Parent–Child Adaptive Responses for Digital Resilience

Abstract

This research investigates U.S. parents’ responses to the rapidly changing, novel environment of the internet, applying evolutionary theory and interdisciplinary methodologies. Novel environments pose potential challenges to existing adaptive strategies, so this research investigates important questions about how parents and children perceive the risks of children’s entry into the virtual world and how they mitigate potential risks. The research focuses on parents of children in middle childhood (children ages 6–12), a significant period in human life history when children start building relationships outside the family. We utilize in-depth interviews (n = 26), cultural domain analysis (n = 32), surveys (n = 199), and participatory co-design (n = 34) to synergize theoretical concepts in evolutionary anthropology with the applied research focus of human–computer interaction. Cultural domain maps and interview results identify and classify perceptions of costs, benefits, and risks, including intrinsic and extrinsic sources of risk and risk tangibility. Survey results further identify platforms and risks of highest priority and confirm parental interest in new kinds of tools for managing the digital experiences of their children. Life history theory informs our approach to the development of parental control software that favors skill building and encourages parent–child discussions supporting child executive function and resilience to risks.

1. Introduction

In the United States, 80 percent of children 5–11 interact with digital devices other than television (Auxier et al. 2020). Children spend a significant portion of their day—more than two hours on average—on digital devices (Qi et al. 2023), while this and other statistics vary by age and household socioeconomics (Auxier et al. 2020). A significant increase in children’s use of the internet occurred during the COVID-19 pandemic (Hmidan et al. 2023). Social media use is close to ubiquitous among U.S. teens and is increasing among 8–12-year-old children—from 31 to 38 percent from 2019 to 2021 in a nationally representative, probability-based online survey (Rideout et al. 2022). Social media use is particularly concerning, as a large cohort study found dose-dependent relationships between the time spent using social media and the odds of teens having mental health problems and comorbidities (Riehm et al. 2019).

The recognition of the mental health effects of social media for young people and parents goes to the highest levels of the U.S. federal government. In 2023, the U.S. Surgeon General issued a new advisory on the impact of social media use on youth and adolescent mental health (Office of the Surgeon General 2023). This advisory recommended immediate action and describes steps for children, parents, technology companies, policymakers, and researchers. For researchers, a key recommendation is to evaluate best practices for social media use. A second U.S. Surgeon General’s advisory on parental stress recognized the impact of online media on parents and outlined steps to reduce stress and promote mental well-being (Office of the Assistant Secretary for Health 2024). Also in 2024, the U.S. Senate overwhelmingly passed two pieces of legislation, the Children and Teens’ Online Privacy Protection Act (COPPA 2.0) and the Kids Online Safety Act (KOSA) (U.S. Senate Committee on Science Technology and Commerce 2024), but the passage of these bills stalled in the House of Representatives prior to the 2024 election. At the state level, more than 50 bills were passed in 2024 (National Conference of State Legislatures 2024). Much of this legislation sets up requirements for social media providers, such as requiring age verification or parental consent when opening a social media account in Tennessee and Utah. The regulation of cellphone use in school was brought up in a number of states in 2024 but failed or is pending. Colorado directed its Department of Education to develop program materials and curricula to educate children about the mental and physical health impacts of social media use by youth. California passed legislation that requires providers to develop a mechanism for parents to prevent notifications from addictive social media and set up other boundaries for their children.

Alongside state legislation and federal advisories, the U.S. Surgeon General prioritizes the role of parents and children in making children’s technology use safer (Office of the Surgeon General 2023). The role of moderator requires parents to learn how their children use technology, how to keep them safe in mostly inscrutable online environments while doing so, and how concerning threats shift as children age (Family Online Safety Institute 2020). Parental strategies for moderation are quite variable and usually entail some means of time and content limitations and checking children’s activity (Auxier et al. 2020). Parent–child conversations about digital life are seen as part of best practices, and industry groups point to the lack of existing infrastructure to support parent–child conversations as an economic opportunity (Family Online Safety Institute 2020).

We utilize a mixed-methods approach in this research, combining anthropological interviewing methods, a survey, and participatory co-design. We describe how parents think about online risks and benefits presented by different types of online platforms and what parents are currently doing to create safe online experiences for their children. We interpret our descriptive findings using human life history theory, a framework developed in evolutionary anthropology, and outline how this lens can aid the development of new approaches to making online environments safer. The insights we gain inform the development of a new approach to parental software that reinforces parent–child communication and reflection rather than control. The resulting app is aimed at helping parents and children adaptively respond to perceived online risks.

1.1. Life History Theory

Online environments and digital technology are complex, artificial interaction spaces that present risks that most users only partially understand. This research aims to develop insights into the landscape of online risks children encounter and the parental responses to these risks. Our primary conceptual framework is drawn from life history theory. Life history theory seeks to explain the trade-offs of major life course processes (e.g., health vs. reproduction) and the environmental influences that affect the timing of life course development (Hill and Kaplan 1999; Sear 2020). Empirical studies taking a life history approach focus on how people adapt to early life stressors, providing greater insight into mechanisms and adaptive plasticity (Nettle 2010, 2014; Nettle and Bateson 2015). As children are exposed to stress, the hypothalamic–pituitary–adrenal (HPA) axis is affected and can become calibrated to persistent stress, leading to greater risk-seeking behavior and other undesirable mental health outcomes across the life course (Flinn et al. 2011; Kim et al. 2020). As with the findings correlating social media time and adolescent mental health issues, the CDC–Kaiser study showed a dose-dependent relationship between adverse childhood experiences (ACEs), unhealthy behaviors, and overall lifetime health (Brown et al. 2009). On the other hand, supportive social relationships can buffer adverse experiences (George 2007) and have been associated with mental health issues going into remission (Hagen 2011). There are life course effects for the magnitude of a child’s social support in reducing the odds of later life course mental and physical health challenges (Snopkowski and Ziker 2020; Ziker and Snopkowski 2020). Such causal models can help to focus the development of new digital applications aimed at improving the resilience of youth in an environment shaped by the ubiquity of digital technology and the stresses associated with it.

This research focuses on parents attending to children in middle childhood (ages 6–12). Middle childhood is the human life–history stage when children typically move into expanding roles and environments, spending more time away from parental supervision in school and other activities and establishing relationships beyond the family (Del Giudice 2018). With the rapid expansion of U.S. youth in middle childhood engaging in the digital environment, the potential for exposure to stressors—such as witnessing or experiencing violence or abuse—once limited to household and community is extended. As middle childhood is a particularly formative period of development and children in this demographic are expanding their use of digital technology, we chose to focus this research on families with children in middle childhood rather than a broader age range.

1.2. Mixed Methods Approach

This research takes a mixed-methods approach, incorporating multiple qualitative methods from cultural anthropology and human–computer interaction (HCI). Methods used include semi-structured in-depth interviews (IDIs), cultural domain elicitation, online surveys, and participatory co-design. Both HCI and anthropological research often study small non-probability samples in order to explore questions of interest in depth (Bernard 2011; Lazar et al. 2017). Participants recruited using non-probability sampling cannot be extrapolated to a larger population. That loss is balanced by depth of insight into particular populations. Rather than imposing researcher-defined questions and categories, these methods endeavor to discover categories and questions of interest to participants, which can later be parleyed into empirical inquiries. In addition to the anthropological interviews with small non-probability samples, we also distributed an online survey to a larger sample of 199 U.S. parents using stratified sampling on gender. This resulted in a 50:50 gender-balanced sample, while other characteristics were not controlled (see Appendix A).

After analyzing and interpreting our findings through the lens of life history theory, the resulting insights informed the application of participatory co-design techniques with a small non-probability sample of parents in an HCI lab setting. Parents and their children were able to inform the prototype software design.

This research has three overarching questions:

RQ1

How can we better understand the cultural domains of risks and benefits in the digital sphere for youth in middle childhood?

RQ2

What are the strategies that parents utilize in addressing their children’s use of digital technology, and how are these influenced by parents’ social characteristics?

RQ3

How can we move beyond existing parental controls driven by apprehension about the risks to new applications that promote trust, communication, and informed responses to risks?

Two graduate students, one in anthropology and the second in human–computer interaction, were involved in all four studies, including the semi-structured interviews and the survey design and analysis. Each student also conducted their individual thesis research to address more focused questions reported here as Study 2 and Study 4. The anthropology student applied cultural domain analysis to parent–child pairs to describe the cognitive organization of the risks and benefits of digital ecology and the extent to which this cognitive organization differs between parents and their children. The HCI student used participatory co-design with parents and children to develop an application that encourages goal-setting and time management skills for children and their parents in using the internet. In addition, several undergraduate researchers participated in the collection, analysis of the data, and write-up.

2. Materials and Methods

The research reported in this paper addresses the research questions through four interrelated mini-studies. The studies were completed in a rough sequence, with the first study informing those that followed, each of which added relevant findings. The first study centered on in-depth semi-structured interviews; the second, cultural domain analysis; the third, surveys; and the fourth, participatory co-design. Below, the methods, recruitment and sampling strategies, and analysis plan of each study are outlined.

2.1. Study 1: Semi-Structured In-Depth Interviews

The first study centered on semi-structured, in-depth interviews with one or both parents of children in middle childhood (ages 6 to 12). The interviews were conducted by graduate research assistants with backgrounds in anthropology and human–computer interaction.

2.1.1. IDIs–Recruitment and Sampling

Recruitment efforts were concentrated in a single U.S. state, and all but one participant resided in that state. Recruitment methods were plural, including social media, researcher outreach, and posters distributed to urban and rural libraries. The resulting participants consisted of a small, non-probability sample of 26 parents. Five lived in rural communities; the remainder lived in urban or suburban areas. Households were overwhelmingly dual-parent, but most participants were mothers. The recruiting effort purposefully sought to include a diverse range of family types; however, 19 of the participants came from two-parent, heterosexual households, and 22 were female. The over-representation of women likely reflects cultural norms and caregiver involvement, and other biases reflect the demography of the state where the study was conducted. That said, as a non-probability sample, the results cannot be extrapolated to either the state studied or the U.S. as a whole with any determinable confidence. Also important to note is that while geographically proximate participants were given the option of meeting either in person or remotely via Zoom, remote interviews were conducted with all but one participant. Lastly, while the recruitment sought parents of children aged 6 to 12, the researchers made note of the influences of younger and older siblings if they were present.

2.1.2. IDIs–Protocol

The interviews were semi-structured: the research team created an interview guide aimed at capturing participants’ day-to-day parenting experiences with technology, such as device types and uses, family rules, and other parental guidance associated with specific or generalized concerns and fears (Boyer et al. 2023) (See Appendix B). The question guide was applied flexibly by the interviewers in order to capture emergent themes. Interviews proceeded in a dialogic manner bordering on collaboration, enabling participants to shape the dialogue by being explicit about the purpose and objectives of the study and inviting personal insights into thematic groundwork. This cooperative approach strengthened our conceptual foundations by anchoring future interpretations in the contextual realities of the participants.

2.1.3. IDIs–Analysis

After digitally transcribing the interviews, we conducted a manual sorting exercise inspired by grounded theory (Bryant and Charmaz 2007). We produced an overarching conceptual map by sorting a set of over 100 artifacts (in the form of quotes and singular micro-themes) into macro-themes and identifying the relationships between them. In analyzing the statements of parents, we focused on why, what, and how they restrict (and enable) the online behavior of their children. The aim was to derive factors influencing parenting strategies. The conceptual map also provided a basis for digitally coding transcripts using NVivo.1 Coded transcripts were used to understand the breadth of the conceptual categories generated in the manual sorting.

2.2. Study 2: Cultural Domain Analysis (Free-Listing and Pile-Sorting Tasks)

The second study applied methods developed for cultural domain analysis in anthropology, specifically methods known as free listing and pile sorting. For the free-listing task, participants generated lists of items they felt were connected to a category. For the pile-sorting task, participants sorted these concepts into groups and ranked them by perceived importance within each group. These two studies were completed separately, and the participants differed between them. Taken together, free listing and pile sorting can help reveal how people think about a particular “domain” of cultural knowledge. The goal of this study was to map how parents and children conceptualize the risks and benefits of digital technology (Boyer 2024). Instructions to participants for both tasks specified that we wanted participants to list and sort risks and benefits specific to children, not to users of the internet more generally.

2.2.1. Free Listing—Sampling and Recruitment

Participants were recruited in a single U.S. state in a region proximal to the largest city. Flyers were developed and posted at libraries and other public places. We were able to recruit a small non-probability sample of five parent–child pairs to participate in the free-listing task. The five children who participated were aged 11–12. Each child was accompanied by one parent, who was also a participant in the study. The children and parents were asked to name the benefits and risks associated with the child’s use of the most commonly mentioned platforms or functionalities identified through the interviews in Study 1. Participants were recruited through social media posts about the research opportunity as well as flyers posted in libraries and grocery stores. A website address and QR code that linked potential participants to a website that described the study in detail.

2.2.2. Free Listing—Protocol

Each participant compiled their lists on a copy of a Google sheet that was designed as an elicitation tool. Column titles on the sheet were online shopping, search engines, messaging/texting, social media, educational services, audio/video streaming, and video games. They were listed in no specific order. Under each column heading, the sheet provided definitions and examples of each platform or functionality. Participants then listed what they perceived to be the risks and benefits of each platform or functionality in separate sub-columns. The listed items were then normalized by applying consistent labels to synonymous concepts.

2.2.3. Free Listing—Analysis

Saliency analysis of the risks and benefits identified across the platform functionalities was conducted using the FLARES Shiny R application (Wencelius et al. 2017). Items with the highest saliency were fed into the second task: pile sorting. The saturation of both risks and benefits was excellent. The log-fit score for the risk lists was R², which was 0.986. The log-fit score for the benefits lists was 0.993. The results indicate a very high level of saturation of concepts.

2.2.4. Pile Sorting—Recruitment and Sampling

For the pile-sorting task, a slightly larger pool of 11 parent/child pairs was recruited using the same methods from the same regional population for the pile-sorting task. As mentioned above, participants were recruited through social media posts about the research opportunity, as well as flyers posted in libraries and grocery stores with a website address and QR code that linked potential participants to a website that described the study in detail. As with free listing, the children participating were aged 11–12, and each child was accompanied by a parent who was also a participant in the study.

2.2.5. Pile Sorting—Protocol

Two randomized sorting decks consisting of high-saliency items generated from the prior free-listing task were created, with unique number codes on the back. Participants then sorted the cards into categories of their own choosing, with the most important items at the top and the least important at the bottom. Participating parent/child pairs completed each task simultaneously at adjacent tables to avoid mutual influence. Participants wrote labels on sticky notes to specify the relationship between the items grouped together in a column. A researcher on the team made a record of the order of items in each pile for each participant.

2.2.6. Pile Sorting—Analysis

A cultural domain is a category that holds shared meaning within a population. The analytical goal is to identify domains that consist of meaningful groupings of individual items from the perspective of the participant (D’Andrade 2004). To identify cultural domains using the pile sorting data, multidimensional scaling (MDS) was applied to each respondent’s card piles using the DOS-based software Anthropac 4.0 (Borgatti 1996). Anthropic makes it possible to generate visual representations of the cultural domains of interest. To accomplish this, Anthropac produces proximity matrices from pile sort data. Proximity matrices measure the relatedness between each individual item and every other individual item. Relatedness captures each time an item is included in a pile with another item. The proximity matrices for risks and benefits were mapped to produce visual representations of the cultural domain. Items clustered together are those that respondents most frequently grouped together within the domain. Items that appear farther away were infrequently (or never) associated with each other. The clusters that emerged were those that respondents most frequently grouped together within the domain.

2.3. Study 3: Online Survey

Based on an analysis of the interviews from Study 1, and concurrent with the implementation of Study 2, a survey was developed for broad online distribution to a representative sample of parents from across the United States.

2.3.1. Online Survey—Recruitment and Sampling

We engaged Prolific2 to recruit a stratified sample of 200 U.S. fathers and mothers. Demographic characteristics, including educational attainment and socio-economic levels, were unspecified and had representation across categories (see Appendix A).

2.3.2. Online Survey—Protocol

The survey consisted of six multi-part questions geared to elicit information on the demographics of the participants, their children’s usage of electronic devices, their fears with regard to the digital space, and the different parental controls they use. The survey was shared in two phases. First, a survey built in Qualtrics,3 was distributed to 100 adults from families with children ranging from 6 to 12 years of age. The distribution was split evenly between men and women. Participants received $5 USD for completion of the survey, which required an average of 10 min to complete. After the first round of 99 adults, the survey was modified to remove a rank-ordering question producing equivocal results and to make another free-listing question optional. For the second round, an additional 100 adults were surveyed. Taken together, a total of 199 participants were recruited who replied to the majority of questions.

2.3.3. Online Survey—Analysis

We asked participants to identify risks associated with different types of platforms in the survey. We analyzed responses to this question as a two-mode network in order to visualize the risks commonly associated with each kind of platform. This work was done in the R programming language (R Core Team 2021) with the tidyr (Wickham et al. 2024) and igraph (Csárdi et al. 2024) packages. In addition, we examined a range of demographic indicators, such as ethnicity and income, to examine whether these factors affected the thinking and actions taken by parents in online environments.

2.4. Study 4: Co-Design of Kids Tech Balance—A Prototype App—And Its Evaluation

Study 4 built on Studies 1 through 3 to create and test a new software prototype, Kids Tech Balance (Wendell 2024). Based on those studies, the new app sought to design and develop habits of intentional technology use and self-moderation. The work was carried out in two phases.

2.4.1. Co-Design—Recruitment and Sampling

The initial phase of work was carried out at the Human-Computer Interaction/Kidsteam lab at Boise State University using co-design methods. The co-design session included nine children ages 6–12, two researchers, and two assistants. Children were recruited from the area surrounding the university, which includes urban, suburban, and rural communities in close approximation. Samples were small and not randomized, but the interaction was extensive and deep. There were four girls and five girls on the team. Many had used technology, and their access to technology at home varied in frequency. Once the prototype, Kids Tech Balance, was co-designed and implemented, the work moved to an evaluation phase. First, nine different families evaluated its usability and efficacy for promoting communication and moderation of technology use. Then, four of the nine families used the prototype app in their homes for one week and provided additional feedback based on their experience.

2.4.2. Co-Design—Protocol

Co-design methods aimed to give children and their parents an authentic voice in shaping this new software. Their work stressed communication rather than restrictive controls. The co-design process (4a) required the cooperation of adults and children working together as a design team to explore issues or ideas using cooperative inquiry, which cycles through action and reflection across the research phases (Heron and Reason 1997). We first explored children’s opinions on parental controls and what tools might help them focus. We then gauged the team’s opinions on parental controls and how the controls could be improved. Over several sessions, we documented and incorporated participants’ responses and ideas into the app design.

2.4.3. Co-Design—Analysis

The co-design sessions were platform agnostic; however, we needed to implement something for testing purposes. For the prototype, we utilized iPads, so it was developed for iOS. In developing the Kids Tech Balance for the iOS environment, we used the SwiftUI4 coding language and Xcode5 for creating and designing applications. To ensure there were no access issues, we provided devices for people to use when using and providing feedback on the application. After creating a functional prototype based on the initial views of the participants, we conducted two evaluations: one (4b), where parents and children were invited to experiment with the app for one hour and provide direct feedback (n = 9 families, 25 individuals), and another (4c), where four families (n = 4 families, 9 individuals) took and used the application (and device) over a period of a week.

3. Results

The following subsections describe results from each of the four studies.

3.1. Study 1: Semi-Structured Parent Interviews Results

Study 1 provided some preliminary concepts that structure the landscape of parent–child concerns and strategies for managing children’s digital activities. These concepts include intrinsic and extrinsic risk, risk tangibility, digital fluency, trust and control, and parental resources.

3.1.1. Intrinsic and Extrinsic Risks

Parental concerns about their children’s digital lives and habits fall along a continuum of intrinsic to extrinsic risk. Intrinsic risks are internal to the individual and are based on a fear that the child’s technology use has the potential to develop into harmful compulsions or reduce self-regulation. Extrinsic risks develop through interactions with others, such as exposure to objectionable content, influence from outsiders with incompatible ideological dispositions, and vulnerability to being accessed in some manner by strangers or bad actors. Screen addiction was almost ubiquitously reported to be the main intrinsic concern of the participants. When asked about the allocation of time and resources to manage digital activities, dealing with screen time usually represented the largest investment. In order to deal with extrinsic risks, several respondents reported that they have rules aimed at limiting engagement with strangers, such as who their children can interact with on the chat features of online games. Parents very rarely specifically called out sexual predation as an extrinsic risk, but every interview participant discussed the general danger of interacting with adult strangers, perhaps couching that fear in less alarmist language. The classification of intrinsic and extrinsic risks is discussed in greater detail in Study 3.

3.1.2. Risk Tangibility

Our analysis of parental statements reveals a difference between threats parents feel empowered to understand and act on and those they do not. Thus, the risks we elicited in Study 1 fell on a continuum of tangibility, or perceptibility, by the senses. Risks that are easy to comprehend and act upon, for example, screen addiction, bullying, sexual content, and revealing locations to strangers, are more tangible. In contrast, less tangible risks include items such as data mining, browser tracking, corporate surveillance, targeted ads, identity theft, user agreements, and feed algorithms. Tangibility is likely a factor in determining parental action for mitigating risks. The costs for action are higher with increasing intangibility. We found that parental tactics are primarily aimed at easily addressable tangible threats, such as how much time kids spend using technologies and which platforms they are permitted to use. Dealing with less tangible risks would require time to understand and develop technical expertise, which did not emerge in these interviews as a pragmatic solution. Otherwise, parents turn to personal experience, anecdotal guidance, and simplified information that may be only partially accurate. Parents did not organically raise the topic of low tangibility risks, so we surmise that such concerns are relatively opaque features of the novel digital environment. We develop the topic of risk tangibility further in Study 2.

3.1.3. Digital Fluency

Parents’ levels of digital fluency varied significantly, and this appears to have a dynamic impact on the types of strategies they deploy to guide and protect their children. Our interviews suggest that low digital fluency is associated with less ability to discern between types of perceived threats and less awareness and use of parental control software. This was reportedly due to uncertainty about selecting from the variety of available software and their fundamentally technical nature, requiring the parent to acquire additional technical skills to operate. High digital fluency is associated with awareness of, if not use of, parental control software and a higher degree of specificity in their articulated perceptions of online threats. Based on this analysis, we hypothesize that the more digitally fluent a parent is, the better they are prepared to understand a greater number of low tangibility risks. However, this expertise is not always deployed. One respondent employed as a network analyst reported acting exclusively on concrete concerns.

3.1.4. Trust and Control

Parents struggled to balance their desire to maintain trust with their children with the need to control their children’s activities. Parents want their children to self-monitor and build critical thinking skills and hope to build and maintain trust: “I want my kid to be able to come to me”, said one parent. But this desire conflicted with wanting to monitor and impose constraints to keep them safe. Consequently, most parents mentioned passive methods of management, such as screen time limits or simply listening in on their children’s activities rather than directly reviewing their browsing history or setting up parental control software. These more passive techniques allowed them to reduce potential exposure to risk with minimal knowledge of the digital environment (low digital fluency) and less direct intervention. This aligns with other literature indicating that limiting time is a common regulation mechanism (Auxier et al. 2020).

3.1.5. Resources

Enrolling kids in extracurricular activities, such as sports, arts, academic clubs, volunteering, or leadership activities, seems to be a preferred tactic to minimize screen time when parents have the resources to do so. Parents articulated a hierarchy of desirable activities, with physical exercise and sports, educational and cultural enrichment, such as chess and piano lessons, and free play with other children valued highly compared with “pointless” (as one parent put it) activities like gaming, video streaming, or social media. Many respondents reported that they fill their kids’ days with diverse “real world” activities with the goal of promoting self-moderation and the ability to disengage from activities regarded as harmful or useless. Our findings in Study 1 motivated our research in Studies 2 through 4.

3.2. Study 2: Cultural Domain Analysis Results

Study 2 entailed an elicitive process to understand in more detail the cultural domains of risk and benefit for children’s presence in the digital environment. In this study we worked with parent–child pairs in order to see how similar or different their perceptions were.

3.2.1. Free-Listing Task

In the free-listing task, parents listed a greater number of values than children overall, and both parents and children listed more risks than benefits. On average, parents listed 17 benefits, while children listed 15 benefits. Parents listed 27 risks, while children listed 19 risks, on average. Parents and children were in greater agreement on the benefits they listed than the risks. This finding itself may reveal one of the challenges of parenting in the digital environment: generational differences in perception of online dangers. To illustrate,

Table 1. The top benefits and risks across platform functionalities from the free-listing task by adults and children and frequency of nomination. Concepts with tied number of nominations are included.

Benefits	Adults	Children
Friend, family connections	7	9
Fun, entertainment	6	15
Creative inspiration	5	-
Learning	5	7
Risks	Adults	Children
Strangers	10	-
Inappropriate content	9	9
Bullying	7	-
Mis/disinformation	7	5
Screen addiction	-	9
Wasting time	-	8
Scams	-	5

compares the most frequently mentioned values by parents versus children. There were three common values in the benefits list (fun/entertainment, family/friend connections, and learning). Children ranked fun/entertainment highest, while parents rated family/friend connections highest. In the top risks, there are two common values (inappropriate content and mis/disinformation). Children ranked screen addiction and wasting time higher than mis/disinformation, while those risks did not appear among parents’ top concerns. This difference may reflect what parents are saying to their children regarding intrinsic risks, as parents focused on extrinsic risks as their top threats. It is also important to note that mis/disinformation is a relatively low tangibility item, and both parents and children ranked this item lowest among their top risks.

In the analysis of free-list results, statistics of saliency can be generated. Saliency captures relevance to the cultural domain of interest. When items are listed by many or all respondents in a sample, these items have higher saliency. When an item’s saliency is higher, this means there is greater agreement about including an item in a cultural domain. Items listed by only one participant (or by no one) have little (or no) saliency within the cultural domain. FLARES provides several statistics on saliency for free lists (Wencelius et al. 2017). After normalizing and combining the lists of parents and children, the free-listing task generated 45 unique benefits and 46 unique risks. However, a number of items in each category were mentioned only once. So, we examined saliency statistics from FLARES and dropped items with the lowest scores. The final lists included 30 unique benefits and 34 unique risks for use in the pile sorting task. We aimed to use only the concepts clearly belonging to the cultural domain for the pile sort, as this task becomes increasingly difficult for participants as the number of cards increases.

3.2.2. Pile Sorting Task

The multi-dimensional scaling (MDS) of respondent pile sort data produces proximity matrices of benefits and risks, which we graphed (see Figure 1 and Figure 2). Where concepts cluster into groups, participants have perceived an underlying similarity. In addition, the overall graph can be structured by conceptual dimensions. Such clusters and dimensions contribute to a model about the cultural domain, in this case, the benefits and the risks of children’s utilization of digital space.

The graph of benefits in Figure 1 illustrates several clusters and two dimensions. On the far left are items associated with purchasing things online. Adjacent to that cluster are several items related to leisure. The top cluster is made up of items that contribute to knowledge and information within two sub-clusters: one specifically having to do with learning and education, and the other more specific to social or cultural learning. The bottom-right cluster contains items referring to social and community connections. Material benefits appear on the left, and nonmaterial benefits on the right. Social aspects of benefits are represented in the bottom half of the graph. For example, gift-giving in the purchasing cluster is at the bottom, indicating a more social aspect. The graph can be broken down into two overall dimensions: material/nonmaterial and nonsocial/social.

The graph of risks in Figure 2 is roughly divided into one dimension, with extrinsic threats on the top left and intrinsic threats on the bottom right. Extrinsic threats cluster into less tangible concerns at the top and more tangible ones at the mid-left. Two items are on the diagonal between intrinsic and extrinsic risks. Overspending is midway between a materialistic mindset and targeted ads. The item “internet is forever/can’t take it back” in the center of the graph has a close-to-equal relatedness to items in the different clusters. Intrinsic risks contain some small clusters, including a group on the left containing physical/emotional harm, comparing oneself to others, and damage to self-worth. Nearest to that cluster are more intrinsic “emotional issues”. There is a small cluster on the bottom right dealing with time issues. It is important to note that the MDS analysis of risks produced an intrinsic/extrinsic dimension and a tangibility dimension, replicating findings from Study 1.

3.3. Study 3: Internet Survey Results

3.3.1. The Landscape of Risk by Platform

Our internet survey included a question that associated the potential risks of children’s activities in the digital environment with seven types of platforms. This matrix-type question provided the same 10 risks for each of the platforms and asked participants to identify all the risks they associated with each platform type. The risks and platforms were derived from the results of Study 1 as the free-listing phase of Study 2 was progressing. We used the same seven platform/functionality types in this question that were used in the free-listing task in Study 2: social media, search engines, audio/video streaming, video games, shopping, educational services, messaging/texting, and others (with an option to add text). Only one participant added a potential platform, “calls”, likely meaning phone calls. Since the remaining 198 participants either left this option blank or added NA or “none”, we dropped this response from our analysis. The risks included are placed in this order: bullying/harassment, screen addiction, physical safety, mental well-being, financial/data privacy, explicit content, impaired self-regulation, harm to future opportunities, disinformation, and inappropriate behavior. Our analysis examines how respondents matched these 10 risks with the seven platforms, producing a two-mode matrix. The results are mapped as a bipartite proximity graph illustrating which risks are most closely associated with which platforms (Figure 3).

In Figure 3, the purple circles (or nodes) represent the risks, and the green nodes represent the platforms. Both risks and platforms are scaled to represent the number of times participants nominated each item. The lines (or edges) connecting each platform to specific risks are scaled to represent the relative number of times participants identified each specific risk with a platform. Because each risk-platform relationship was nominated at least once, relationships with fewer than 20 nominations were left off the graph. Otherwise, the figure would include every possible relationship and be difficult to interpret. Ties nominated between 21 and 100 times are represented with a thin line (weaker tie). Ties nominated between 101 and 199 times are represented with a thicker line (stronger tie). Thus, the graph represents both the weaker and strongest associations.

The relative location of platforms and risks in Figure 3 indicates their similarity. Considering platforms alone, social media is in the center of the graph and co-located near audio/visual streaming, messaging/texting, and video games. Thus, social media, audio/visual streaming, messaging/texting, and video games have similar risk profiles. This indicates that the types of risks associated with these platforms are more similar to one another than they are to the risks associated with educational services, for example. If we look more closely, we can see that social media has the strongest relationship with all of the risks except for physical safety and harm to future opportunities, where the relationship is weaker. Shopping, located nearest the financial/data privacy risk, has one strong relationship with that risk and several weaker relationships with other risks. Shopping, educational services, and search engines appear as more peripheral platforms on the graph—they have fewer strong connections to risks.

The risk in Figure 3, with the strongest relationship to the seven platforms, is mental well-being. It has strong relationships with four platforms. Bullying/harassment, inappropriate behavior, explicit content, and screen addiction each have strong relationships with three platforms. Disinformation has strong relationships with two platforms. Four risks are less strongly nominated across platforms: financial/data privacy and impairing self-regulation have the strongest relationships to just one platform each. Lastly, physical safety and harm to future opportunities only have weaker ties to platforms. It is important to note that of the six risks with strong connections to platforms, five are more tangible, and only one (disinformation) is less tangible. Of the four risks with few strong connections or mostly weak connections, three are less tangible, and only one (physical safety) is more tangible. This suggests that parents are identifying more tangible risks across platform functionalities.

A number of these risks are prioritized in ways that mirror the results of Study 2, including explicit (inappropriate) content, bullying/harassment, screen addiction, and disinformation. As far as the similarity of risks, screen addiction and mental well-being are in close proximity. Explicit content and inappropriate behavior are also in close proximity. Overall, the results of this question help inform a cultural model of risks in the digital space.

3.3.2. Parental Controls

One of our survey questions asked participants to select all the types of parental controls that they currently use or have used. We provided 12 options, plus “other” and “currently do not use parental controls”. All 199 respondents gave at least one response. Eight respondents indicated that they do not use parental controls. The frequency of nominations for the first through fourth responses is presented in

Table 2. The frequency of the first four respondents answers to the prompt “please indicate all types of parental controls you currently use or have used”.

Parental Controls	1st Response	2nd Response	3rd Response	4th Response
Child Profiles	85	79	0	0
Lockout Timers	85	0	0	0
Active Review of History	9	59	58	0
Passive Monitoring	7	25	58	44
Change Privacy Settings	2	9	14	32
Control Online Purchases	1	3	10	14
Malware Protection	0	5	11	13
Manual, Cellphone Timer	1	1	3	11
Monitor Texts, Calls	1	0	9	13
Web, App Blocking	0	4	1	12
Hardware	0	0	0	1
I Do Not Use Parental Controls	8	1	0	0

.

Two choices were tied as the most common choice for this question (first column on the left). Respondents most frequently selected either child profiles (such as YouTube Kids, Microsoft Family, Google Family Link, and Amazon Family/Household) or lockout timers. When respondents chose lockout timers first, they most frequently listed child profiles or active reviews of watch/browsing history as their second choice. Active and passive monitoring were tied for the most popular third response. Fourth was passive monitoring, followed by changing privacy settings and control online. For the fifth response (not included in the table), the most common choices were changing privacy settings (28 responses), controlling online purchases (23 responses), malware protection (17 responses), and using a manual/cellphone timer (17 responses). Thus, a typical response to this question began with lockout timers and/or child profiles, then added child profiles, active monitoring, passive monitoring, changing privacy settings, controlling online purchases, malware protection, and manual/cellphone timers. The remaining items, including hardware, were mentioned at low frequencies in the 6th through 13th responses. Notably, the more technically oriented solutions were least frequently nominated, indicating digital fluency is a factor in the choice of parental control strategy. There were no significant relationships between the respondent’s gender, income bracket, or education level and the number of controls selected in the survey (see Data Availability Statement).

As a follow-up question, we asked parents whether or not they felt that their children would benefit from more or fewer digital parental controls. Half (50%) of the 199 respondents said they felt more parental control would be beneficial. Eighty-five (43%) said they would benefit from neither more nor less parental controls, implying they were using the appropriate level. Approximately seven percent of respondents felt their children would benefit from less parental control. Although more men (n = 55) felt they would benefit from more parental controls than women (n = 45), this difference is not statistically significant (t = 0.9905, df = 197, p = 0.3231).

Since half of the online survey participants indicated that their children would benefit from more parental controls, Study 4 (discussed next) focused on the development of a parental control system with features that align children and parents with discussion and mutual understanding as children use technology.

3.4. Study 4: Prototype Software

Study 4 focused on the co-design of the prototype software and evaluation by families of its effectiveness in creating moderation habits for children.

3.4.1. Co-Design Results

The co-design team worked with nine children to co-create various tools and prototypes. Co-design sessions were platform agnostic as low-tech prototyping techniques were utilized. In these sessions, children communicated that they wanted tools that help them focus and set goals and facilitated mutual understanding between them and their parents. The children voiced that they wanted to be included in the process of determining device usage, not just an overall time limit. The team’s insights were applied to create an initial prototype as a multi-step software that brought together parents and children in setting up their device usage goals and associated time needs. When children’s device time ended, both parents and children were involved, creating a sense of mutual accountability. Overall, the team was pleased with the prototype design and felt the prototype app was “fair” to children, more so than other controls they had used.

The working prototype of the final design is shown in Figure 4. Users are first greeted with a page that prompts them to start a session. First, the goal-setting screen prompts the families to create a list of tasks or goals using the provided text box and associated time slider. Once families have created a list of goals, they move to the app selection screen to select relevant applications using the activities picker. After app selection, the families can start a session, which secures the device for the designated time and is ready for the child’s use on the use screen. The child can then use the selected apps and track their goals and time using the wiidget. Once the session is complete, the device will lock until parents approve the finished session and review the children’s usage on the Review screen.

3.4.2. Usability Evaluation

The usability evaluation of the prototype app involved two types of studies: usability interviews and a longitudinal study. The interviews consisted of nine families, including one adult and one or more children, who were able to attend an hour-long in-person demo. In all cases, participants were provided with a tablet with the working prototype installed on it. The families were able to briefly use the app and provide feedback on the experience. For the longitudinal study, four families were provided with iPads to take home for a week, during which they were asked to use the app three or more times. At the start and end of the week, the families completed a Digital Usage Questionnaire (see Appendix C) and participated in interviews to assess their pre- and post-usage perceptions. The survey consisted of ten questions targeted at assessing both the parent’s and children’s perceptions of technology use by the child. For both studies, the team documented the family’s feedback, sentiment, and whether they found it to be suitable for them.

The overall response to the Kids Tech Balance was positive from a parental perspective and mostly positive for children as well. The families’ responses indicated that goal setting and familial involvement were helpful in creating moderation habits when using devices. Many parents mentioned that the app helped create more discussion about device usage with their children than they had previously.

The families that participated in the longitudinal study provided detailed feedback on its applicability to their typical use. All four participating families found the app helpful and said they would use it if it was available on all of their devices. The four families stated that its simplicity and accountability features set it apart from prior controls they had used. Three of the four families were able to fully complete the Digital Usage Questionnaire. Two families completing pre- and post-usage questionnaires reported small increases in absent-minded use. One family reported a large decrease in absent-minded use. As the sample was small, the results are equivocal. This approach to evaluation is promising, and more quantitative data would be needed to validate a decrease in absent-minded use. The qualitative data signal advantages for the prototype approach.

4. Discussion

This section discusses how our findings address the research questions presented in Section 1.

4.1. The Landscape of Risks and Benefits

Our research contributes to understanding the cultural domains of risks and benefits of the digital sphere as perceived by U.S. parents of youth in middle childhood (RQ1). Semi-structured interviews in Study 1 and cultural domain tasks in Study 2 revealed distinctive features of parental concern for children’s digital technology use, predominantly in one U.S. state. Study 3 extended these findings to a larger sample within the U.S., linking a consolidated list of risks to common digital platform functionalities.

Major findings we generated relevant to RQ1 include the following:

First, parents differentiate intrinsic and extrinsic risks. Intrinsic risks include risks related to children’s mental health, time management, and other skills and capabilities. Extrinsic risks include perceptions that the internet is an environment in which there are dangers to be navigated and resources to exploit. These are not limited to sexual predators and include corporations, scammers, and advertisers. These constellations of risks are visible in Figure 2, which displays the similarity of risks to each other.

Second, like risks, benefits also break down into constellations of similar items. However, the overall domain of benefits appears less complicated than risks. The benefits domain breaks down more neatly into three clusters, as seen in Figure 3.

Third, parents prioritized more tangible risks over less tangible risks. This finding is in line with predictions from optimal foraging theory in evolutionary anthropology, which models decisions assuming perfect information. If perfect information is not available due to uncertainty, then people either exert effort to gain that information, or people make decisions based on what they think might be optimal, or play it safe (Smith 1983). Considering the relative novelty of the emerging digital sphere, it is likely that information about some kinds of risks—such as loss of privacy—will not be immediately clear, and so most parents are making decisions based on what they think might be optimal or are playing it safe by exerting more parental controls. New risks are developing all the time in the digital sphere, such as the quick expansion of applications using large language models (LLMs) and artificial intelligence (AI), which complicates efforts to gain information.

Fourth, parents and children agreed more on benefits than risks. While this and the other findings from Study 1 and Study 2 are based on a very small sample of participants, we did find consistency across the two studies with different participants. Further, seeing differences between parents and children is consistent with anthropological studies showing learning in age-peer groups plays an important role in practicing skills in a social environment where scaffolding can occur (Reyes-García et al. 2016).

4.2. Current Parental Responses

Part of the results from Study 3 inform the strategies parents take to address their children’s use of digital technology while also providing insights into the prioritization of these strategies and the role of digital fluency and other social characteristics in the choice of strategies (RQ2).

The frequency of parental control strategies ranges from relatively widespread, tried-and-true methods to relatively infrequent technological approaches. This indicates that digital fluency is a likely determinant of choice of strategy. Common parental controls include lockout timers and child profiles, using products that are ubiquitous and do not require a great deal of digital fluency. Next, parents might add “analog” methods of control, such as active and passive monitoring of texts/calls and online activities. More technically demanding approaches requiring web/app blocking and additional hardware, such as routers and WiFi configurations, were relatively rare.

The results of Study 1 suggest parental strategies might be informed by risk tangibility and digital fluency. Interestingly, however, we found no correlation between gender, education level, or income bracket with the number of parental controls survey participants selected. It seems reasonable to expect that digital fluency increases with education level, but we found no relationship. It is likely that the concept of digital fluency needs to be better operationalized to understand how digital fluency informs parental strategies. Nevertheless, it is clear that parents spend time, effort, and money implementing controls. Minimally, resources spent on online monitoring are not directed to other dimensions of child development. This aligns with the digital ecology of fear concept.

Based on the survey results from Study 3, most parents are utilizing some kinds, if not multiple, systems of online controls, and many want more. This finding also aligns with the digital ecology of fear concept, in that parents are investing in vigilance against potential threats. Developing new approaches for parent involvement in their child’s online activity based on building trust also needs to address the potential for harm to the child’s (and parent’s) mental health.

4.3. Cultivating Parent–Child Adaptive Responses

Study 4 addressed how to move beyond top-down parental controls. We developed and piloted a new approach in a prototype app based on co-design sessions involving both children and adults. We discuss the outcomes of this exercise considering life history theory. We found that co-designed solutions provide a promising approach that supports parent–child adaptive responses to digital risks and promotes positive mental health outcomes. This addresses RQ3.

Regular, continued usage of digital devices can have a negative impact on both the physical and the mental health of the user, and these effects are particularly concerning with children. A common parental strategy is to place limits or controls on children’s devices. Controls only go so far, however, because nothing positive about the experience is being generated.

Our prototype app, Kids Tech Balance, structures the digital experience of parent and child into three phases: before the child is online, during their online activity, and after their activity concludes. The pre-session includes prioritizing activities and setting time limits through discussion before a screen time session begins. After the child’s session with the digital device, the app sets aside time for a post-session reflection and discussion, which allows parents and children to talk about what went well and what did not go according to plan, as well as the content. This approach aligns with best practices (Gruchel et al. 2024) and likely supports parent and child mental health.

Prioritizing activities is a key feature in the development of executive function (EF). Executive function is the group of skills that support self-regulation and help children regulate the flow of information, pay attention, plan ahead, and remember and follow rules. Early life stresses are known to negatively affect EF and self-regulation (McClelland et al. 2018). Importantly, our prototype’s pre-session supports developing these skills. The review and discussion during the post-session support a healthy developmental environment known as “serve and return” (National Scientific Council on the Developing Child 2009). Positive interactions involving back-and-forth play between children and caregivers help build strong brain architecture, providing a foundation for learning healthy behavior. Empirical studies testing the adaptive response hypothesis (Snopkowski and Ziker 2020; Ziker and Snopkowski 2020) show that such social interactions in early and middle childhood are the strongest factor working against adverse childhood experiences and outcomes associated with dysregulation.

The results of Study 4 substantiate the proposition that co-design can be utilized to develop new infrastructure that supports EF skills, discussion, and relationships that will ultimately build resilience to the risks of using digital devices. The results of the pilot studies (4b and 4c) reveal positive responses from parents and children that enable a cooperative approach to managing technology use. This approach goes beyond control to develop trust and relationships around children’s digital experiences.

The development and testing of a new approach to parental controls that favors skill building and encourages parent–child discussions can support executive function and increase resilience to risks. Bringing together goal setting and time management with pre-session prioritization, discussion, and post-session reflection helps support social relationships (and trust) between parents and children.

4.4. Features and Limitations

Our project used a mixed-method approach to understand parent and child perspectives on costs, benefits, and risks in the digital space and control strategies for parental engagement. While three of our studies center on small non-probability samples (Study 1, Study 2, and Study 4), our findings are robust in their triangulation, producing similar results using different methods. The unbalanced proportions of women and men in Study 1 do not detract from our overall findings that were replicated in later studies using other methods. The saturation of concepts elicited in Study 2 was excellent. The survey analyzed in Study 3, utilizing a much larger sample and stratified with women and men in equal proportions, reproduced the dimensions of the landscape identified in Study 1. Study 3’s participant sample recruited through an online service was not nationally representative in other respects, however, and may be biased to higher income brackets (see Appendix A).

In addition, unlike other national surveys, such as those of the Pew Research Center and FOSI, we focused on children in middle childhood (ages 6 to 12). One difference in our results, compared with the FOSI survey, is that the number one concern of parents of children in middle childhood is strangers, while the FOSI survey reported child predators in the middle of the list of concerns with inappropriate content at the top, followed by bullying (Family Online Safety Institute 2020). The FOSI survey addressed a broader age range of children (ages 2–17) than we did. The differences in results can be interpreted from a life history perspective. Middle childhood is the time period when children begin to build relationships with people outside the family, while development toward sexual maturity occurs in adolescence. Thus, according to life history theory, parental concerns and attention should also shift with the changing risks. In addition, our findings, such as the prioritization of risks based on their tangibility, align with considerations from optimal foraging theory in evolutionary anthropology.

The qualitative evaluation of our prototype app is encouraging. Future work will focus on developing a larger sample of evaluators to employ the quantitative evaluation. In addition, future work can focus on measures of EF and mental health outcomes for longer-term use of the app.

5. Conclusions

In this project, we successfully integrated anthropological methods and theoretical concepts and the concepts and methods of human–computer interaction (HCI) to learn about parental fears and strategies for guiding children’s online activities. We described major dimensions of parental perceptions of the benefits and risks posed by children’s internet use and current parental responses and tested a new co-designed app that can improve parent–child engagement in relation to online activities.

Proximity graphs produced through multidimensional scaling revealed related groupings of benefits and risks associated with children’s use of digital technology. Parents and children tend to agree more on the benefits but have more diverse perspectives on the risks. Children appear to prioritize what parents are saying with regard to time spent on digital devices and the potential for digital addiction, while parents themselves rate that risk much lower. Intrinsic and extrinsic dimensions of risks and tangibility of risks are apparent in the mapping of risks.

Our online survey of 199 U.S. adults found parents appear to prioritize more tangible risks and prefer to use strategies that are easily implemented. Parental concerns for their children’s safety align with predictions from life history theory about the effects of early and middle childhood stress that can lead to the development of dysfunctional behaviors, such as decreased ability to delay rewards, prioritize goals, and create positive social relationships. These concerns shape the digital ecology of fear that parents react to using parental controls, including child profiles, lockout timers, and active and passive monitoring. The good news is that many parents are actively monitoring their children’s use of digital technology. Hopefully, many are engaging in conversations with their children about what their children are experiencing online. Kids Tech Balance, the co-designed prototype app developed as part of this project, specifically supports such conversations as well as skills important for executive functioning, helping to build digital resilience.

Technological approaches that encourage parent–child discussions about their digital experiences can help build digital resilience. The potential impact of the proliferation of technologies—including the recent rampant proliferation of AI tools—on how children interact with the digital environment redoubles the importance of developing strong digital resilience strategies, including relationships between parents (and other caretakers) and children, particularly with regard to goal setting prior to use, and reflection after use about what the child is experiencing.