducted with a small, nonclinical sample of female Android phone users, was not intended to provide population-level conclusions about the prevalence of media and mental health relationships nationally or within subgroups. Avoiding generalization at this stage helps circumvent the small sample fallacy []. Our intent was to first understand the unique media dosage and mental health indicators important for individuals without prematurely investing in a research strategy aimed at extensive between-subject variability that we believe has resulted in considerable ambiguity in this literature. This bottom-up inductive approach celebrates idiosyncrasy and prioritizes individual models as a first step toward building models that might apply to larger groups of people. We have successfully demonstrated that it is possible to find strong, individualized relationships with granular measures and longitudinal data.
Our analysis focused on a limited set of variables: 6 media use metrics and 5 mental health measures. Many additional factors could influence within-person associations between media use and mental health. Notably, data collection coincided with the height of the COVID-19 pandemic. Although none of the participants contracted the virus, the pandemic’s broader effects may have shaped their mental health and media use behaviors. Future analyses could examine how contextual factors, such as global or personal life events, moderate these relationships. For example, exploring interactions between media use, sleep patterns, and mental health outcomes [,], or identifying whether media use acts as a supportive resource [,] or an additional stressor [,] could inform context-aware models that better capture individual contingencies.
Future Work and Conclusions
While new computational methods enable the collection and analysis of large datasets, they often focus on aggregated measures, obscuring person-specific longitudinal insights and yielding ambiguous population-level conclusions. Our method emphasizes the importance of a person-specific approach applicable to new datasets. A key recommendation for extending this investigation is to maintain this individual focus and compare people based on similarities between models derived for single individuals, rather than relying solely on priori groupings based on demographics or other psychosocial characteristics. When idiographic models are available for many individuals, researchers can use clustering, latent class, and mixture models to quantify and examine between-person similarities and differences [-]. One such method suggested by Ram et al [] involves calculating Tucker congruence coefficients [] that compare parameter estimates (eg, the canonical variates) for each pair of individuals in a larger sample. Importantly, this approach entails specifying and modeling each person’s data separately, as done in this study, before identifying shared patterns. Rather than assuming a priori commonalities (eg, based on sex or personality), this bottom-up paradigm uncovers the full range of person-specific solutions, which can then inform theory, policy, and interventions [,,,].
Future work will expand on the metrics presented here, integrating richer variables through our computational assay [] and identifying screen content characteristics—such as words, objects, emotional arousal, and environmental features—using advanced artificial intelligence models. These granular data can help refine our understanding of how different media use patterns related to mental health. While traditional screen time measures, often derived from self-reports, have been criticized for their imprecision [], this study suggests that accurate, granular recordings of screen behavior can reveal strong associations for some individuals. Similar increments in the precision of assessing media use have paid off in the study of other media effects; for example, exposure to political information [], advertising [], and food content in digital media []. Our ongoing work also includes open-sourcing the most up-to-date Screenomics app and toolset [].
The idiographic filter approach, producing individual models 1 person at a time, may initially seem to risk overfocusing on small samples. However, there may not be a shorter route to useful answers about media use and mental health. This is especially true if we want to retain media use and mental health as important and psychologically complex categories of digital and non-digital life. The present results suggest that to advance research addressing media use and mental health problems generally, we may need to seek answers applicable to one person at a time; that is, we must embrace idiosyncrasy, not ignore it.
Acknowledgments
This project was partially funded by an internal Stanford University seed grant from the Stanford Medical School (Spectrum SPADA).
Data Availability
All deidentified data used and reported in this study, along with the code for analysis and visualizations, are available from the corresponding author upon reasonable request.
Authors' Contributions
TNR, NR, and BR did the inception and design of the study; MC, AYL, and BR drafted the manuscript and did data interpretation; NR, MC, and NH did data analysis and strategy; AM and JH advised on literature review, mental health assessments, and interpretations; DM and AAGT organized data acquisition and data management. All authors have edited the manuscript in multiple versions and accept responsibility for the entire content of this manuscript.
Conflicts of Interest
DM is a founder of Screenlake Inc, a startup company focusing on smartphone data collection and analysis for the purpose of market intelligence. His involvement in the present study was the design and management of data collection during his doctoral studies, which took place prior to the founding of the company. JH consults for YouTube and Google on issues related to social media and receives compensation for these services. However, his contribution to this paper is solely his own work. The other authors declare no competing interests.
Multimedia Appendix 2
Permutation distribution histograms. This figure displays histograms of the permutation distributions of 1,000 permuted canonical correlations for each participant (A to E). To assess the statistical significance of the observed canonical correlations, we conducted permutation tests using Wilks' lambda by randomly permuting each individual’s data, creating a null distribution. This null distribution represents correlations expected by chance—meaning the values that would arise if there were no systematic relationship between media use and mental health variables and any observed associations were purely random. The histograms display these chance-based distributions under the null hypothesis, with observed canonical correlations indicated by red dashed lines to evaluate their statistical significance. The test statistic shown in each histogram is Wilks' lambda. For Participant E, the result is nonsignificant (P> 0.05), suggesting that the observed correlation cannot be confirmed as statistically significant. For the other participants, the observed correlations are confirmed by the permutation test.
PNG File , 5001 KBReferences
- Sohn SY, Rees P, Wildridge B, Kalk NJ, Carter B. Prevalence of problematic smartphone usage and associated mental health outcomes amongst children and young people: a systematic review, meta-analysis and GRADE of the evidence. BMC Psychiatry. 2019;19(1):356. [FREE Full text] [CrossRef] [Medline]
- Kushlev K, Leitao MR. The effects of smartphones on well-being: theoretical integration and research agenda. Curr Opin Psychol. 2020;36:77-82. [CrossRef] [Medline]
- Hancock J, Liu SX, Luo M, Mieczkowski H. Psychological well-being and social media use: a meta-analysis of associations between social media use and depression, anxiety, loneliness, eudaimonic, hedonic and social well-being. SSRN J. 2022:84. [CrossRef]
- Marciano L, Ostroumova M, Schulz PJ, Camerini AL. Digital media use and adolescents' mental health during the Covid-19 pandemic: a systematic review and meta-analysis. Front Public Health. 2021;9:793868. [FREE Full text] [CrossRef] [Medline]
- Nawi A, Zameran FA. Exploring the relationship between smartphone use and adolescent well-being: a systematic literature review. In: Rajakumar G, Du KL, Vuppalapati C, Beligiannis GN, editors. Intelligent Communication Technologies and Virtual Mobile Networks. Singapore. Springer; 2023.
- Harari GM, Müller SR, Aung MS, Rentfrow PJ. Smartphone sensing methods for studying behavior in everyday life. Curr Opin Behav Sci. 2017;18:83-90. [CrossRef]
- Brinberg M, Bodie GD, Solomon DH, Jones SM, Ram N. Examining individual differences in how interaction behaviors change over time: a dyadic multinomial logistic growth modeling approach. Psychol Methods. 2023. [CrossRef] [Medline]
- Mao C, Bayer JB, Ross MQ, Rhee L, Le HTK, Mount J, et al. Perceived vs. observed mHealth behavior: a naturalistic investigation of tracking apps and daily movement. Mobile Media Commun. 2023;11(3):526-548. [CrossRef]
- Barnett ML, Gonzalez A, Miranda J, Chavira DA, Lau AS. Mobilizing community health workers to address mental health disparities for underserved populations: a systematic review. Adm Policy Ment Health. 2018;45(2):195-211. [FREE Full text] [CrossRef] [Medline]
- Chancellor S, De Choudhury M. Methods in predictive techniques for mental health status on social media: a critical review. NPJ Digit Med. 2020;3:43. [FREE Full text] [CrossRef] [Medline]
- Torous J, Choudhury T, Barnett I, Keshavan M, Kane J. Smartphone relapse prediction in serious mental illness: a pathway towards personalized preventive care. World Psychiatry. 2020;19(3):308-309. [FREE Full text] [CrossRef] [Medline]
- Orben A, Przybylski AK, Blakemore SJ, Kievit RA. Windows of developmental sensitivity to social media. Nat Commun. 2022;13(1):1649. [FREE Full text] [CrossRef] [Medline]
- Orben A, Blakemore SJ. How social media affects teen mental health: a missing link. Nature. 2023;614(7948):410-412. [CrossRef] [Medline]
- Beyens I, Pouwels JL, van Driel II, Keijsers L, Valkenburg PM. Social media use and adolescents’ well-being: developing a typology of person-specific effect patterns. Commun Res. 2021;51(6):691-716. [CrossRef]
- Valkenburg PM, Meier A, Beyens I. Social media use and its impact on adolescent mental health: an umbrella review of the evidence. Curr Opin Psychol. 2022;44:58-68. [FREE Full text] [CrossRef] [Medline]
- Bryant J, Oliver MB. Media Effects: Advances in Theory and Research. New York, NY. Routledge; 2009.
- McLeod JM, Reeves B. On the nature of mass media effects. In: Withey SB, Abeles RP, editors. Television and Social Behavior: Beyond Violence and Children/A Report of the Committee on Television and Social Behavior, Social Science Research Council. Oxfordshire, United Kingdom. Routledge; 2013.
- Nesselroade JR, Gerstorf D, Hardy SA, Ram N. Focus article: idiographic filters for psychological constructs. Meas Interdiscip Res Perspect. 2007;5(4):217-235. [CrossRef]
- Thompson B. Canonical Correlation Analysis: Uses and Interpretation. Washington, DC. Sage Publications; 1984.
- Ram N, Haber N, Robinson TN, Reeves B. Binding the person-specific approach to modern AI in the human screenome project: moving past generalizability to transferability. Multivariate Behav Res. 2024;59(6):1211-1219. [CrossRef] [Medline]
- Reeves B, Robinson T, Ram N. Time for the Human Screenome Project. Nature. 2020;577(7790):314-317. [CrossRef] [Medline]
- Kessler RC, Adler L, Ames M, Demler O, Faraone S, Hiripi E, et al. The World Health Organization Adult ADHD Self-Report Scale (ASRS): a short screening scale for use in the general population. Psychol Med. 2005;35(2):245-256. [CrossRef] [Medline]
- Chen R, Mias GI, Li-Pook-Than J, Jiang L, Lam HYK, Chen R, et al. Personal omics profiling reveals dynamic molecular and medical phenotypes. Cell. 2012;148(6):1293-1307. [FREE Full text] [CrossRef] [Medline]
- Rozgonjuk D, Levine JC, Hall BJ, Elhai JD. The association between problematic smartphone use, depression and anxiety symptom severity, and objectively measured smartphone use over one week. Comput Hum Behav. 2018;87:10-17. [CrossRef]
- Alloy LB, Urošević S, Abramson LY, Jager-Hyman S, Nusslock R, Whitehouse WG, et al. Progression along the bipolar spectrum: a longitudinal study of predictors of conversion from bipolar spectrum conditions to bipolar I and II disorders. J Abnorm Psychol. 2012;121(1):16-27. [FREE Full text] [CrossRef] [Medline]
- McInnis MG, Andreassen OA, Andreazza AC, Alon U, Berk M, Brister T, et al. Strategies and foundations for scientific discovery in longitudinal studies of bipolar disorder. Bipolar Disord. 2022;24(5):499-508. [FREE Full text] [CrossRef] [Medline]
- Keller MB, Lavori PW, Mueller TI, Endicott J, Coryell W, Hirschfeld RM, et al. Time to recovery, chronicity, and levels of psychopathology in major depression. a 5-year prospective follow-up of 431 subjects. Arch Gen Psychiatry. 1992;49(10):809-816. [CrossRef] [Medline]
- Crocetti E, Klimstra T, Keijsers L, Hale WW, Meeus W. Anxiety trajectories and identity development in adolescence: a five-wave longitudinal study. J Youth Adolesc. 2009;38(6):839-849. [CrossRef] [Medline]
- Buecker S, Luhmann M, Haehner P, Bühler JL, Dapp LC, Luciano EC, et al. The development of subjective well-being across the life span: a meta-analytic review of longitudinal studies. Psychol Bull. 2023;149(7-8):418-446. [CrossRef]
- Ram N, Yang X, Cho MJ, Brinberg M, Muirhead F, Reeves B, et al. Screenomics: a new approach for observing and studying individuals' digital lives. J Adolesc Res. 2020;35(1):16-50. [FREE Full text] [CrossRef] [Medline]
- Humphreys L. The Qualified Self: Social Media and the Accounting of Everyday Life. Cambridge, MA. MIT Press; 2018.
- Conway CC, Forbes MK, Forbush KT, Fried EI, Hallquist MN, Kotov R, et al. A hierarchical taxonomy of psychopathology can transform mental health research. Perspect Psychol Sci. 2019;14(3):419-436. [FREE Full text] [CrossRef] [Medline]
- Nock MK, Kessler RC, Franklin JC. Risk factors for suicide ideation differ from those for the transition to suicide attempt: the importance of creativity, rigor, and urgency in suicide research. Clin Psychol Sci Pract. 2016;23(1):31-34. [CrossRef]
- Nock MK, Banaji MR. Prediction of suicide ideation and attempts among adolescents using a brief performance-based test. J Consult Clin Psychol. 2007;75(5):707-715. [FREE Full text] [CrossRef] [Medline]
- Robinaugh DJ, Brown ML, Losiewicz OM, Jones PJ, Marques L, Baker AW. Towards a precision psychiatry approach to anxiety disorders with ecological momentary assessment: the example of panic disorder. Gen Psychiatr. 2020;33(1):e100161. [FREE Full text] [CrossRef] [Medline]
- Chahal R, Ho TC, Miller JG, Borchers LR, Gotlib IH. Sex-specific vulnerability to depressive symptoms across adolescence and during the COVID-19 pandemic: the role of the cingulum bundle. JCPP Adv. 2022;2(1):e12061. [FREE Full text] [CrossRef] [Medline]
- Hardoon DR, Szedmak S, Shawe-Taylor J. Canonical correlation analysis: an overview with application to learning methods. Neural Comput. 2004;16(12):2639-2664. [CrossRef] [Medline]
- Orben A. Teenagers, screens and social media: a narrative review of reviews and key studies. Soc Psychiatry Psychiatr Epidemiol. 2020;55(4):407-414. [CrossRef] [Medline]
- Schimmack U, Krause P, Wagner GG, Schupp J. Stability and change of well being: an experimentally enhanced latent state-trait-error analysis. Soc Indic Res. 2009;95(1):19-31. [CrossRef]
- Liu X, Ping S, Gao W. Changes in undergraduate students' psychological well-being as they experience university life. Int J Environ Res Public Health. 2019;16(16):2864. [FREE Full text] [CrossRef] [Medline]
- Odgers CL, Jensen MR. Annual research review: adolescent mental health in the digital age: facts, fears, and future directions. J Child Psychol Psychiatry. 2020;61(3):336-348. [FREE Full text] [CrossRef] [Medline]
- Yoon S, Kleinman M, Mertz J, Brannick M. Is social network site usage related to depression? A meta-analysis of Facebook-depression relations. J Affect Disord. 2019;248:65-72. [CrossRef] [Medline]
- Houben M, Van Den Noortgate W, Kuppens P. The relation between short-term emotion dynamics and psychological well-being: a meta-analysis. Psychol Bull. 2015;141(4):901-930. [CrossRef] [Medline]
- Meier A, Reinecke L. Computer-mediated communication, social media, and mental health: a conceptual and empirical meta-review. Commun Res. 2020;48(8):1182-1209. [CrossRef]
- Vanden Abeele MMP, Halfmann A, Lee EWJ. Drug, demon, or donut? Theorizing the relationship between social media use, digital well-being and digital disconnection. Curr Opin Psychol. 2022;45:101295. [CrossRef] [Medline]
- Goodman LS, Gilman A. Brunton LL, Knollman BC, editors. Goodman and Gilman's The Pharmacological Basis of Therapeutics. New York, NY. McGraw Hill; 2022.
- Schleidgen S, Friedrich O, Gerlek S, Assadi G, Seifert J. The concept of “interaction” in debates on human-machine interaction. Humanit Soc Sci Commun. 2023;10(1). [CrossRef]
- Orben A, Meier A, Dalgleish T, Blakemore SJ. Mechanisms linking social media use to adolescent mental health vulnerability. Nat Rev Psychol. 2024;3(6):407-423. [CrossRef]
- Lee AY, Katz R, Hancock J. The role of subjective construals on reporting and reasoning about social media use. Soc Media Soc. 2021;7(3):205630512110353. [CrossRef]
- Yeykelis L, Cummings JJ, Reeves B. The fragmentation of work, entertainment, e-mail, and news on a personal computer: motivational predictors of switching between media content. Media Psychol. 2017;21(3):377-402. [CrossRef]
- Reeves B, Ram N, Robinson TN, Cummings JJ, Giles CL, Pan J, et al. Screenomics: A framework to capture and analyze personal life experiences and the ways that technology shapes them. Hum Comput Interact. 2021;36(2):150-201. [FREE Full text] [CrossRef] [Medline]
- Radloff L. The CES-D Scale: A self-report depression scale for research in the general population. Appl Psychol Meas. 1977;1(3):385-401. [FREE Full text] [CrossRef]
- Marteau TM, Bekker H. The development of a six-item short-form of the state scale of the Spielberger State-Trait Anxiety Inventory (STAI). Br J Clin Psychol. 1992;31(3):301-306. [CrossRef] [Medline]
- Wilkinson D. Screen time trends in the age of COVID-19. SimpleTexting. 2021. URL: https://simpletexting.com/blog/screen-time-survey [accessed 2023-10-22]
- Flynn J. 20 Vital Smartphone Usage Statistics (2023): facts, data, and trends on mobile use in the U.S. Zippia.com. Zippia. 2023. URL: https://www.zippia.com/advice/smartphone-usage-statistics/ [accessed 2024-02-03]
- Kerai A. Cell phone usage statistics: mornings are for notifications. Reviews.org. URL: https://www.reviews.org/mobile/cell-phone-addiction/ [accessed 2024-02-03]
- Pezeshki A, Scharf LL, Azimi-Sadjadi MR, Lundberg M. Empirical canonical correlation analysis in subspaces. 2004. Presented at: Conference Record of the Thirty-Eighth Asilomar Conference on Signals, Systems and Computers,; November 7-10, 2004:994-997; Pacific Grove, CA. [CrossRef]
- Julian LJ. Measures of anxiety: State-Trait Anxiety Inventory (STAI), Beck Anxiety Inventory (BAI), and Hospital Anxiety and Depression Scale-Anxiety (HADS-A). Arthritis Care Res. 2011;63:S467-S472. [FREE Full text] [CrossRef] [Medline]
- Vargas-Terrones M, Nagpal TS, Perales M, Prapavessis H, Mottola MF, Barakat R. Physical activity and prenatal depression: going beyond statistical significance by assessing the impact of reliable and clinical significant change. Psychol Med. 2021;51(4):688-693. [CrossRef] [Medline]
- Spielberger CD, Gorsuch RL, Lushene RE. State-Trait Anxiety Inventory (STAI): Test Manual for Form X. Washington, DC. Consulting Psychologists Press; 1968.
- Adler L, Cohen J. Diagnosis and evaluation of adults with attention-deficit/hyperactivity disorder. Psychiatr Clin North Am. 2004;27(2):187-201. [CrossRef] [Medline]
- Kessler RC, Berglund P, Demler O, Jin R, Merikangas KR, Walters EE. Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the national comorbidity survey replication. Arch Gen Psychiatry. 2005;62(6):593-602. [CrossRef] [Medline]
- JoMingyu. Google Play scraper. Github. URL: https://github.com/JoMingyu/google-play-scraper [accessed 2024-02-05]
- Rhee L, Bayer JB, Lee DS, Kuru O. Social by definition: how users define social platforms and why it matters. Telematics Inform. 2021;59:101538. [CrossRef]
- Hotelling H. Relations Between Two Sets of Variates. In: Kotz S, Johnson NL, editors. Breakthroughs in Statistics: Methodology and Distribution. New York, NY. Springer; 1992.
- Weenink D. Canonical correlation analysis. 2003. Presented at: Proceedings of the Institute of Phonetic Sciences of the University of Amsterdam ;25?99; 2003:81-99; University of Amsterdam. URL: https://uvafon.hum.uva.nl/archive/2003/2003-ifaproc25-Weenink.pdf
- González I, Déjean S. Package CCA: Canonical correlation analysis. R Foundation for Statistical Computing. 2021. URL: https://cran.r-project.org/package=CCA [accessed 2024-04-04]
- Braghieri L, Levy R, Makarin A. Social media and mental health. Am Econ Rev. 2022;112(11):3660-3693. [CrossRef]
- Müller SR, Peters H, Matz SC, Wang W, Harari GM. Investigating the relationships between mobility behaviours and indicators of subjective well–being using smartphone–based experience sampling and GPS tracking. Eur J Pers. 2020;34(5):714-732. [CrossRef]
- Johnson AL. Changes in mental health and treatment, 1997-2017. J Health Soc Behav. 2021;62(1):53-68. [CrossRef] [Medline]
- Bailey ER, Matz SC, Youyou W, Iyengar SS. Authentic self-expression on social media is associated with greater subjective well-being. Nat Commun. 2020;11(1):4889. [FREE Full text] [CrossRef] [Medline]
- Brailovskaia J, Delveaux J, John J, Wicker V, Noveski A, Kim S, et al. Finding the "sweet spot" of smartphone use: reduction or abstinence to increase well-being and healthy lifestyle?! An experimental intervention study. J Exp Psychol Appl. 2023;29(1):149-161. [CrossRef] [Medline]
- Gerbner G, Gross L, Signorielli N, Morgan M. Aging with television: images on television drama and conceptions of social reality. J Commun. 1980;30(1):37-47. [CrossRef] [Medline]
- Clark JL, Green MC. Self-fulfilling prophecies: perceived reality of online interaction drives expected outcomes of online communication. Personal Individual Differ. 2018;133:73-76. [CrossRef]
- Boers E, Afzali MH, Newton N, Conrod P. Association of screen time and depression in adolescence. JAMA Pediatr. 2019;173(9):853-859. [FREE Full text] [CrossRef] [Medline]
- Molenaar PCM, Campbell CG. The new person-specific paradigm in psychology. Curr Dir Psychol Sci. 2009;18(2):112-117. [CrossRef]
- Williams SZ, Chung GS, Muennig PA. Undiagnosed depression: a community diagnosis. SSM Popul Health. 2017;3:633-638. [FREE Full text] [CrossRef] [Medline]
- Handy A, Mangal R, Stead TS, Coffee RL, Ganti L. Prevalence and impact of diagnosed and undiagnosed depression in the United States. Cureus. 2022;14(8):e28011. [FREE Full text] [CrossRef] [Medline]
- Boerema AM, Ten Have M, Kleiboer A, de Graaf R, Nuyen J, Cuijpers P, et al. Demographic and need factors of early, delayed and no mental health care use in major depression: a prospective study. BMC Psychiatry. 2017;17(1):367. [FREE Full text] [CrossRef] [Medline]
- Nahum-Shani I, Smith SN, Spring BJ, Collins LM, Witkiewitz K, Tewari A, et al. Just-in-time adaptive interventions (JITAIs) in mobile health: key components and design principles for ongoing health behavior support. Ann Behav Med. 2018;52(6):446-462. [FREE Full text] [CrossRef] [Medline]
- Poulton EC. Small sample fallacy. In: Behavioral Decision Theory: A New Approach. Cambridge, MA. Cambridge University Press; 1994.
- Li X, Buxton OM, Lee S, Chang AM, Berger LM, Hale L. Sleep mediates the association between adolescent screen time and depressive symptoms. Sleep Med. 2019;57:51-60. [FREE Full text] [CrossRef] [Medline]
- Guerrero MD, Barnes JD, Chaput JP, Tremblay MS. Screen time and problem behaviors in children: exploring the mediating role of sleep duration. Int J Behav Nutr Phys Act. 2019;16(1):105. [FREE Full text] [CrossRef] [Medline]
- Zhang K, Kim K, Silverstein NM, Song Q, Burr JA. Social media communication and loneliness among older adults: the mediating roles of social support and social contact. Gerontologist. 2021;61(6):888-896. [CrossRef] [Medline]
- Lisitsa E, Benjamin KS, Chun SK, Skalisky J, Hammond LE, Mezulis AH. Loneliness among young adults during COVID-19 pandemic: the mediational rolea of social media use and social support seeking. J Soc Clin Psychol. 2020;39(8):708-726. [CrossRef]
- Zhao N, Zhou G. Social media use and mental health during the COVID-19 pandemic: moderator role of disaster stressor and mediator role of negative affect. Appl Psychol Health Well Being. 2020;12(4):1019-1038. [FREE Full text] [CrossRef] [Medline]
- Seo J, Lee CS, Lee YJ, Bhang SY, Lee D. The type of daily life stressors associated with social media use in adolescents with problematic internet/smartphone use. Psychiatry Investig. 2021;18(3):241-248. [FREE Full text] [CrossRef] [Medline]
- Ram N, Brose A, Molenaar PCM. Dynamic factor analysis: modeling person-specific process. In: The Oxford Handbook of Quantitative Methods. New York, NY. Oxford University Press; 2013:441.
- van der Nest G, Lima Passos V, Candel MJJM, van Breukelen GJP. An overview of mixture modelling for latent evolutions in longitudinal data: modelling approaches, fit statistics and software. Adv Life Course Res. 2020;43:100323. [CrossRef] [Medline]
- Gates KM, Lane ST, Varangis E, Giovanello K, Guskiewicz K. Unsupervised classification during time-series model building. Multivariate Behav Res. 2017;52(2):129-148. [FREE Full text] [CrossRef] [Medline]
- Lorenzo-Seva U, ten Berge JMF. Tucker's congruence coefficient as a meaningful index of factor similarity. Methodology. 2006;2(2):57-64. [CrossRef]
- Brose A, Ram N. Within-person factor analysis: Modeling how the individual fluctuates changes across time. In: Mehl MR, editor. Handbook of Research Methods for Studying Daily Life. New York, NY. Guilford Press; 2012:459.
- Yang X, Ram N, Lougheed JP, Molenaar PCM, Hollenstein T. Adolescents' emotion system dynamics: network-based analysis of physiological and emotional experience. Dev Psychol. 2019;55(9):1982-1993. [FREE Full text] [CrossRef] [Medline]
- Lee J, Hamilton JT, Ram N, Roehrick K, Reeves B. The psychology of poverty and life online: natural experiments on the effects of smartphone payday loan ads on psychological stress. Inf Commun Soc. 2022;26(14):2775-2796. [CrossRef]
- Muise D, Hosseinmardi H, Howland B, Mobius M, Rothschild D, Watts DJ. Quantifying partisan news diets in web and TV audiences. Sci Adv. 2022;8(28):eabn0083. [FREE Full text] [CrossRef] [Medline]
- Yee AZH, Yu R, Lim SS, Lim KH, Dinh TTA, Loh L, et al. ScreenLife Capture: an open-source and user-friendly framework for collecting screenomes from android smartphones. Behav Res Methods. 2023;55(8):4068-4085. [FREE Full text] [CrossRef] [Medline]
Abbreviations
| ADHD: attention-deficit/hyperactivity disorder |
| CCA: canonical correlation analysis |
| HIPAA: Health Insurance Portability and Accountability Act |
| pCCA: p-technique canonical correlation analysis |
Edited by A Mavragani; submitted 24.04.24; peer-reviewed by A AL-Asadi, T Seibers; comments to author 20.08.24; revised version received 15.10.24; accepted 17.12.24; published 26.02.25.
Copyright©Merve Cerit, Angela Y Lee, Jeffrey Hancock, Adam Miner, Mu-Jung Cho, Daniel Muise, Anna-Angelina Garròn Torres, Nick Haber, Nilam Ram, Thomas N Robinson, Byron Reeves. Originally published in JMIR Formative Research (https://formative.jmir.org), 26.02.2025.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on https://formative.jmir.org, as well as this copyright and license information must be included.