Examining the Relationship between Broadband Access, Parent Technology Beliefs, and Student Academic Outcomes

Abstract

This study explores the relationship between community broadband access, parent technology use and beliefs, and student academic outcomes in a Southeastern U.S. school district during and after the COVID-19 pandemic. By applying a quantitative exploratory approach and multiple regression analysis, the research revealed that parents’ technology beliefs and use were significantly associated with students’ standardized Math and ELA scores. While parents’ positive beliefs about technology’s developmental impact showed an inverse relationship with student performance, greater concerns about familial challenges related to technology were associated with lower scores. The results explored persistent social inequities, whereby students who were identified as Black historically score lower than White-identified students, emphasizing the role and importance of digital equity to mitigate disparities and enhance outcomes. Both the methodology and results encourage researchers, educational leaders, and policymakers to deeply consider the role of parental access to and engagement with digital resources, and how they contribute to shaping student outcomes.

1. Introduction

Over the last few decades, the proliferation of connected personal devices in households and education settings has profoundly changed educational opportunities for teachers and students as well as how parents engage and interact with their children’s school. Although low-cost internet connected technology holds promise for correcting historical inequities in access and use of educational content and resources, many unfortunate inequities across our global society have pervaded into the digital age. To bridge the persistent digital gap especially among underrepresented groups, the United States Digital Equity Act represents a massive policy promise to provide USD 2.75 billion in support of digital equity programs. For example, the National Telecommunications and Information Administration approved the California State Digital Equity Plan in April 2024 that aims to provide affordable internet service and digital literacy training to state residents [1].

Given the substantial investments in such programs, it is crucial to evaluate their impact on addressing digital inequities and related educational inequities. Evaluation study results can provide insights into how similar initiatives can be implemented and improved in broader settings across regions and nations.

2. Literature Review

According to the United States Digital Equity Act, the digital gap can be defined as the inequity between those who have affordable access, skills, and support to effectively engage online and those who do not [2]. In the years preceding the COVID-19 pandemic, the National Center for Education Statistics (NCES) estimated that 70% to 90% of K-12 students in the U.S. had access to home internet and digital devices [3], still leaving approximately 16 million students with limited access/connectivity [4]. More recent estimates from [5] and the U.S. Census Bureau [6] suggest more than 90% of all U.S. households now have home broadband access, with notable variations observed across geographic regions, as well as by household racialized/ethnic identity, parental education, and household income.

The gap in broadband accessibility has disproportionately impacted families from traditionally disenfranchised communities, including racial and ethnic minority groups and low-income families [7]. In 2017, the U.S. Department of Commerce’s National Telecommunications and Information Administration (NTIA) estimated there were about 7 million school-age children without broadband access at home, with over 60% representing families with an annual income below USD 50,000 [8]. Additionally, in many U.S. communities, Black and Brown students have been found to have significantly less access to home technologies than Asian or White students [8,9,10]. In a Los Angeles study by Galperin et al. [11], African American and Hispanic students were found to be 35% and 50% less likely to have resources for remote learning compared to White students, a pattern observed regardless of location or family income. Similarly, in a pre-pandemic Michigan investigation, students receiving free and reduced lunch assistance were found to be 25% less likely to have home broadband access compared to students not receiving free and reduced lunch assistance [12]. For many, historic inequities have only been perpetuated by limited access to technology resources and broadband.

Using data from the Tennessee Department of Education, economically disadvantaged students across the state scored about 17% lower in Math and 18% lower in ELA compared to non-economically disadvantaged students in 2023. Furthermore, across all grade levels, students who were identified as Black scored about 1% lower for ELA and 1.5% higher for Math than the economically disadvantaged student averages [13]. Research has shown the test score gap between Black and White students has persisted for decades and exists in every age group [14]. One explanation for the Black–White gap in ELA test scores was the effect of vocabulary differences under the influence of culture, habits, social classes, parenting styles, skills, and knowledge [14]. According to Farkas [14], ‘by the age of three, the professional parents had spoken 35 million words to their children, the middle- and working-class parents had spoken 20 million words to their children, and the lower-class parents had spoken only 10 million words to their children’. Traditionally, Black households in the U.S. have also suffered inequitable employment and income opportunities. As such, research studies examining test score gaps have found that family background explains as much as half of the Black–White student test score differences [14].

Other scholars would emphasize cultural and psychological differences such as how parents and children react to circumstances rather than emphasizing socioeconomic differences such as income and resources to explain the Black–White gap [15]. Additionally, Rothstein and Wozny [16] found that permanent household income (long-term average income) explained substantially more variance of the Black–White test score gap than a household’s current annual income, since permanent income was found to be a more stable indicator of long-term income and spending. When predicted using permanent income, the Black–White test score gap typically shrank to 0.2 to 0.3 standard deviations, indicating that the gap might be more associated with long-term household income and economic spending habits than the household’s cultural and racial background [16]. Annual income as an indicator of parental socioeconomic status (SES) has its limitations, but permanent income as an independent variable provides a more nuanced understanding of the social phenomenon [16], showing that many important educational outcomes can be influenced by a student’s family income and exposure to poverty [17].

Despite the lofty promises of digital equity, access to digital resources may be serving to widen the educational inequities evidenced by test-score performance gaps. The digital divide has a long-term impact on students’ academic performance, digital skills, and STEM career interests [12]. In a study of rural Michigan middle and high school students, Hampton et al. [12] used a multilevel modeling approach to show that students in districts with high-speed internet access at home acquired more digital skills, exhibited more interest in pursuing STEM careers, and performed better on SAT standardized tests regardless of their socioeconomic status. However, the relationship between household internet access and classroom GPA was minimal [18]. After publishing their initial study of pre-pandemic home broadband conditions, Hampton et al. [19] later found major increases in household internet access, time spent using technology, and students’ self-reported digital skills during and after the pandemic, but they obtained little post-pandemic data on students’ academic performance.

In addition to studies of home broadband access and parent use, researchers have additionally explored the evolution of stakeholder attitudes and beliefs towards technology [20,21]. Helsper [21] suggested attitudes and beliefs about availability, appropriateness, regulations, and the effects of technology are an integral part of understanding and measuring the digital divide. Over decades of research, it has been recognized that parents’ attitudes and engagement are important factors in influencing students’ motivation and achievement in school and their growth in general [22,23,24]. The role and influence of parent engagement in the post-pandemic landscape has undoubtedly shifted with increased school and household technology resources and internet access, but this has not been well-documented in the literature.

Few studies on community broadband program impacts have empirically examined their relationship with student achievement using community or school-level data. It is noteworthy that the current paper aims to be the first published research examining broadband impacts on individual student/household level use and achievement data. For example, Caldarulo et al. [25] examined 2009 to 2018 county-level aggregate measures of grade 3 to 8 standardized test results across 784 U.S. counties collected from the Stanford Center for Education Policy Analysis (SEDA). The average county-level broadband subscription rate was combined with estimates of county-wide broadband connectivity rates using the Current Population Survey (CPS) and the American Community Survey (ACS) during this same time period. Using two fixed effect models to explore the combined national dataset, the researchers found that broadband adoption rates were positively associated with standardized test scores in both Math and English language arts (ELA). Additionally, the authors observed Black and Hispanic students realized greater academic benefits from higher broadband subscription rates than White or higher SES students.

Community broadband programs not only benefited students during the pandemic, enabling many to continue their education, but they also provided students’ families with opportunities to utilize the internet and digital devices to stay connected with schools and support their children’s education. Previous studies support the positive influence of parents’ engagement on students’ academic motivation and achievement [26,27,28].

In Anthony and Ogg’s [26] longitudinal study, they examined how students’ attitudes and behaviors as well as academic achievement were related to parents’ engagement from kindergarten to third grade. The authors compared parents’ school-based involvement, home-based involvement, and home–school communication with respect to students’ reading achievement and approach to learning. With more than 12,000 students drawn from the U.S. Early Childhood Longitudinal Study dataset, their results showed that home-based involvement was not statistically significantly associated with children’s reading achievement, and there was only a weak association between home–school communication and reading achievement. The researchers suspected home-involvement instruments lacked validity and that the items did not fully focus on measuring home-based academic activities. Recognizing this limitation, the current study includes both home-based involvement measures such as parents checking students’ homework and grades and home–school communication measures such as communicating with teachers and checking school events.

Although Anthony and Ogg [26] found a limited association between home-based parent involvement and student academic achievement, the paradigm of parental involvement with student education changed dramatically during the pandemic [29]. A more recent exploratory mixed methods study conducted by Nobis and Caparroso [30] underscored how leveraging technology can be an important strategy for Filipino parents to help improve their high school children’s Math homework completion rates. In student interviews, distractions such as social media were the most frequently mentioned factors preventing students from completing homework. Although many students felt technology can negatively impact homework completion, students and parents both recognized technology was also a valuable resource for homework. In their parent survey, one of the top strategies parents employed to improve students’ Math homework completion was ‘leveraging technology’. The value of technology in helping students with their homework was rated higher by parents than hiring a tutor, setting academic goals, or even communicating with teachers. Nobis and Caparroso [31] cautioned that using technology to boost student learning requires parental guidance. Recognizing that parents’ use of technology is often shaped by their attitudes and beliefs, the current study explores how both parents’ practices and beliefs about technology may potentially relate to their students’ academic achievement.

In summary, the lack of broadband access can exacerbate the existing achievement gap, especially for students from underrepresented and low-income households, as students without reliable internet and technology at home face significant barriers in completing homework and engaging with online educational resources. In addition to access, parental beliefs about technology and their own use of it also influence students’ academic performance. Parents who view technology positively and use it to support their children’s learning are more likely to foster an environment where technology is used as a tool for educational enrichment. These parents may be more engaged in their child’s education through online resources, educational apps, and communication with teachers, which can further boost student achievement. Together, broadband access and parental engagement with technology shape students’ educational experiences and outcomes, highlighting the importance of addressing both access and attitudes to close the achievement gap.

3. Research Questions

The need for remote learning during the COVID-19 pandemic necessitated governments, schools, and organizations across the globe to address broadband connectivity disparities. As one response to the pandemic, a community broadband initiative at Hamilton County Schools was launched in Summer 2020 with a 10-year commitment to provide free high-speed home internet access across a Southeastern United States school district. In partnership with the local broadband provider, numerous community and civic organizations, as well as the school district, the HCS EdConnect program enabled many families and students to more fully engage in remote learning. By May 2023, the EdConnect program had connected more than 16,000 students and their families to no-cost high-speed broadband internet and was considered to be one of the largest and most far-reaching broadband programs in the United States.

To better understand this effort to close the broadband access gap, an evaluation study was commissioned to explore the community-wide impact of the EdConnect broadband initiative. Guided by the Culturally Responsive Evaluation (CRE) framework and part of this larger study, the current investigation seeks to assess the effectiveness of the broadband program in promoting equity, contribute to the broader efforts of closing the digital divide, and improve educational outcomes for students across this historically disenfranchised community by addressing the following research questions:

RQ 1: How do households’ participation in a free community broadband program relate to their children’s Math and ELA achievement?

RQ 2: How do parents’ beliefs and use of technology relate to students’ Math and ELA achievement, controlling for household participation in a free community broadband program?

RQ 3: How do students’ personal and household characteristics relate to their Math and ELA achievement, given their parents’ beliefs and use of technology, and their households’ participation in a free community broadband program?

4. The EdConnect Study

Summarizing prior study results from the EdConnect broadband initiative as early as Spring 2021, it was found that the vast majority of randomly sampled parents used technology consistently and frequently for supporting the education of their school-aged children. Overall, the first household survey results found that 93% of all parents (including both program participants and non-participants) had used technology at home in the last month to interact with their child’s school. Among surveyed practices, parents reported the most frequent use of home technology was accessing information about their child’s grades or performance, followed by obtaining information about homework or assignments. In addition, about two-thirds of parents reported they had communicated directly with their child’s teacher or school at least several times during the past month.

Table 1. Scheme 2022. Parent frequency of technology use related to their child’s school.

During the Past Month, about How Often Have You Used Technology in Your Home for the Following:	Never	Once or Twice	Several Times
Accessing information about your child’s grades or performance in school.	16%	14%	71%
Obtaining information about your child’s homework or assignments.	21%	16%	63%
Communicating with your child’s teacher or school.	18%	20%	61%
Obtaining information about a school event or schedule.	18%	17%	65%

summarizes the 2022 parent survey results used for the current analyses and demonstrates that the majority of households across the study continued regular technology practices related to their child’s school.

As more thoroughly described in prior research, an exploratory factor analysis (principal axis factoring with varimax rotation) was employed to consolidate the household survey items into four quantitative scales [32,33]. A parent use scale explained 78% of the variance across four survey items asking parents to report their frequency of using technology to check students’ homework, grades, and school events, and to communicate with teachers [32,33]. A second scale, developmental impact, explained 19% of total variance among survey items asking belief and attitude questions. This scale aggregated parents’ beliefs about the effects of digital devices and technology across children’s academic skills (reading, Math) and non-academic skills (social, attention span, creativity). A technological utility scale captured parental views on the utility of digital devices, including increased efficiency, communication, entertainment, and teaching, explaining 16% of total variance across the belief and attitude questions. Lastly, the familial challenges scale focused on perceived negative impacts of digital devices such as being a catalyst for strife between parents and children and their domination of children’s attention span. This scale explained 12% of the total variance. Cumulatively, the three parent belief scales explained a total of 47% of the variance among survey items focused on parent beliefs. The mean, standard deviation, and range for the parent use and belief scales are displayed in

Table 2. Characteristics of the parent use and belief measures.

	Mean	Std. Dev.	Min	Max
Parent Use	0.05	0.89	−2.00	0.71
Developmental Impact	0.06	0.95	−2.63	1.91
Technological Utility	−0.03	0.91	−3.82	1.33
Familial Challenges	−0.006	0.96	−3.0	1.60

.

Plotted using a standardized z score, larger parent use scores indicate a greater frequency of technology use to support their children’s education. Likewise, larger developmental impact scores infer more positive perceptions towards the cognitive benefits afforded by digital devices. Greater technological utility scale scores indicate more positive parent beliefs towards the pragmatic advantages of technology, while greater familial challenges scores infer increased parent perceptions towards difficulties regulating their children’s technology use. The reported range across parent beliefs was greater than that observed for parent use, indicating parents’ beliefs had more overall variability than parent practices.

Examining the first two years of EdConnect household data, researchers found that parent beliefs exhibited a weak but significant relationship with household background, and inconsistent impacts across program enrollment. For example, parents from households categorized as Hispanic/Latinx or Black held significantly more positive attitudes toward technology’s developmental impact on their children compared to parents from households categorized as White. Similarly, parents from program-enrolled households reported statistically more positive attitudes regarding the impact of technology on their children’s Math skills than parents from non-enrolled households. Overall, researchers found some evidence of a relationship between households’ broadband access and parents’ supportive use and beliefs towards technology during the first years of the program [32,33].

Providing the household data for the current investigation, the EdConnect household survey was administered to a stratified random sample of 400 households containing at least one student enrolled in grades five through eight in June 2022. The household sample was stratified to reflect varying participation levels across the greater community/population in the free broadband program.

5. Culturally Responsive Evaluation Framework

From its inception, the EdConnect study has employed a Culturally Responsive Evaluation Framework (CRE) to emphasize equity and power balance within this diverse cultural setting [34]. By including the perspectives of historically marginalized groups and involving a diverse and representative group of stakeholders, the aim was to ‘create accurate, valid, and culturally-grounded understanding of the evaluand’ [34]. As one step, a Community Advisory Board composed of community organizers and activists were consulted across each step of the research to better comprehend and capture the challenges faced by the diverse communities served by the program.

In line with the Culturally Responsive Evaluation (CRE) framework, this study also draws on multiple data sources from the local broadband provider, the school district, and a parent survey, aiming to accurately reflect the realities of underrepresented and historically marginalized groups.

6. Methodology

Embedded in a larger IRB-approved study of a community broadband program, the current study explores the relationship between parent practices and students’ academic performance through multiple regression analysis using de-siloed household survey results and student achievement data. Specifically, the current study examines the relationship between various student, parent, and household factors including parents’ beliefs and use of technology, households’ broadband participation, students’ racial/ethnic identity, gender, economically disadvantaged status, and English language learner status, and two outcome variables: student-level Math and ELA Tennessee state test scores. To provide the most nuanced examination of student achievement, individualized, student-level scale scores are employed throughout the current analysis, rather than performance levels or cohort data.

Study Setting and Data Sources

Hamilton Public Schools in an Tennessee anonymous State in the USA enroll approximately 45,000 students across 79 urban, suburban, rural, and virtual schools in a geographically large (over 550 square miles) and culturally diverse community. Working closely with the community, the research team established data sharing agreements and protocols to access the requisite data to quantitatively address the research questions. Specifically, the local broadband utility company provided data showing household enrollment in the community broadband program, the school district provided student-level achievement data, and the research team collected parent opinions through a June 2022 survey of 400 randomly selected households.

R was employed for merging parent survey data with household participation and student-level achievement data, as well as Microsoft Excel 2408 for additional data cleaning and STATA BE 18 for data analysis. The parent survey data and student achievement data were merged based on unique, anonymized identifiers provided by the district. The 400 household survey records matched 716 students with district-supplied Spring 2022 assessment data. After removing students who left the school district, non-participating-grade-level students, and students lacking complete assessment data, a total of 511 students from 382 households remained with both matched standardized Math and ELA data for subsequent data analysis.

Student-level background characteristics such as grade level, socioeconomic level, and ELL status were also shared by the district and are summarized below in

Table 3. Sample characteristics of the student-level Hamilton County school district data.

Variable Name		N	%
Econ disadvantaged
	Yes	272	53%
	No	239	47%
	Total	511	100%
Student gender
	Female	255	50%
	Male	256	50%
	Total	511	100%
ELL
	Yes	35	7%
	No	476	93%
	Total	511	100%
Student grade level
	3	36	7%
	4	26	5%
	5	98	19%
	6	121	24%
	7	127	25%
	8	103	20%
	Total	511	100%

for the resulting study sample.

As summarized in Table 3, nearly seven in ten students in the study sample were middle school students. Across the Spring 2022 sample, the majority of students were native English speakers (93%), while nearly half were deemed ‘economically disadvantaged’ by the district.

Using the broadband provider-sourced data, each households’ program participation was dichotomized based on enrollment status at the time of the parent survey. For example, ‘Enrolled’ households were actively enrolled in the program and connected to no-cost broadband prior to the survey, while ‘Unenrolled’ households, although eligible, had declined service, were non-responsive to offers to join the program, or had not completed enrollment in the program.

In addition, the district provided household-level demographic information including racialized identity, ethnic identity, and free and reduced lunch status for all students in the study. As such, this paper employs the use of the phrase ‘racialized/ethnic identity’ when generally referring to a student’s identified race and/or ethnicity in recognition that race is a social construct and racialized categories are constructed for social purposes.

Table 4. Distribution of household participation in the free community broadband program by student racialized/ethnic identity.

Student Racialized/Ethnic Identity	Non-Participating	Participating	Total
Black (48%)	117	126	243
Hispanic (14%)	37	32	69
White (35%)	138	42	180
Other BIPOC (4%)	12	7	19
Total (100%)	304 (59%)	207 (41%)	511 (100%)

shows the distribution of the student-level racialized/ethnic identity data by household-level broadband participation rates.

As Table 4 shows, about 48% of study-wide students’ racialized/ethnic identity were categorized as Black, while 35% of students were categorized as White, and 14% were categorized as Hispanic/Latinx. The remaining racialized/ethnic identity categorizations provided by the district (Asian and Native American) had very small resulting sample sizes and were combined into an ‘Other BIPOC’ category to help compensate for analytic limitations. Compared to the overall population of the school district with 73% of households categorized as White, the study sample was disproportionately Black and Hispanic. This was because program-eligible households tended to have lower SES, which, in this historically inequitable community, was associated with racial/ethnic identity.

The participating school district provided the research team with anonymized, student-level Tennessee state assessment data for the 2021–2022 academic year. As the state sanctioned an annual achievement indicator used across the state, the Anonymized State Test (AST) was designed to measure students’ learning outcomes annually for all students across grades 3–8 and for high school students in more limited subjects. Given prior local, state, and national education research and reform emphasis, the current study focuses on students’ Math and ELA scores in grades 3–8 [35]. The standardized Math exam is designed to measure students’ ‘understanding of mathematics, number sense, fluency, [and] problem solving’ [36], while the ELA test assesses students’ ‘ability to read closely, analyze text, answer text-dependent questions, write a response to a prompt, and demonstrate command of the English language’ [37].

As widely documented across the educational literature, students’ standardized achievement scores are often related to socioeconomic status and racialized/ethnic identity. Exemplifying these patterns in the study community, Figure 1 and Figure 2 show the distribution of 2022 district-wide grade 4 student ELA and Math achievement across the community’s socioeconomic indicator (economically disadvantaged) and the three largest major racialized/ethnic student categories.

Figure 1 shows how the distribution of district-wide ELA and Math achievement scores for 4th grade students who were identified as economically disadvantaged (ED) were generally lower than non-ED students. Figure 2 additionally shows how students’ racialized/ethnic identity related to achievement patterns across the district, with 4th graders who were identified as Black scoring lower in both ELA and Math, on average, than students who were identified as Hispanic/Latinx or White.

7. Results

7.1. RQ 1: How Do Households’ Participation in a Free Community Broadband Program Relate to Their Children’s Math and ELA Achievement?

Using the four parent use and beliefs scales as indicators of general parent practices and attitudes in the community, Figure 3 plots the relationship between Spring 2022 parent use and beliefs and their Spring 2022 students’ ELA and Math scale scores.

Figure 3 shows the distribution of the four technology use and beliefs factors related to students’ Math and ELA scores. Overall, the distribution of Math and ELA scores and the four factors exhibit the same pattern. The data points are vertically spread out, indicating that for similar levels of parent use, technological utility, developmental impact, or familial challenges, there was a wide range of student Math and ELA scores. This variability suggests that such factors do not fully explain students’ Math academic performance. Apart from the developmental impact factor, most data points are clustered towards the right side of the plots, suggesting a generally high frequency of technology use and a tendency to agree with technology’s utility value and associated familial challenges. The fitted lines on the scatter plots indicate a slight positive relationship between Math and ELA scores and the factors of parent use and technological utility. Conversely, there is a weak inverse association between Math scores and the factors of developmental impact and familial challenges. Collectively, the plots provide limited evidence of linear relationships. Summarizing all of the relationships in Figure 1 and Figure 2,

Table 5. Correlation between scaled scores and parent use, developmental impact, technological utility, and familial challenges.

	ELA	Math	Par Use	Dev Imp	Tech Uti	Fam Chal
ELA	1.00
Math	0.66	1.00
Par Use	0.04	0.03	1.00
Dev Imp	−0.16	−0.12	0.10	1.00
Tech Uti	0.06	−0.02	0.33	−0.06	1.00
Fam Chal	−0.14	−0.17	−0.05	0.04	−0.01	1.00

displays all of the Pearson’s correlations between students’ Math and ELA scores and their parents’ use of technology for supporting/connecting with their school as well as three parent beliefs scales.

Generally, the relationship between the Spring 2022 parent use and belief scales and their students’ achievement indicators exhibited fairly small correlations overall. As expected, the strongest correlation was observed across the outcome measures for Math and ELA scale scores (r = 0.66), such that students’ performance on one subject was often very similar to their performance in other subjects. The next strongest correlation was a moderate relationship observed between parent use and technological utility (r = 0.33), indicating that parents who used technology more frequently were also more likely to recognize the utility of technology. Comparing the parent use and beliefs scales to the student ELA and Math outcomes, relatively weak correlations were observed between students’ Math and ELA scores and developmental impact and familial challenges, respectively.

To summarize the overall impacts for RQ1, a simple linear regression examined the relationship between household participation (enrolled in the broadband program vs. not enrolled) and their students’ Math and ELA scores in Spring 2022.

Table 6. Regression results for math and ELA achievement by household enrollment in EdConnect broadband program.

Math Scores as the Outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Household Participation	−7.69	3.91	−1.97	0.05 *	−15.37	−0.02	3.88 (1, 486)	0.006
Constant	309.62	2.52	122.84	<0.001 ***	304.67	314.57
ELA Scores as the Outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Household Participation	−9.75	3.60	−2.71	0.007 **	−16.82	−2.68	7.33 (1, 501)	0.0125
Constant	318.39	2.28	139.27	<0.001 ***	313.90	322.88

* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

shows the magnitude and direction of the relationship as evidenced by the estimated regression coefficients and 95% confidence intervals.

The regression results showed a small, negative statistically significant difference between Math test scores of students from program-participating households and non-participating households. Students who were participating in the program performed about 7.7 points lower, on average, in Math compared to non-participating students. Household participation explained only about 0.6% of the variance in students’ Math achievement, with an adjusted R-squared value of 0.006. Similarly, a statistically significant difference was observed in students’ ELA scores between participating and non-participating students. Participating students were about 9.7 points lower than non-participating students, on average. However, the effect size was relatively small in that only 1.3% of the variance in ELA scores was explained by household participation. Overall, the small effect sizes suggest that household program participation is only one of numerous factors that can potentially be associated with students’ test scores.

7.2. RQ 2: How Do Parents’ Beliefs and Use of Technology Relate to Students’ Math and ELA Achievement, Controlling for Household Participation in a Free Community Broadband Program?

To explore factors that may be associated with students’ test scores in addition to household participation in the free community broadband program, scales measuring parents’ use of and beliefs and attitudes toward technology were added to a stepwise regression model. Any non-significant predictors from the model were removed and the results are presented in

Table 7. Regression results for math and ELA achievement by parent use and beliefs (controlling for household program participation).

Math Scaled Scores as the Outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Household Participation	−6.94	3.90	−1.79	0.075	−14.59	0.70	8.14 (3, 484)	0.042
Fam Chal	−7.22	1.92	−3.75	<0.001 ***	−11.00	−3.44
Dev Imp	−4.76	2.02	−2.35	0.019 **	−8.73	−0.78
Constant	309.17	2.18	124.63	<0.001 ***	304.30	314.04
ELA scaled scores as the outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Household Participation	−8.37	3.59	−2.33	0.02 **	−15.42	−1.33	9.54 (3, 499)	0.049
Fam Chal	−5.54	1.78	−3.11	0.002 **	−9.04	−2.05
Dev Imp	−6.00	1.86	−3.20	0.001 **	−9.60	−2.30
Constant	317.88	2.25	141.21	<0.001 ***	313.45	322.30

Note: Tech utility and parent use were removed (p > 0.1) from the regression model for Math, and Tech utility and parent use were removed (p > 0.1) from the regression model for ELA. ** p ≤ 0.01, *** p ≤ 0.001.

.

For Math scores, among the four factors examined, parent use and tech utility did not emerge as statistically significant. However, beliefs regarding development impact and familial challenges did show significant associations with student Math scores when controlling for other variables. Together, these factors explained 4.2% of the variance in student Math scores. Interestingly, as parent perceptions of development impact and familial challenges increased, student Math scores tended to decrease.

Household participation in the community broadband program was not a statistically significant predictor of students’ Math scores but did predict ELA scores. Following the same approach of predicting students’ Math scores, parent use and tech utility did not appear to be significant predictors, but development impact and familial challenges were significantly associated with students’ ELA scores. For every unit increase in developmental impact, students scored approximately six fewer ELA points on average, and there was a similar magnitude of change for a unit increase in familial challenges. Together, all analyzed variables explained about 5% of the variance in the outcome of student ELA scores. In summary, more positive parental beliefs about technology’s developmental impact and stronger agreement that technology brought familial challenges were both associated with students’ lower Math and ELA performance.

7.3. RQ 3: How Do Students’ Personal and Household Characteristics Relate to Their Math and ELA Achievement, Given Their Parents’ Beliefs and Use of Technology, and Their Households Participation in a Free Community Broadband Program?

Students’ racialized/ethnic identity, gender, grade level, English language learner (ELL) status, and economic disadvantage status were added to the RQ2 regression model, in addition to parent use and belief factors and household participation, to understand how students’ personal and household characteristics further associate with student achievement. A stepwise regression analysis was performed to model the relationship between these variables and students’ Math and ELA scores, removing non-significant factors. The results are presented in

Table 8. Regression results for math and ELA scores and household participation, demographic variables, and parent use and beliefs.

Math Scaled Scores as the Outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Dummy_B	−27.5	3.81	−7.21	<0.001 ***	−34.94	−19.98	23.83 (5, 482)	0.19
Econ dis	−8.31	3.67	−2.26	0.024 *	−15.51	−1.10
Grade	−5.71	1.23	−4.65	<0.001 ***	−8.13	−3.30
ELL	−20.39	6.91	−2.95	0.003 **	−34.0	−6.81
Fam Chal	−4.64	1.79	−2.59	0.01 *	−8.16	−1.12
Constant	359.42	7.93	45.34	<0.001 ***	343.85	375.0
ELA scaled scores as the outcome
Predictor	b	SE	t	p	95% CI		F (df)	Adj R2
Econ Dis	−9.93	3.34	−2.96	0.003 **	−16.52	−3.34	20.12 (8, 494)	0.23
ELL	−18.35	6.31	−2.91	0.004 **	−30.75	−5.94
Dev Imp	−3.73	1.69	−2.21	0.027 *	−7.05	−0.42
Tech Uti	4.86	1.83	2.65	0.008 **	1.26	8.45
Fam Chal	−3.20	1.63	−1.96	0.05 *	−6.40	0.002
Grade	−2.03	1.09	−1.86	0.063	−4.18	0.11
Gender	−13.13	3.15	−4.16	<0.001 ***	19.32	−6.94
Dummy_B	−30.69	3.54	−8.67	<0.001 ***	−27.64	−23.74
Constant	367.20	8.78	41.80	<0.001 ***	349.94	384.46

Note: Tech utility, parent use, developmental impact, gender, household participation, Dummy_otherBIPOC, Dummy_Hispanic were removed (p > 0.1) were removed from the regression model for Math, and Parent use, household participation, Dummy_otherBIPOC, Dummy_Hispanic were removed (p > 0.1) from the regression model for ELA. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001.

.

The VIF values for all predictor variables, including the economic disadvantage indicator and racialized/ethnic identity indicators, were around 1, below the commonly accepted threshold of 5, indicating that multicollinearity is not a major concern for the current regression models.

For the Math achievement model, tech utility, parent use, developmental impact, Enrollment, Gender, and the Hispanic and other BIPOC racialized/ethnic identity indicators were removed from the model, as their p-values were larger than 0.1 and there was insufficient evidence to suggest a meaningful association of these factors with students’ Math achievement. The negative coefficient for the Black racialized/ethnic identity indicator shows that, compared to the reference category of White, Black students in the study sample had, on average, lower Math achievement (27.5 points), holding other variables constant. This difference is statistically significant (p-value < 0.001), indicating a substantial racial disparity in Math achievement. Similarly, students who were identified as English language learners (ELLs) scored approximately 20 points lower in Math, on average, compared to non-ELL students. Economically disadvantaged students scored around eight points lower in Math, on average, then their non-economically disadvantaged peers. Notably, the disparity in Math achievement between Black and White students was even greater than the gap observed between ELL and non-ELL students, or between economically disadvantaged and not disadvantaged students.

Familial challenges remained a significant factor related to student Math performance. Students’ scores tended to be poorer when parents found more challenges related to technology at home. For the ELA model, household participation, parent use, grade, and the Hispanic and Other BIPOC racialized/ethnic identity indicators were removed. Gender, developmental impact, and tech utility did not exhibit a significant relationship with students’ Math scores but did for ELA scores.

The pattern for students labeled/categorized as economically disadvantaged, ELL, Black, and facing familial challenges were similar with ELA scores and Math scores. This suggests that when parents perceive more challenges related to technology use in their daily interactions with their children, their children tend to achieve lower scores in Math and ELA tests. Similarly, students with lower test scores were associated with parents who perceived a more positive impact of technology on their children’s cognitive development.

However, a larger disparity in ELA scores between students who were identified as Black and students who were identified as White was observed compared to other predictors. Specifically, when all other variables (economically disadvantaged, ELL, parent use and belief factors, and gender) were held constant, students who were identified as Black scored about 30 points lower than White students. This considerable achievement gap has been well-documented by the Hamilton county school district and entire state whereby students who were identified as Black typically score 20% lower than White-identified students (Tennessee Department of Education, n.d.-c) across Math and ELA outcomes.

8. Discussion

This research explored the relationship between community broadband access, parent technology use and beliefs, and student academic outcomes in a Southeastern U.S. school district during and after the COVID-19 pandemic. By applying a quantitative exploratory approach and multiple regression analyses to newly de-siloed student and household data sources, the study leveraged the unique combination of 400 parents’ surveyed technology use and beliefs, their households’ connectivity status to the free broadband program, and their students’ demographic characteristics and standardized Math and ELA scores. That said, there was little expectation from the program leaders (or the existing research literature) that broadband access or parent technology use or beliefs would appreciably change deeply embedded inequities associated with racialized/ethnic identity and socioeconomic status. Although the overall patterns of achievement for historically disenfranchised student groups remained largely unchanged by program access and surveyed parental practices, the use and combination of these diverse data sources provides some unique insights and shows great promise to support future programs and initiatives.

8.1. Program Participation

In contrast to findings on a high-speed internet access program implemented in rural Michigan [12], the current study did not find a positive relationship between broadband program participation and standardized test scores. When not controlling for other demographic variables, household participation in the broadband program was significantly associated with both Math and ELA scale scores, with lower scores for participating students compared to non-participating students. Instead of attributing these findings to negative program impacts, they likely underscore pre-existing disparities among lower-income households the program is meant to target. For example, the observed variance in household participation rates was found to be closely associated with measures of students’ racialized/ethnic identity, income, and English language learner (ELL) status. Rather than influencing students’ scores, household participation appears to reveal existing social and resource inequities long associated with academic performance.

8.2. Parents’ Beliefs and Use of Technology

Overall, the current data did not reveal any statistically significant relationship between parent technology use and student achievement outcomes. However, parent beliefs towards technology’s impacts on children’s cognitive and non-cognitive skills and on family dynamics, as well as technology’s usefulness, were associated with students’ Math and/or ELA scores, even after controlling for student characteristics, aligning with previous research that demonstrates how parental attitudes and engagement with technology influence student academic outcomes [22,23,24]. Both greater familial challenges and greater development impacts tended to be negatively related to achievement. Meanwhile, a greater sentiment toward technology’s utility was associated with increased ELA scores. While the reasons for these nuanced relationships are unclear, simply being able to link parent voices to academic outcomes in the context of a community broadband program is an important step forward in recognizing how social programs may relate to academic outcomes. Additionally, prior research in this community has found that parent use of and beliefs toward technology were often greatest among traditionally disadvantaged groups in the community, which may be a long-term mechanism for more equitable student achievement [32,33].

8.3. Racialized/Ethnic Identity and Systemic Inequality

Among the students’ personal and household characteristics explored here, racial/ethnic identity, economic disadvantage status, and English learner status emerged as robust predictors of both Math and ELA scores. Interestingly, gender displayed a significant association exclusively with ELA scores. The extent of racial disparities was stark; across the study-wide sample, Black students scored approximately 28 points lower in Math and 30 points lower in ELA compared to their White counterparts, on average, holding other demographic and parent belief measures constant. This finding partially aligned with what was documented by the state government: in 2023, students who were economically disadvantaged or identified as Black tend to score about 16% to 19% lower than their non-economically disadvantaged peers in both ELA and Math standardized tests [13]. The notable gaps in achievement based on demographic factors reveal the limits of the academic outcome measures as well as the deep-seated historical educational equity issues within the community.

9. Conclusions and Limitations

Overall, the current study underscores three key points. First, the study revealed community-wide parent sentiments about managing and guiding the use of technology for educational purposes. Parents’ engagement and family environment have been factors influencing students’ achievement. Although the observed significant association between parents’ belief factors and students’ ELA scores was inverse, it reflected parents’ dual perception of both the benefits and challenges associated with technology. While parents reported a high frequency of technology use and positive attitudes toward its educational value, they also experienced challenges related to effectively controlling their children’s technology use. It is important to note that parent survey data are self-reported and could be subject to social desirability or other biases.

The community broadband program has been a significant step in addressing existing inequalities and providing valuable resources to parents. Moving forward, communities could likely bolster the educational reach of community broadband initiatives through increased support and resources for both parents and students to access and leverage home technology for improved educational outcomes.

Policymakers are encouraged to partner with internet providers and local organizations to provide affordable devices and internet access to communities in need, paying particular attention to adapting programs that meet local infrastructure, cultural contexts, and educational priorities. Additionally, offering digital literacy training for parents enables them to support their children’s education more effectively, further solidifying this shift.

While these efforts may not transform educational achievement immediately, they have empowered these groups by providing access to resources and fostering connections with schools and teachers. Schools are now seeing new levels of digital engagement after the pandemic. With time and the continued bridging of gaps, the landscape of digital resources and student education could look entirely different.

Second, the study revealed the persistent achievement gap across racialized/ethnic student groups under the lens of a community broadband program and parental technology use and beliefs. Students categorized as Black generally scored substantially lower than students categorized as White across both Math and ELA state assessments. Given that student achievement was one of many potential long-term outcomes from this free community broadband program, the results here suggest that accessibility alone does not effectively narrow persistent social inequality. Although the current study did not find sufficient evidence suggesting the broadband program narrowed the achievement gap, without a control group it is uncertain how the achievement gap might have changed without this program. Even though this data concerned only the second full year of the broadband implementation, households’ access to free internet clearly provided many families with an important resource and mechanism for engaging with their children’s education and connecting with their teachers and schools.

In conclusion, the study design of merging de-siloed data sources offers a unique contribution to the literature by integrating multiple data sources to address community experiences and impacts. The current study sought to advance current methodological practices by integrating three sources of previously disconnected student and household data to better understand the community-wide implementation and address three research questions. Specifically, the study accessed and merged student-level data including student achievement and student demographic data culled from the school district, a random sample of surveyed households collected by the research team, and household-level program participation data supplied by the broadband provider. Although many of these data sources are found in many communities and research studies, what was unique here was that the three sources of information were integrated at both the student and household levels, providing a powerful opportunity to explore how such data interact, and seek insights to better inform the program and community. By bridging such data silos, the study hopes to encourage other researchers and communities to advance a more holistic understanding of their program’s impact and better use available data sources to more fully address implementation and explore potential outcomes across their own diverse communities.