Socioeconomic and gender inequalities in home learning during the COVID-19 pandemic: examining the roles of the home environment, parent supervision, and educational provisions

ABSTRACT

Objective

School closures due to the COVID-19 pandemic left millions of pupils to continue their education at home. We investigated gender and socioeconomic inequalities in pupils’ home learning, and some mechanisms underlying those inequalities.

Method

We analysed online survey responses from 3,167 parents collected between May and July 2020 in the UK, when most schools were closed.

Results

Boys, pupils who were eligible for free school meals (FSM), pupils from families that were financially struggling, and pupils whose parents had not graduated from university were less engaged and spent less time home learning. Pupils of non-graduate parents found home learning challenging because they were less likely to have someone in their home who could supervise their work. Pupils eligible for FSM and from financially struggling families found home learning challenging because of noise, a lack of space, lack of technology and insufficient internet in their homes. The quality of educational resources schools provided positively predicted engagement and learning for all pupils.

Conclusion

Pupils from lower socioeconomic status families and boys were less involved with home learning, although for different reasons. We discuss how these findings can inform policy and practice to reduce educational inequalities resulting from school closures.

KEY POINTS

What is already known about this topic?

1. There are existing gender and socioeconomic inequalities in educational outcomes.

2. School closures that were implemented due to the COVID-19 pandemic are expected to exacerbate existing educational inequalities.

3. There is not yet published peer-reviewed evidence that investigates inequalities in home learning or the reasons for those inequalities.

What this topic adds:

1. Our analyses of over 3,000 parents suggests boys and pupils from families with lower socioeconomic status were less engaged and spent less time home learning, although for different reasons.

2. Lower socioeconomic status families need additional support during school closures, particularly through the provision of sufficient technology and support with supervising home learning.

3. The quality of the educational resources that schools provided were key in supporting home learning. Schools should therefore be provided with additional support to create engaging and high-quality educational resources to facilitate remote learning.

KEYWORDS:

• Educational inequality
• COVID-19
• pandemic
• socioeconomic status
• gender
• home learning

Significance

In 2020, in the wake of the coronavirus pandemic, many governments attempted to slow the rate of transmission of the virus by closing schools or allowing them to remain open only for certain pupils. As a result, in many places, pupils had to continue their learning at home, parents were suddenly expected to supervise their children’s learning, and schools were tasked with rapidly providing remote educational support. In many countries, therefore, pupils were not able to attend school but were still expected to make educational progress. There was inevitably huge variation in the ability of pupils, parents, and schools to conduct these goals effectively.

School closures therefore drastically disrupted the learning of a generation of children, although concerns have been raised that those from economically disadvantaged backgrounds will bear the brunt of the detriment (Goudeau et al., 2021). Supporting these concerns, one meta-analysis of relevant studies conducted before the pandemic suggested that the existing social class attainment gap at age 16 in England (equivalent to 18.1 months of developmental progress; Hutchinson et al., 2020) is likely to increase because of the pandemic by 36%, with estimates ranging from 11–76% (Education Endowment Foundation [EEF], 2020). Part of the reason for these increases in inequalities is likely to be that – akin to the phenomenon of “summer learning gap” (e.g., Alexander et al., 2007) – some groups of students were more engaged and spent more time learning from home during the school closures than others. However, there is not yet a clear body of evidence regarding which groups were more or less engaged in home learning or why that may be so.

In the current study, we investigate whether there were inequalities in the time spent on, and engagement with, home learning during the school closures in the UK according to pupils’ gender and different indicators of their family’s socioeconomic status. We also examine whether certain factors – including the home environment, access to technology, parent supervision, and the quality of the resources provided by schools – help to explain those inequalities. We therefore offer a systematic investigation of which groups of pupils are likely to have had their education most disrupted by the pandemic, and some indications of why that is so.

Background

It has been estimated that the pandemic will have caused a 32–37% reduction in the reading progress – and a 50–63% loss in the mathematical learning progress – that pupils in US schools would have normally been expected to have made during the 2019–2020 school year (Kuhfeld et al., 2020), with similar effects expected across the Western world (EEF, 2020). However, the detrimental effects are not uniform and have been shown to be particularly pronounced in economically disadvantaged areas and for pupils who are considered to be vulnerable or from economically disadvantaged families (Bayrakdar & Guveli, 2020; Borman, 2020; Domingue et al., 2021; Kuhfeld et al., 2020). Some have even argued that pupils from advantaged backgrounds may evidence enhanced educational outcomes as a result of increased educational support they received during the school closures (Borman, 2020). These results imply that inequalities in educational outcomes between different socioeconomic groups will increase considerably as a result of the pandemic (Goudeau et al., 2021).

The pandemic may also exacerbate existing gender inequalities in educational outcomes, in which girls outperform boys (Department for Education, 2020). Boys tend to have more controlled or extrinsic motivation to learn than girls, who tend to be more autonomous (De Blide et al., 2011; Midgley et al., 2001). Such autonomous motivation might be more important in the absence of teacher supervision and the structure of a school day, and thus one may speculate that girls have greater motivational resources to learn at home. Other studies have also shown that girls tend to report more overall motivation to learn (Clifton et al., 2008; Sheard, 2009), spend significantly more time studying (Mau & Lynn, 2001), and manifest more interest in school subjects and activities than do boys (Gentry et al., 2002; Okun et al., 1990; Verkuyten & Thijs, 2001), all of which may become more important during school closures when self-motivation is paramount to learning.

Several reports have documented how existing socioeconomic and gender inequalities in educational attainment are likely to be compounded by differences in home learning engagement during the school closures. In England, pupils from more economically disadvantaged backgrounds – who have poorer performance nationally than their peers (Department for Education, 2020) – were reportedly less engaged and spent less time home learning during the closures than their more advantaged peers (Bayrakdar & Guveli, 2020; Green, 2020; Lucas et al., 2020). Boys also spent less time on schoolwork than girls (Green, 2020). This is important because the amount of time pupils spend on their schoolwork at home is a positive predictor of their academic achievement, even outside of school closures (Clark & Hawkins, 2010; Cooper et al., 2006).

One reason for social class inequalities in home learning may be the “digital divide” (Borman, 2020; Goudeau et al., 2021), as economically disadvantaged pupils typically lack the IT – such as laptops or fast internet – essential for remote learning (Clark & Hawkins, 2010; Eyles et al., 2020). Pupils’ wider home learning environments are also likely to play a key role, as those from disadvantaged backgrounds are more likely to have had to work in noisy and/or cramped environments that were less conducive to learning (Andrew et al., 2020; Clark & Hawkins, 2010; Sammons et al., 2015; Shield & Dockrell, 2008). We investigate whether noise, lack of space, and insufficient IT and internet help to explain some of the associations between pupil characteristics and backgrounds on engagement and time spent home learning, and expect that family financial circumstances may be most strongly associated with these aspects of the home environment.

The role of parents in their child(ren)’s education is also likely to be exaggerated during school closures because parents were tasked with supervising their child(ren)’s schoolwork. Parental support and involvement in their child’s education are robust positive predictors of children’s attainment as well as the effectiveness of homeschooling (Castro et al., 2015; Desforges & Abouchaar, 2003; Guterman & Neuman, 2018). Yet, parental involvement also tends to show socioeconomic differences, with more affluent parents tending to be more engaged in their child’s education (Hill & Taylor, 2004). However, early reports of home schooling during the pandemic suggest that, while parents with higher levels of education reported being more confident home schooling their children (Cullinane & Montacute, 2020), parents in more affluent families were also more likely to continue working and to be working from home (Andrew et al., 2020), and so may have had less time to support their child’s education. Indeed, many parents juggled the demands of work and the responsibility of supervising their child’s education, which parents and children alike found stressful and emotionally trying (Thorell et al., 2021). We measure whether parents report having someone at home to supervise their child’s learning, in terms of time and ability. Whereas having the time to supervise home learning may be more of an issue for more affluent parents, those with higher levels of education are expected to be more confident and motivated to supervise home learning, and to have the relevant knowledge to do so effectively.

Many schools provided resources to help parents and pupils with home learning, although reports from parents and family members suggest the quality of the provision varied considerably across schools (Green, 2020). This is important because parental ratings of the quality of these resources have been shown to be associated positively with pupil engagement (Green, 2020). We therefore include parents’ ratings of the quality of the resources provided by schools in our models as predictors of pupil engagement and time spent home learning.

In the current study, we investigate whether there were inequalities in home learning between pupils of different genders and of different socioeconomic statuses, as well as which factors contribute to these inequalities. We expect that boys and pupils from families of lower socioeconomic status will spend less time home learning and be less engaged than girls and pupils from families of higher socioeconomic status. We also expect that features of the home learning environment, access to sufficient technology, the ability and confidence of parents to supervise their children’s work, and the quality of the educational resources that schools provided, will be associated with pupils’ gender, socioeconomic status, and home learning. We will also explore whether these factors partially explain the associations between pupils’ socioeconomic status, gender, and their engagement with and time spent home learning. We hope that our results will guide recommendations for policies offering support to schools and families to mitigate some of the negative consequences of pandemic-related school closures.

Current study

The current study analyzes over 3,000 responses from parents in the UK to an online survey conducted between May and July 2020. Schools were closed for several months from 23 March 2020 to all but those considered vulnerable or who were the children of key workers. We investigate whether pupils’ gender and different indicators of their family socioeconomic status are associated with their engagement with home learning and the time they spent learning from home. While much prior research tends to focus on a single indicator of family socioeconomic status, we include several different indicators – self-reported financial situation, eligibility for free school meals, and parental level of education – and investigate whether these are differentially associated with different home learning experiences. We then include indicators of the home learning environment, parental supervision, and the quality of schools’ educational resources as potential mechanisms to explain inequalities in pupils’ engagement and time spent learning from home during the school closures.

Methods

The results are based on responses to an online survey of parents of school-aged children in the UK about their own and their children’s experiences of home schooling. The survey ran from 1 May to 31 July 2020, a period during which UK schools were closed to most or all pupils as a result of the Covid-19 pandemic. We received institutional ethical approval for the study.

Participants

Our sample consisted of 3,167 parents (81% mothers, 16% fathers, 3% other), although the number who completed each question varied. The demographic characteristics of the sample are shown in Table 1, and the responses to each item are shown in Table 2. Respondents with more than one school-aged child were randomly asked to respond about their eldest, youngest, or one of their middle child(ren). Recruitment was mostly opportunistic via Facebook adverts, placed on relevant interest group pages (e.g., parent forums, educational forums, school and academy forums, education interest groups). We also recruited a smaller proportion of the sample from Twitter adverts. We boosted the size and diversity of our sample by recruiting 36.7% of our total sample from Academic Prolific, with a particular focus on recruiting lower socioeconomic status parents and those from ethnic minorities. However, ethnic minorities were still underrepresented in our sample (89% were White/Caucasian), and so we did not have enough statistical power to analyse inequalities by ethnicity.

Table 1. Sample details.

	n	%
Eligibility for free school meals
Not eligible	534	18.8
Eligible	2300	81.2
Gender of pupil
Boy	1459	51.9
Girl	1350	48.1
English as a first language
No	110	4.4
Yes	2387	95.6
Is child considered vulnerable
No	195	7.8
Yes	2300	92.2
Ethnicity
White/Caucasian	2503	88.6
Asian	127	4.5
African	43	1.5
Caribbean or Black	37	1.3
Arab	3	.1
Mixed	79	2.8
Other	34	1.2
Gender of parent
Male	503	17.8
Female	2315	81.8
Other	1	.0
Prefer not to say	12	.4
Immigration status
UK citizen	2553	92.4
EU citizen	143	5.2
Leave to remain	47	1.7
Visa	20	.7
Current employment status
In paid work	1929	68.2
On furlough	304	10.7
In education	27	1.0
Unemployed	77	2.7
Permanently sick or disabled	46	1.6
Retired	18	.6
In community or military service	5	.2
Doing housework or looking after children	347	12.3
Other	76	2.7
Parent a key worker
Yes	970	43.7
No	1251	56.3
At least one parent has a bachelor degree
No	708	27.0
Yes	1919	73.0
Household income
Struggling	344	12.3
Average	838	29.9
Comfortable	1621	57.8

Table 2. Response rates.

	n	%	M	SD
Time spent home learning
None	97	3.3
0–30 minutes	229	7.7
30 minutes–1 hour	350	11.8
1–2 hours	617	20.9
2–3 hours	669	22.6
3–4 hours	516	17.5
4–5 hours	292	9.9
5 hours or more	186	6.3
Pupils’ engagement and motivation	2898		3.07	1.22
Parental confidence to supervise home learning	2852		4.23	0.89
Parental time to supervise home learning	2851		3.65	1.35
Quality of home learning materials
Poor	200	6.7
Inadequate	303	10.1
Average	724	24.1
Good	1054	35.1
Excellent	664	22.1
Don’t know	62	2.1
Did noise make it harder for child to complete homework
No	1526	53
Yes, a little	959	33.3
Yes, quite a lot	261	9.1
Yes, definitely	135	4.7
Did lack of space make it harder for child to complete homework
No	2115	73.6
Yes, a little	518	18
Yes, quite a lot	150	5.2
Yes, definitely	89	3.1
Did lack of computers or tablets make it harder for child to complete homework
No	2104	73
Yes, a little	476	16.5
Yes, quite a lot	165	5.7
Yes, definitely	136	4.7
Did lack of internet make it harder for child to complete homework
No	2226	77.3
Yes, a little	469	16.3
Yes, quite a lot	114	4
Yes, definitely	69	2.4

Measures

We included the following measures for parents within a larger survey that also included a section for teachers. In this paper, we focus exclusively on the parent responses. The teacher section included items asking about the educational provisions and pastoral support their school offered and which groups of pupils were likely to be most disadvantaged by the closures. Analyses of these will be reported elsewhere.

We adopted a pragmatic approach to measurement, which prioritized the face-validity of the measures and our ability to recruit a large sample size within the limited and uncertain time frame of the initial school closures. This approach informed our decisions regarding the development of instruments and led us to prioritize single-item, face-valid measures over lengthier ones, and to develop new items that were worded with reference to the specific situation rather than use existing ones that were not orientated towards the pandemic school closures. This is in line with evidence that single item measures may constitute psychometrically reliable alternatives for long scales and thus serve educational research purposes well (Gogol et al., 2014). Unless otherwise noted, the following items were therefore designed for this survey. Descriptive statistics and correlations are shown in Table S1 in the supplementary online material (SOM).

Time spent home learning

Respondents read the following text “Please now answer these questions about the school work your child is doing at home”, before answering the question: “How much time did your child spend on their homework during an average day in the last two weeks?” with response options: None, 0–30 minutes, 30 minutes–1 hour, 1–2 hours, 2–3 hours, 3–4 hours, 4–5 hours, 5 hours or more. Responses to this variable were normally distributed (see SOM).

Pupils’ engagement and motivation

We asked parents: “How engaged with home learning is your child?” and “How motivated to home learn is your child?” Responses were given on a 5-point scale (Not at all engaged/motivated to Very engaged/motivated). Although engagement and motivation are independent constructs, the items were highly correlated, r = .83, and so were averaged for analyses.

Gender

Gender of the child was measured by parent report (Boy (51.6%), Girl (47.5%), Other (0.1%), Prefer not to say (0.7%)). To maintain sufficient sample and cell sizes, we focus on those responding boy or girl in our main analyses.

Household income

We adapted an item from the European Social Survey (2021) to ask parents whether or not their household had enough income to get by on. Responses were treated as three categories: struggling, average and comfortable.

Eligibility for free school meals

Parents reported whether their children were eligible to receive free school meals, an often-used indicator of socioeconomic disadvantage (Gorard, 2012).

Parental education

Respondents indicated their highest level of education and that of any second adult that lived in their home. These responses were then divided into two groups: graduate (households in which at least one parent had obtained a bachelor’s degree or higher) and non-graduate (neither parent had a university degree).

Parental supervision

Respondents stated how much they agreed that there was someone in their home who had the 1. Confidence, 2. Time, 3. Knowledge and 4. Motivation to help with their children’s home learning. Responses were given on a 5-point scale (strongly disagree to strongly agree). The reliability of the scale improved if the item for Time was removed, which also had weaker correlations with the other items (see Table S2). Accordingly, we combined the items for knowledge, motivation and confidence into a single measure (α = .82) of ability to supervise while also retaining time to supervise as a single separate item that we expected to be an important factor for home learning.

Home environment

Respondents rated whether noise, lack of space, lack of computers or tablets, or insufficient internet made it harder for their children to complete their schoolwork from home. Responses were No; Yes, a little; Yes, quite a lot; and Yes, definitely.

Quality of resources

Parents rated the quality of the home learning resources their school provided in the two weeks before completing the survey. Responses were on a 5-point scale (poor, inadequate, average, good, excellent) with an additional don’t know option. The don’t know responses were excluded from analyses (2.1%).

Results

We employed MPlus version 8 (Múthen & Múthen, 1998–2017) using robust maximum likelihood estimation, which is robust against violations of the assumptions of regression analyses. In the models testing our main predictions, the pupil characteristics (gender and family SES) were specified as covarying exogenous predictors of the learning environment variables (family members’ time and ability to supervise, noise, lack of space, lack of technology, lack of internet) except quality of the resources provided by schools, as there was no rationale for expecting pupil characteristics to be associated with the quality of the resources provided by the schools. We also specified direct effects from the pupil characteristics and from the learning environment variables to pupil engagement. Engagement with home learning, in turn, predicted the time pupils spent home learning. We also estimated all indirect effects from the pupil characteristics to engagement and to time spent home learning. The model is illustrated in Figure 1, full results for the direct effects are shown in Table 3, the indirect effects in Table 4, and the significant paths are shown in Figure 2.

Table 3. Direct effects for the model.

			95% CIs
Regression paths	Coefficient	SE	Lower	Upper	p-value
Time to supervise on …
Girl	0.025	0.051	−0.074	0.125	.619
Financial situation	0.062	0.037	−0.011	0.135	.094
Parents’ education	−0.354	0.059	−0.470	−0.239	<.001
nonFSM	−0.237	0.063	−0.360	−0.113	<.001
Ability to supervise on …
Girl	0.018	0.033	−0.046	0.082	.584
Financial situation	0.162	0.026	0.111	0.213	<.001
Parents’ education	0.349	0.045	0.262	0.437	<.001
nonFSM	0.100	0.046	0.009	0.191	.032
Noise on …
Girl	−0.045	0.031	−0.106	0.015	.144
Financial situation	−0.101	0.025	−0.150	−0.051	<.001
Parents’ education	−0.078	0.039	−0.155	<.001	.046
nonFSM	−0.244	0.046	−0.333	−0.155	<.001
Space on …
Girl	−0.023	0.027	−0.076	0.030	.392
Financial situation	−0.190	0.024	−0.237	−0.143	<.001
Parents’ education	0.045	0.036	−0.025	0.115	.212
nonFSM	−0.184	0.042	−0.266	−0.102	<.001
Technology on …
Girl	−0.063	0.030	−0.122	−0.005	.035
Financial situation	−0.193	0.025	−0.242	−0.143	<.001
Parents’ education	0.003	40.000	−0.075	0.080	.944
nonFSM	−0.157	0.044	−0.244	−0.070	<.001
Internet on …
Girl	−0.012	0.025	−0.061	0.037	.063
Financial situation	−0.121	0.021	−0.161	−0.080	<.001
Parents’ education	−0.017	0.032	−0.081	0.046	.596
nonFSM	−0.090	0.036	−0.161	−0.020	.012
Engagement on …
Girl	0.347	0.040	0.270	0.429	<.001
Financial situation	0.089	0.031	0.027	0.150	.005
Parents’ education	0.087	0.052	−0.014	0.188	.091
nonFSM	0.057	0.054	−0.049	0.163	.029
Time to supervise	0.135	0.018	0.100	0.170	<.001
Ability to supervise	0.288	0.028	0.234	0.342	<.001
Noise	−0.098	0.027	−0.152	−0.044	<.001
Space	−0.018	0.034	−0.085	0.049	.605
Technology	−0.055	0.030	−0.113	0.004	.068
Internet	0.132	0.032	0.069	0.196	<.001
Resources	0.274	0.019	0.236	0.312	<.001
Time spent home learning on …
Engagement	0.678	0.027	0.625	0.730	<.001
Girl	0.017	0.056	−0.092	0.126	.757
Financial situation	−0.031	0.043	−0.115	0.053	.474
Parents’ education	0.272	0.068	−0.139	0.404	<.001
nonFSM	0.308	0.077	0.157	0.459	<.001
Time to supervise	−0.050	0.024	−0.097	−0.004	.033
Ability to supervise	0.067	0.038	−0.007	0.141	.077
Noise	−0.040	0.035	−0.109	0.029	.252
Space	−0.032	0.045	−0.119	0.056	.477
Technology	−0.214	0.040	−0.292	−0.135	<.001
Internet	0.298	0.047	0.207	0.390	<.001
Resources	0.101	0.027	0.048	0.153	<.001

Table 4. Indirect effects from the model.

			95% CIs
Indirect paths	Estimate	SE	Lower	Upper	p-value
Girl to Time Spent Home Learning via …
Time to supervise	−0.001	0.003	−0.006	0.004	.627
Ability to supervise	0.001	0.002	−0.003	0.006	.609
Noise	0.002	0.002	−0.002	0.006	.364
Space	0.001	0.001	−0.002	0.003	.591
Technology	0.014	0.007	<.001	0.027	.050
Internet	−0.004	0.007	−0.018	0.011	.627
Engagement	0.237	0.029	0.180	0.294	<.001
Time to supervise and engagement	0.002	0.005	−0.007	0.011	.620
Ability to supervise and engagement	0.003	0.006	−0.009	0.016	.586
Noise and engagement	0.003	0.002	−0.001	0.007	.182
Space and engagement	>.001	0.001	−0.001	0.001	.646
Technology and engagement	0.002	0.002	−0.001	0.006	.171
Internet and engagement	−0.001	0.002	−0.005	0.003	.629
Financial situation to Time Spent Home Learning via …
Time to supervise	−0.003	0.002	−0.008	0.002	.187
Ability to supervise	0.011	0.006	−0.002	0.024	.095
Noise	0.004	0.004	−0.003	0.011	.271
Space	0.006	0.009	−0.011	0.023	.479
Technology	0.041	0.009	0.023	0.059	<.001
Internet	−0.036	0.008	−0.052	−0.020	<.001
Engagement	0.060	0.022	0.018	0.102	.005
Time to supervise and engagement	0.006	0.004	−0.001	0.013	.105
Ability to supervise and engagement	0.032	0.006	0.020	0.043	<.001
Noise and engagement	0.007	0.003	0.002	0.012	.008
Space and engagement	0.002	0.004	−0.006	0.011	.605
Technology and engagement	0.007	0.004	−0.001	0.015	.081
Internet and engagement	−0.011	0.003	−0.017	−0.005	.001
Parents’ education to Time Spent Home Learning via …
Time to supervise	0.018	0.009	0.001	0.035	.043
Ability to supervise	0.023	0.013	−0.003	0.050	.083
Noise	0.003	0.003	−0.003	0.009	.322
Space	−0.001	0.002	−0.006	0.003	.546
Technology	−0.001	0.008	−0.017	0.016	.944
Internet	−0.005	0.010	−0.024	0.014	.597
Engagement	0.059	0.035	−0.009	0.127	.091
Time to supervise and engagement	−0.032	0.007	−0.046	−0.019	<.001
Ability to supervise and engagement	0.068	0.011	0.046	0.090	<.001
Noise and engagement	0.005	0.003	−0.001	0.011	.087
Space and engagement	−0.001	0.001	−0.003	0.002	.633
Technology and engagement	<.001	0.001	−0.003	0.003	.944
Internet and engagement	−0.002	0.003	−0.007	0.004	.598
Not eligible for FSM to Time Spent Home Learning via …
Time to supervise	0.012	0.006	−0.001	0.024	.061
Ability to supervise	0.007	0.005	−0.003	0.016	.170
Noise	0.010	0.009	−0.007	0.027	.264
Space	0.006	0.008	−0.010	0.022	.479
Technology	0.034	0.011	0.011	0.056	.003
Internet	−0.027	0.012	−0.050	−0.004	.021
Engagement	0.039	0.037	−0.033	0.110	.029
Time to supervise and engagement	−0.022	0.007	−0.034	−0.009	.001
Ability to supervise and engagement	0.019	0.009	0.001	0.038	.037
Noise and engagement	0.016	0.005	0.005	0.027	.003
Space and engagement	0.002	0.004	−0.006	0.011	.606
Technology and engagement	0.006	0.004	−0.001	0.013	.105
Internet and engagement	−0.008	0.004	−0.016	−0.001	.035
Resources to Time Spent Home Learning via …
Engagement	0.186	0.015	0.157	0.215	<.001

Figure 1. The theoretical model specified for both primary and secondary school pupils.

Note: Covariances between pupil characteristics, the direct effects from the pupil characteristics to engagement, and all direct effects leading to time spent home learning, were also included but are not shown here for clarity of presentation.

Figure 2. Significant paths for the model specified.

Note: The different pupil characteristics have different type of arrows.

Gender

Being a girl was not directly associated with time spent home learning but was associated directly and positively with engagement; girls were more engaged than boys. Being a girl was also associated weakly and negatively with lack of technology, suggesting that boys were more likely than girls to struggle with home learning because of insufficient technology. Being a girl was associated indirectly and positively with time spent home learning via increased engagement.

Financial situation

Being from a family that reported a more comfortable financial situation was not directly associated with time spent home learning but was associated directly and positively with engagement; pupils from more financially comfortable families were more engaged in home learning. It was also associated negatively with a lack of internet, a lack of technology, a lack of space, and a noisy learning environment, suggesting that those with greater financial security had better access to conducive home learning environments and the necessary technology. Financial situation was also associated positively with having a household member who had the ability to supervise home learning. It was also associated indirectly and positively with time spent home learning via sufficient technology and via engagement, and indirectly and negatively with time spent home learning via sufficient internet, suggesting financial comfort was related to less home learning because of greater access to the internet. There were also positive and significant indirect paths from financial situation to time spent home learning via ability to supervise and engagement, and from financial situation to time spent home learning via lack of noise and engagement.

Parents’ education level

Having at least one parent with a bachelor’s degree was associated positively and directly with time spent home learning but was not directly related to engagement. It was associated positively and directly with ability to supervise, and negatively and directly with time to supervise and (more weakly) with noise, suggesting graduate parents had quieter homes and felt more able but had less time to supervise home learning. It was associated indirectly and positively with time spent home learning via time to supervise but associated indirectly and negatively via time to supervise and engagement. It was associated positively and indirectly with time spent home learning via ability to supervise and engagement.

Eligibility for free school meals

Not being eligible for FSM was associated positively and directly with time spent home learning but not with engagement. It was associated negatively with lack of internet, lack of technology, lack of space, and noise, suggesting pupils eligible for FSM were more likely to have noisy and cramped home environments and lack sufficient technology and internet. Not being eligible for FSM was positively associated with ability to supervise but negatively with time to supervise, suggesting parents of pupils eligible for FSM had the time but lacked the ability to supervise home learning. Not being eligible for FSM was indirectly and positively related to time spent home learning via sufficient technology, via ability to supervise and engagement, and via lack of noise and engagement. It was associated indirectly and negatively with time spent home learning via sufficient internet, via time to supervise and engagement, and via sufficient internet and engagement.

Home environment, supervision, and resources

Engagement, insufficient internet, and the quality of resources were associated positively and directly with time spent home learning, whereas lack of technology and (weakly) time to supervise were associated negatively and directly with time spent home learning. Time to supervise, ability to supervise, lack of internet, and quality of resources were also associated positively and directly with engagement, whereas noise was associated negatively and directly with engagement. Finally, quality of resources was indirectly and positively associated with time spent home learning via engagement.

Discussion

Our analyses of data collected from a survey of 3,167 parents revealed that there were inequalities in the home learning experience following the school closures that were implemented by the UK Government in response to the COVID-19 pandemic. Overall, our results showed that boys and pupils from lower socioeconomic status families were less engaged and spent less time home learning, although for different reasons.

Our results showed that girls spent more time home learning than boys, and that this was primarily because girls were more engaged and motivated to learn from home. Boys tend to have more controlled versus autonomous motivation to learn than girls (De Blide et al., 2011; Midgley et al., 2001), which, in the absence of the imposed structure of the school day and close supervision by teachers, may have resulted in boys being less motivated than girls to learn at home. Our findings are somewhat in line with other studies that show that girls are generally more motivated to learn (Clifton et al., 2008; Sheard, 2009), spend more time studying (Mau & Lynn, 2001), and are more interested in school subjects and activities than boys (Gentry et al., 2002; Okun et al., 1990; Verkuyten & Thijs, 2001). Our results show that these gender differences are also pronounced when analysing pupils’ involvement in home learning during school closures and imply that that this may exacerbate existing gender inequalities in educational outcomes.

We also found that those from lower SES families tended to be less engaged and spend less time home learning, but that different indicators of SES were related to engagement and time spent home learning in various ways and through different mechanisms. A family’s financial situation was most clearly related to the home environment, with parents from financially struggling families or whose child was eligible for FSM – an indicator of severe economic disadvantage – being more likely to report that a lack of internet, insufficient technology, a lack of space, and a noisy working environment, made home learning more challenging for their child. Self-reported financial situation and eligibility for FSM were also both associated with parents’ self-reported ability to supervise their child’s work, even while accounting for parents’ level of education.

Parents’ level of education was most clearly associated with the variables related to supervision: children with at least one parent who had a bachelor’s degree felt better able to supervise their child’s home learning, being more likely to report that someone in the home was confident, motivated and/or knowledgeable enough to supervise home learning. This, in turn, was associated with greater pupil engagement. This is in line with previous research that found a higher level of education among parents is associated with their tendency to take a more active role in the education of their children (Hill & Taylor, 2004), and that parental involvement is an important factor associated with increased educational attainment (Guterman & Neuman, 2018; Park & Holloway, 2017; Wilder, 2014), particularly among those of low SES (Domina, 2005). These findings show the responsibility that learning from home places on parents, as well as the importance of differentiating different indicators of socioeconomic status when investigating home learning and educational outcomes more generally.

However, graduate parents and parents of pupils not eligible for FSM were more likely to report that they struggled to find the time to supervise their child’s home learning, which in turn was positively related to engagement.¹ This may be because those parents were more likely to be able to continue their work from home, and less likely to be on furlough (Andrew et al., 2020). Thus, they were more likely to be working full-time in paid employment while also being expected to supervise their children’s work. These results also suggest a striking irony for parents. On the one hand, parents from lower socioeconomic families are more likely to have the time to supervise home learning, but, on the other, are less likely to have the confidence, knowledge, and motivation to do so. This conflict is likely to be emotionally challenging and may partly explain parents’ negative experiences of home learning (Thorell et al., 2021).

One unexpected but interesting finding is that a lack of internet was associated with greater engagement and motivation to home learn and more time spent home learning. Although speculative, this may be because the internet is a great source of distraction for pupils and takes their attention away from their schoolwork (bearing in mind that our results are based on parents’ reports). Importantly, we also found that the quality of the educational resources that schools provided to support home learning – such as live or recorded video lessons, worksheets, and/or feedback – was positively related to the time pupils spent home learning and to their engagement. There was also a strong indirect effect from the quality of resources to time spent home learning via pupil engagement and motivation. Schools therefore played an important supportive role by providing quality resources to aid home learning.

Limitations

We recruited respondents primarily through social media advertising. The opportunistic sample may therefore overrepresent parents who were sufficiently engaged or interested in, or opinionated about, home learning to voluntarily complete an online survey without remuneration. We also recruited a sample of paid participants using Academic Prolific, who presumably will have had different motivations for completing the survey, although this represents only a minority of our overall sample. We must therefore bear this limitation in mind when interpreting the results, as analyses of responses from other samples may have shown different results reflecting alternative experiences of home learning.

We also had to create many of our measures because the novelty of the topic meant that few relevant measures existed. To keep what was a rather detailed survey to a manageable length, we often opted for single-item measures. Although our analyses were robust against violations of the assumptions of linear models, this pragmatic approach to measurement means that many of the measures have not been psychometrically validated. These measurement limitations should be considered when interpreting the findings, although the interpretability of the results and the consilience with previous evidence gives us reason to believe our results are reliable.

We did not exploit a full array of variables that could relate to pupils’ engagement and time spent learning. For example, we did not control for pupils’ actual talent or level of their academic skills. The COVID-19 pandemic presents unique challenges for pupils because of disruptions in school, and a frequently documented challenge is the lack of sufficient academic skills necessary for self-study (Scott et al., 2020). It is noteworthy that – even after accounting for pathways via the home environment, parental supervision, and engagement variables – there were direct effects of eligibility for FSM and of parental education level on time spent home learning, suggesting that there are mechanisms beyond those we measured that account for inequalities in home learning.

Due to the relatively small samples and cell sizes of pupils of certain demographics, our study did not enable us to estimate the effects of homeschooling on specific vulnerable groups, such as children with learning difficulties, migrants, those of certain ethnicities, and those with English as an additional language. Nor could we investigate effects among children who were identified as being neither a boy nor a girl. As such, our findings cannot confidently be generalized to all groups. Our results are also based on parents’ reports and do not therefore take account of teachers’ or pupils’ views. We also surveyed parents only during the first English lockdown, between May and July 2020, and so our results represent a snapshot of the initial home learning experience. The experience in subsequent lockdowns may have been different, as schools, parents, and pupils became more familiar with the learning format. Nevertheless, the early experiences of home learning that our results capture offer some insight into inequalities in home learning at the start of the pandemic and some plausible explanations for these inequalities. The results should be used to help identify which groups need additional support to catch up with their peers in the aftermath of the pandemic.

Recommendations

Where they were in place at the start of the pandemic – and often they were not – policies and practices to guide the responses to school closures and to mitigate their potential negative effects were inconsistent and often inadequate. Based on our results, we offer the following recommendations regarding policies and support that could be developed to mitigate some of the negative consequences of this and future pandemic-related school closures.

Lower SES families struggled with home learning in part because their children were attempting to learn in noisy and cramped spaces and with insufficient technology. These findings cohere with other work that has shown that around a million children and families in England lack sufficient technology (Horrocks, 2020), and that up to a quarter of teachers in the most deprived schools report that at least a fifth of their pupils have insufficient technology for remote learning (Cullinane & Montacute, 2020; Teach First, 2020). Although many schemes to provide laptops to pupils who needed them were put in place during the school closures, many of these were informal or did not meet the demand. To counteract these detrimental effects on engagement and home learning, practical support should be made available rapidly to those who lack the appropriate environments and technology to participate fully in remote learning (EEF, 2020; Horrocks, 2020). More efficient and targeted schemes should be prepared so they can be immediately and effectively implemented when required. It may also help to set up learning bubbles, in which several families with children of similar ages can learn together, share resources and technologies, and get targeted guidance from teachers. Schools could offer sanitized and isolated rooms to these bubbles in cases where home environments are not conducive to learning.

Learning bubbles could also help to counter the implications of another of our findings; that many parents did not feel able to provide adequate supervision of their child(ren)’s learning. Family learning bubbles, in which several families are able to remote learn together, could be created to allow parents to offer and receive support from each other and to share supervision duties and resources between families. Parents will need support and training from teachers about how best to supervise their child(ren)’s work. Teachers and tutors could offer group-based assistance to bubbles rather than individual families, making the support they offer more efficient.

The quality of the educational resources that schools provided were key to supporting children’s home learning. Many schools, however, struggled to rapidly adjust to remote learning, meaning that – initially at least – many pupils did not have high-quality educational resources at home. In the future, schools must be provided with rapid guidance and support to enable them to quickly adjust to remote learning and provide high-quality educational resources. Teachers need training so that they can create the most effective digital educational resources, navigate remote learning systems, and understand how to engage pupils from different backgrounds remotely. Schools require guidance and support regarding remote learning practices, platforms, and safeguarding. Clear and realistic baseline standards for remote learning provision should be put in place, with appropriate technological support, allowing schools to maintain the necessary levels of autonomy to deliver a creative curriculum and work responsively to adapt to the specific needs of their students.

We hope the work and recommendations we have presented here help to inform planning and to mitigate some of the worst effects of this – and future – pandemics on educational inequalities.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The authors have made the data available on a public repository which can be accessed at https://osf.io/9vztu/?view_only=4afc9dc586d441f99da72ae4523d4b1e

Supplementary material

Supplemental data for this article can be accessed at https://doi.org/10.1080/20590776.2021.2014281.

Additional information

Funding

This research was supported by awards from the European Association of Social Psychology, the University of Sussex’s Higher Education Innovation Fund, and the Economic and Social Research Council’s Impact Acceleration Account Fast Track Engagement Fund [ES/T502042/1].

Notes

1. We interpret the negative effect of time to supervise on time spent home learning as representing the effect of time to supervise only once the indirect effect of time spent supervising on time spent home learning via engagement was accounted for.