Covid-19 has profoundly changed the way we live our lives. To contain the spread of the virus, governments across the world have imposed partial or complete lockdowns at various time points, including the closing of non-essential businesses and childcare settings, banning of public and private meetings, and mandating working from home wherever possible. Because of the pandemic, many people started to spend substantially more time at home.
In the UK, in the first lockdown starting in March 2020, everybody but keyworkers (ONS 2020d) had to work from home, non-essential shops were closed, a stay-at-home order was imposed, and schools and nurseries were shut to all children but those from keyworkers. Hence, a likely expectation would be an increased usage of energy in the home and a decrease in national demand. Indeed, national electricity demand dropped quickly when confinement measures were introduced (IEA 2020). For example, electricity demand in China dropped by 13% in February 2020, in Italy by about 28% in March–April 2020 and in the UK by 12–13% during the same period. Focusing on the UK and controlling for general trends in the consumption of electricity, an overall reduction in electricity was estimated, with reductions of up to 20% in the morning peak period, 11% in the daytime and 9% in the evening peak period in April (Anderson & James 2021).
In Europe, from January to May 2020, demand for natural gas reduced by about 8%; likely due to a mixture of factors, e.g. comparatively warm weather, lots of renewables in power generation and Covid-19 (Honoré 2020). Gas use showed a reduction of about 15% during the first two weeks of lockdown in Ireland compared with the same period in 2019; however, this was not weather corrected (SEAI 2020). For the UK, annual demand for gas in 2020 fell by 6.2% compared with 2019 (BEIS 2021a).
Hence, on national levels both electricity and gas demand were lower during the pandemic. Focusing on the residential sector, a different picture emerges: Residential electricity consumption increased by about 30% during lockdown in the US (Krarti & Aldubyan 2021). Data from Spain indicate energy increases of 15% (García et al. 2021) and 7% (Santiago et al. 2021), together with a more even distribution of consumption over the day. The UK reported an increase of 2.1% in total domestic energy consumption and of 0.8% in domestic demand for gas in 2020 compared with 2019, even though 2020 was warmer on average (BEIS 2021a). Focusing on the second quartile (April–June), i.e. the period that contained most of the first lockdown, total domestic energy consumption in the UK increased by 6.5% on a seasonally and temperature-adjusted basis (BEIS 2020c). Using per circuit monitoring revealed an increase in cooking and digital device use in Australia; however, total energy use reduced during lockdown because of lower air-conditioning use as external temperatures decreased (Snow et al. 2020).
An online survey amongst respondents in New York, US, showed that almost 50% of respondents reported higher electricity use during the pandemic, whilst about 40% reported no change (Chen et al. 2020). As per self-report, there were no longer morning or evening usage peaks on weekdays in lockdown.
Taken together, national energy demand decreased during lockdown, but residential demand increased; however, correcting for weather and seasonality is crucial. Little is known about how exactly energy use in the home changed during lockdown. It is expected that working from home will remain more widespread even after the pandemic (Hern 2020); hence, understanding the implications for energy use of an increased time spent at home are important.
The UK government started the first Covid-19 lockdown on 23 March 2020. All non-essential businesses were closed or operated in a delivery mode only; schools and nurseries were only open to children of keyworkers; and people were ordered to stay at home, but permitted to leave for essential purposes only. First easing of lockdown restrictions occurred on 11 May 2020.
Park mobility as a proxy for outdoor recreational activity decreased significantly during the initial phase of lockdown (Burdett et al. 2021), and traffic counts in April 2020 were 69% lower than in April 2019 (Jephcote et al. 2021). In April 2020, 46.6% of people in employment conducted some work at home, and the majority of those (86.0%) because of the pandemic (ONS 2020a). About one-third of those employed are key worker occupations and industries (ONS 2020c). Hence, a substantial number of employees continued to work outside the home. Also, those of working age classified as economically inactive (about 20%) or unemployed (about 4%) (ONS 2020b) would not have changed their place of work. About 18% of the population is aged 65 and above (ONS 2019). This is very close to the average exit age from the job market for both women (64.3 years) and men (65.3 years) (ONS 2021). Hence, whilst many people will have spent more time at home because of the pandemic, this is not the same across the population.
A second important contextual factor is the weather during lockdown. Figure 1 shows average daily temperature and average daily solar radiation for the location of the dwellings in this sample, with the green vertical lines indicating the lockdown period. Data were taken from the Copernicus/ECMWF ERA5 hourly re-analysis data, using the 2m_temperature_K and surface_solar_radiation_downwards variables (Copernicus Climate Change 2020).
Conditions during the first lockdown were both warmer and sunnier than on the same dates the previous year. Average temperature in lockdown was 9.4°C (SD = 2.34) compared with 8.6°C (SD = 2.29) the year before; solar radiation was 133.7 Wm–2 (SD = 50.6) in lockdown compared with 105.1 Wm–2 (SD= 43.5) the year before. Both external temperature and solar radiation can impact the energy use for space heating in a building (Kreider et al. 2010; Mihalakakou et al. 2002; Uglow 1981), so the comparatively higher ambient temperatures and higher solar radiation might have decreased the need for space heating during lockdown.
This paper explores how self-reported energy-use behaviours changed in UK homes during the first lockdown of the pandemic, from 23 March to 11 May 2020, with a survey administered after the end of the first lockdown. The survey focuses on heating-related questions and self-reported changes in the frequency and timing of appliance-use data and explores changes in other energy-related practices such as the usage of in-home displays (IHDs). All hypotheses were prespecified in advance of analysing the data (Huebner et al. 2020) to avoid selective reporting and cherry-picking of results. Data-cleaning procedures are also described in the preregistration. A reporting guideline is used for this paper to ensure that all relevant details are reported to increase the transparency of the work conducted.
This study made use of the natural experiment created by Covid-19. The SERL project1 has been recruiting participants with smart meters in their homes, and is accessing and collating their smart meter data in a secure data environment continuously since initial sign-up. As part of the pilot sample recruitment, SERL has enlisted 1711 homes in England and conducted a survey with them in September 2019 (called background survey subsequently). These participants were recontacted during the UK’s first lockdown in April–May 2020 and asked to fill in an additional survey related to their household and energy use during Covid-19 (called the ‘Covid survey’). Both postal and online surveys were used to contact participants. The initial postal survey was sent out on 22 May 2020, with a reminder mailed on 8 June; the initial email survey was sent on 27 May 2020, with reminders emailed to unfinished responders on 3 and 10 June 2020.
Surveys were sent to 1711 households, i.e. the number of households who had signed up to SERL by September 2019. They received a postal survey or an online survey (depending on household preference); those who received a postal survey based on their previously stated preferences were also given the option to answer online (called ‘push to web’). Of these households, 1084 filled in the Covid survey. Only one person from each household was instructed to fill in the survey on behalf of the household. After removing duplicates, households who were no longer living at the same address during lockdown and households who had not filled in the background survey, a final sample size of N = 1016 households remained.2 Respondents were able to skip questions, hence the number of respondents can be < 1016; the valid N for each question are stated.
Participants were located in England and Wales. The sample was biased towards areas of low deprivation, detached homes and owner-occupied dwellings. Two-person households were most common, followed by single-person households. Table 1 shows a comparison with national statistics based on the 2011 Census (ONS 2011), the English Housing Survey (EHS) 2019/20 (MHCLG 2021), and the Ordnance Survey’s (OS) Address Base (OS n.d.). For how these were derived in detail, see Webborn et al. (2021).
|VARIABLE||SAMPLE (COUNT)||SAMPLE (%)||POPULATION (%)|
|Accommodation type||Source: Census 2011|
|Tenure||Source: EHS 2019/20|
|Household size (persons)||Source: EHS 2019/20|
|Other (> 5/missing)||68||6.7||2.3b|
|Index of multiple deprivation (quintile)||Source: OS Address Base|
|1 (most deprived)||94||9.3||20.5|
|5 (least deprived)||292||28.7||18.2|
|Push to web||128||12.6||n.a.|
Both surveys are shown in full in Appendices 1a, 1b and 2 in the supplemental data online. The Covid survey asked in particular about how heating and other energy-use behaviour had changed given that residential energy use is dominated by heating in the UK. It also covered questions around household composition, financial satisfaction, work status, window-opening, the attempt to save energy and usage of IHDs, which provide easy access to smart meter data for the occupants and, in Great Britain, are offered to all homes receiving a smart meter (BEIS 2021b). The background survey was much broader, covering dwelling characteristics, appliance ownership, tenure and other variables. Some questions were identical across both surveys, allowing the direct comparison of responses pre- and during lockdown.
All data collected as part of the SERL project are accessible to all accredited UK academic researchers with access to pseudo-anonymised data via a Secure Lab environment (for details, see SERL 2021). The background survey is already available; the Covid survey will be released in the first quarter of 2022, as will the code underlying the analysis.
Note that for statistical disclosure control reasons, this paper can only show data with a frequency of at least 10 cases for each value.
The analysis was mainly focused on descriptive statistics. However, the following five hypotheses were also tested:3
Hypothesis 1. Thermostat settings have increased during lockdown: Assessed through a paired sample t-test.
Hypothesis 2. The attempt to save energy has increased during lockdown: Assessed through a Wilcoxon signed-rank test.
Hypothesis 3. Those who report using an IHD more are more likely to say they attempt to save energy: Assessed through a Spearman-rank correlation coefficient.
Hypothesis 4. Those who indicate less satisfaction with their financial situation are more likely to say they attempt to save energy: Assessed through a Spearman-rank correlation coefficient.
Hypothesis 5. Window-opening has increased during lockdown: Assessed through a Wilcoxon signed-rank test.
All hypotheses were prespecified in advance of analysing the data (Huebner et al. 2020); data-cleaning procedures were also described in the preregistration. Any deviations from the preregistration are noted in Appendix 3 in the supplemental data online. Data were analysed in R (Core R Team 2019) using the base stats package ggplot2 (Wickham 2016) and tidyverse (Wickham et al. 2019). The STROBE reporting guideline were used to ensure necessary details are reported (see Appendix 4 in the supplemental data online). Appendix 5 online contains a summary of the pre-specification, data and code availability, reporting guidelines and preprints (Huebner et al. 2021).
Risk of bias describes whether the results are affected by flaws in the design, conduct or analysis of a study. This study is observational and largely descriptive; however, it does aim at linking findings to the first Covid-19 lockdown. It is possible that in some households other events happened that might have impacted the answers, such as the birth of a child, or the death of a household member or a loss of income due to factors other than the Covid lockdown. This was not explicitly measured. However, given the low likelihood of these events—e.g. 153,464 live births occurred in England and Wales in the second quarter of 2020 across about 24.5 million households (ONS 2020f)—and the fact there is no clear evidence how those events might impact the replies, this was not considered a crucial bias. The sample was biased in its selection. However, this was beyond the control of the researchers and care was taken to limit statements about the generalisability of findings. To overcome selective reporting, the analysis and data cleaning were prespecified in advance.
The respondent of each household had to indicate for each adult in the house (Nvalid = 869) whether they always worked from home, never worked from home, sometimes worked from home, or did not work before lockdown (Nadults = 1426) and during lockdown (Nadults = 1472). The proportion of adults working from home more than tripled during lockdown, and the proportion of those never working from home halved during lockdown; however, both before and during lockdown about half of the adults in the sample did not work (Figure 2).
What Figure 2 does not reveal are within-household changes. In order to account for those, a point system was developed. Always working from home and not working were given 2 points for each adult; sometimes working from home 1 point per adult, and never working from home 0 points. For each household the sum of points was calculated for before and during lockdown, and the difference between before lockdown and during lockdown was calculated. A positive difference indicated more time at home during lockdown, a negative difference less time, and no difference the same. The largest group was no change (61.4%), followed by more time at home (30.9%); 7.7% spent less time at home.
Participants were asked for how many hours they heated their home during lockdown on a cold day (Nvalid = 842). If a household gave a non-integer answer (possible in the postal survey and decimals were allowed in the online survey), values were rounded to the nearest half-hour. If minutes were given (e.g. 9 h, 20 min; 9h20), anything < 30 min was rounded down and anything ≥ 30 min rounded up. For decimals, 0.3 was considered to indicate 30 min, as was 0.5. If anyone gave a fraction, the respective minutes were calculated and rounded to the nearest half-hour. If a respondent gave a range, e.g. 2–3 h, the midpoint was calculated and rounded if needed. If a household gave an answer > 24 h or < 0 h, the data would be flagged as an error; however, no household did this.
Mean (SD) hours were 7.46 (5.66); median hours were 6 (interquartile range (IQR) = 7). As becomes clear from the measures of dispersion and as visualised in Figure 3, responses were widely spread out.
Respondents then indicated whether this was more or fewer hours than before lockdown (Nvalid = 839). The vast majority reported no change (59.5%), with 13.6% indicating an increase and 7.6% a decrease. A total of 19.9% did not remember or indicated the question as not applicable.
However, those who spent more time at home indicated an increase in heating-hours (Figure 4).
Participants were also asked to state to what temperature they set their thermostat on a cold day during lockdown. This same question had also been asked in the background survey. Values < 5 and > 40°C were considered as outliers and excluded. Mean thermostat settings were very similar in both surveys, with mean (SD) = 20.05°C (2.13; Nvalid = 809) in lockdown and mean (SD) = 20.40°C (2.14; Nvalid = 799) pre-lockdown. Contrary to Hypothesis 1, thermostat settings were slightly lower before lockdown than during lockdown, paired sample t-test, t(710) = –5.55, p < 0.001; mean difference = –0.336 (95% confidence interval (CI) for the mean difference: –0.454 to –0.217). However, the effect was small (Cohen’s d = 0.2). When repeating the results only for those who stayed at home more, the finding was the same, i.e. thermostat settings were slightly lower during lockdown than before, t(229) = –4.33, p < 0.001; mean difference Mdiff = –0.495°C (95% CI: –0.720 to –0.270).
A third heating related question asked about the number of rooms heated during lockdown on a cold day (Nvalid = 990). The vast majority indicated no change in the number of rooms heated (86.9%), with about 3% each indicating more and fewer rooms (the remaining answers fell into categories of not applicable or did not remember). There was no difference depending on whether the respondent spent more time at home during lockdown.
Participants had to indicate how often (Table 2) or for how long (Table 3) they used certain appliances. For any answer with a range (e.g. once or twice) the midpoint was calculated. Negative numbers were coded as an error. Note that the spread of answers was very wide, and data were non-normally distributed, hence the median and IQR are used to describe the data. N = 961 respondents provided at least one answer.
|Washing machine||3 (2)|
|Cooker, oven, grill||7 (5)|
|Laptop, computer, tablet||5 (10)|
|Electric gym equipment||0 (0)|
They were then asked to indicate if this was more or less than before lockdown (Figure 5).
For all activities, doing it for the same amount or with the same frequency in lockdown compared with pre-lockdown was the most common answer, except for using laptops, computers and tablets. Here, most respondents indicated using it more often in lockdown than beforehand. The only activity that respondents indicated doing less in lockdown was the use of the tumble-dryer. Pronounced increases in lockdown were seen in television, cooking and dishwasher use.
A second block of questions, only present in the online survey version, asked about the timing of activities (Figure 6). Respondents indicated both pre-lockdown (Nvalid = 509) and during lockdown (Nvalid = 522) when they carried out the activities from the previous question.
No changes occurred in the timing of having baths, and using both the dishwasher and tumble-dryer. For cooking, about 80 additional responses were given for cooking around midday. Changes were most pronounced for watching television and using laptops, computers and tablets. These activities increased in frequency and were spread out across the day during lockdown.
Respondents were asked how often they use the IHD that shows their energy consumption (an exact valid N cannot be released for statistical disclosure control, but was between 980 and 1016). The most common answer was ‘Never’ (30.3%), followed by ‘Most days’ (22.8%) (Figure 7); ‘DK_NA’ stands for ‘don’t know’ or not applicable as answers.
Most respondents indicated that their usage of the IHD had not changed during lockdown (80.3%), with 8.8% indicating using it more often, and 10.9% less often (Nvalid = 852).
Respondents indicated how much effort they made to save energy during lockdown (Nvalid = 989). This question had also been asked in the background survey (Nvalid = 1004), and it was hypothesised that the effort to save energy would have increased in lockdown. As already clearly visible in Figure 8, this was not the case: the majority indicated having made no effort at all during lockdown, whereas pre-lockdown that answer was hardly chosen, and most respondents chose ‘some effort’ or a ‘great deal of effort’.
A Wilcoxon signed-rank test confirmed that the difference was significant, V = 288,885, p < 0.001, samples estimates: (pseudo-) median = 1.500 (95% CI 1.499–1.500), based on 980 cases. Hence, Hypothesis 2 of greater usage of the IHD during lockdown compared with pre-lockdown was not supported.
To understand if those who reported using the IHD more were more likely to say they attempted to save energy, the Spearman rank-correlation coefficient (Nvalid = 934) was calculated. Anyone who had stated ‘don’t know’ to either question or reported not having an IHD was excluded. There was a significant positive correlation between the effort to save energy and the usage of the IHD, r = 0.244, p < 0.001, supporting Hypothesis 3. However, the correlation coefficient is small.
Respondents also indicated how well they were managing financially during lockdown (for statistical disclosure control reasons, the exact valid N cannot be released, but it was between 975 and 1016). The answer categories ‘finding it very difficult’ and ‘finding it quite difficult’ were combined into ‘difficult’ for statistical disclosure reasons. Figure 9 shows that about 80% of the sample were ‘living comfortably’ or ‘doing alright’.
To understand if those who indicate doing less well financially were more likely to say they attempted to save energy, the Spearman rank-correlation coefficient (Nvalid = 948) was calculated. Anyone who had stated ‘don’t know’ to either question, did not provide an answer or chose the option of ‘prefer not to say’ for the question on financial status were excluded. There was a significant negative correlation between the effort to save energy and managing financially, r = –0.196, p < 0.001; hence, Hypothesis 4 was supported. However, the correlation coefficient was small.
In both the Covid-survey and the background survey, respondents indicated how often they opened windows on a cold and on a warm day (Figure 10).
Complete cases across both surveys were N = 938 (cold days) and N = 980 (warm days); exact valid N for the individual surveys cannot be released for statistical disclosure control reasons, but were above the numbers of complete cases across survey and < 1016. A Wilcoxon signed-rank test was used to test Hypothesis 5 that window-opening increased during lockdown. It was supported for cold days, V = 77,757, p < 0.001, sample estimates: pseudo-median = 0.4999, 95% CI = 0.00–0.500. The hypothesis was not supported for warm days.
The survey carried out with N = 1016 respondents about their energy use during the first Covid-19 lockdown showed comparatively few changes in energy-use practices. Contrary to expectations, for heating behaviour, neither an increase in heating-hours, in the number of rooms heated nor in thermostat settings was reported. Some changes in appliance use were reported, such as increased usage of cooking appliances, televisions and laptops/computers/tablets, and their usage was spread out more over the course of the day. Those who were doing less well financially reported using their IHD more often, as shown by a significant negative correlation between the two variables and a greater attempt to save energy correlated with greater usage of the IHD; overall usage of the IHD did not change. Given that national statistics implied an increase in residential energy use, these results are somewhat surprising.
One possible explanation is that although respondents spent more time at home, they did not increase their heating-hours, number of rooms heated and demand temperature. For gas (which is predominantly used for heating in the UK), only weather and seasonality-adjusted data showed an increase during the quartile that contains all but one week of the initial lockdown (BEIS 2020c). However, the quartile covered in those data (April–June) also went beyond the period of the strict lockdown that ended on 11 May, providing only limited comparability. In any case, an absolute increase in those heating behaviours is not necessary to be congruent with these data, only a lesser decrease than expected based on weather and season.
About half the sample was not working before or during lockdown, i.e. lockdown might have had a lesser impact on how much time this cohort spent at home and hence their heating behaviour. For those who stayed home more during lockdown than before, there was a likely increase in heating-hours. Hence, studying changes in subgroups will be important in understanding the implications of increased working from home.
In general, home occupants might not be particularly willing to change heating patterns. One study among 52 social housing tenants showed that the majority of participants did not change the settings of the thermostatic radiator valves during the heating season (Huebner et al. 2013).
A second explanation is that self-reported changes in space heating are not a very reliable indicator of people’s actual energy behaviour. Given that there was a time lag of at least two months between the lockdown’s start (March 2020) and the survey (May–June 2020), self-reporting might not have been very accurate, in particular for heating behaviours for which respondents were asked to think back about cold days during lockdown. If this explanation is true, it brings into question the value of self-reported data on energy use which are widely used in domestic energy research.
Data on an increased frequency and duration of the use of certain appliances (e.g. television, computing equipment and cooking) and a wider spread over the course of the day are in line with other findings on electricity use during lockdown (Chen et al. 2020; García et al. 2021; Santiago et al. 2021; Snow et al. 2020). For both heating behaviour and electrical appliances use, survey findings ought to be linked to actual energy consumption to understand their actual effect on energy used during lockdown. For all households in this survey, smart meter data pre- and during lockdown are available.
Overall IHD usage had not increased in lockdown. Snow et al. (2020) reported that respondents indicated using it even less during lockdown, given the stressful situation the pandemic posed. Usage of the IHD correlated negatively with being able to manage financially and positively with the effort to save energy. Receiving real-time feedback on energy consumption can lead to significant energy savings (Delmas et al. 2013). Indeed, the UK government has adopted smart meters and IHDs at least in part to ‘help give consumers more control over energy use and spending’ (Navigator 2012: 4), and consumers are expected to realise financial savings from smart meters and IHDs (DECC 2013). The correlations around IHDs found in this study support such aims, though it needs to be emphasised that correlations are small, do not imply causality and some findings support scepticism around the ability of current IHDs to deliver a decrease in domestic energy use (Buchanan et al. 2015).
The results around the effort to save energy are puzzling. The effort to save energy was substantially lower in lockdown compared with pre-lockdown. A new wave of recruitments for the SERL project in September 2020 showed replies to the question on the effort to save energy very similar to the baseline survey (Elam et al. 2021), i.e. both before and after lockdown the effort to save energy was comparable and much higher than during lockdown. One possible explanation is that respondents misunderstood the question as to whether lockdown caused them to make an extra effort to save energy beyond what they were already doing anyway. The second explanation is that the responses to the question on the effort to save energy in the background survey were preceded by other questions on energy-saving behaviours which might have influenced the answer on the effort to save energy. Those questions were not used in the Covid survey. However, nationally representative surveys show that overall climate change concern was high in March and June 2020 with about 80% of respondents concerned about climate change (BEIS 2020a), and over half of respondents indicated in March 2020 that they tried to minimise the amount of energy they use in the home (BEIS 2020b); hence, it seems more likely that respondents misinterpreted the question. Finally, similar to the usage of the IHD (Snow et al. 2020), it could be that participants were simply too preoccupied with the pandemic to give energy saving any particular attention.
The hypothesis that window-opening had increased during lockdown was supported for cold days, though the effect was small. The main rationale for including this hypothesis was that increased occupancy of a home will lead to the greater production of moisture and hence mandates the need for more ventilation, i.e. window-opening. Dampness leads to increased mould growth and dust mites, which might have implications for building and occupant health (Clark 2004; Koskinen et al. 1999). The results imply that occupants only increased window-opening by a very small amount, which might increase the chance of issues around damp in the homes. The longitudinal element of the SERL project will allow tracking this over time.
In general, it needs to be kept in mind that the sample consisted of participants who had opted in to a study on smart meters and the sample was biased towards those who are better off, living in non-deprived areas and in detached homes. Hence, the generalisability of the findings to other parts of the UK population is limited. Especially, financial dissatisfaction might have been more pronounced in a more representative sample (ONS 2020e). Also, as the data were solely based on retrospective self-report, it is not possible to estimate how accurate the answers were. Any future study with a similar aim, i.e. understanding changes in a crisis situation, should collect data whilst the particular crisis situation is still ongoing. This was not possible for this study given the need for survey development, the ethical approval process and the partly postal roll-out but would likely increase fidelity of results.
Self-reported energy-use showed comparatively little change in heating behaviour during the first Covid-19 lockdown in England and Wales; appliance-use data showed that the demand was spread out more over the day and also showed an increased usage of cooking, television and computers. This study also supports the notion that in-home displays (IHDs) may support energy-saving behaviour and might be of particular importance to those doing less well financially. Hence, promoting their use should be a key consideration at point of smart meter installation.
Only a subgroup of the sample studied spent more time at home during lockdown than before, as judged on their work status, e.g. because they were retired, continued to go to work or already worked from home. To understand the implications of homeworking explicitly, it will be important to focus on subgroup analysis. The findings suggest it would be oversimplified to assume that electricity and gas use have increased in all homes because of a stay-at-home order. The crucial next step is to compare actual energy consumption normalised for weather with self-reported energy behaviours, which is the next planned analysis of the SERL Covid-19 data.
GH: Conceptualisation; methodology; investigation; formal analysis; visualisation; writing—original draft; writing—review and editing; validation; software.
NW: Conceptualisation; investigation; methodology; writing—review and editing.
KD: Conceptualisation; data curation; writing—review and editing; validation; software.
EMcK: Conceptualisation; resources; writing—review and editing; software.
EW: Conceptualisation; data curation; writing—review and editing; validation; software.
FH: Formal analysis; visualisation; writing—review and editing.
SE: Conceptualisation; writing—review and editing; supervision; project administration.
TO: Conceptualisation; writing—review and editing; supervision; funding acquisition; project administration.
The authors have no competing interests to declare.
All data collected as part of the SERL project are accessible to all accredited UK academic researchers with access to pseudo-anonymised data via a Secure Lab environment (for details, see SERL 2021).
The study was approved by the departmental ethics committee at the Bartlett School of Energy, Environment and Resources (BSEER), University College London, which has the authority to approve low-risk studies such as this one.
The authors gratefully acknowledge support from UK Research and Innovation through the Centre for Research into Energy Demand Solutions (grant number EP/R035288/) and the Smart Energy Research Lab (grant number EP/P032761/1). The funders had no role in the study design; in the collection, analysis and interpretation of the data; in the writing of the report; and in the decision to submit the article for publication.
Supplemental data for this article can be accessed at:Appendix 1a
SERL Covid-19 survey (email respondents). DOI: https://doi.org/10.5334/bc.162.s1Appendix 1b
SERL Covid-19 survey (paper respondents). DOI: https://doi.org/10.5334/bc.162.s2Appendix 2
SERL Consent form. DOI: https://doi.org/10.5334/bc.162.s3Appendix 3
Deviations from the pre-specification. DOI: https://doi.org/10.5334/bc.162.s4Appendix 4
STROBE statement. Checklist of items that should be included in reports of cross-sectional studies. DOI: https://doi.org/10.5334/bc.162.s5Appendix 5
The TReQlist: a checklist for reporting of tools that promote transparency, reproducibility and quality of research. DOI: https://doi.org/10.5334/bc.162.s6
Anderson, B., & James, P. (2021). Covid-19 lockdown: Impacts on GB electricity demand and CO2 emissions. Buildings and Cities, 2(1), 134. DOI: https://doi.org/10.5334/bc.77
BEIS. (2020a). BEIS public attitudes tracker (June 2020, Wave 34, UK). Department for Business, Energy and Industrial Strategy (BEIS). https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/906452/BEIS_PAT_W34_-_Key_findings.pdf
BEIS. (2020b). BEIS public attitudes tracker (March 2020, Wave 33, UK). Department for Business, Energy and Industrial Strategy (BEIS). https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/884028/BEIS_PAT_W33_-_Key_findings_Final_.pdf
BEIS. (2020c). Energy trends—September 2020. 44(1), 1–90. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/920544/Energy_Trends_September_2020.pdf
BEIS. (2021a). Energy trends UK, October to December 2020 and 2020. Department for Business, Energy and Industrial Strategy (BEIS). https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/972741/Energy_Trends_March_2021.pdf
BEIS. (2021b). Smart meters: A guide for households—GOV.UK. Department for Business, Energy and Industrial Strategy (BEIS). https://www.gov.uk/guidance/smart-meters-how-they-work
Buchanan, K., Russo, R., & Anderson, B. (2015). The question of energy reduction: The problem(s) with feedback. Energy Policy, 77, 89–96. DOI: https://doi.org/10.1016/j.enpol.2014.12.008
Burdett, A., Davillas, A., & Etheridge, B. (2021). Weather, mental health, and mobility during the first wave of the COVID-19 pandemic. Health Economics, 30(9), 2296–2306. DOI: https://doi.org/10.1002/hec.4371
Chen, C. f., Zarazua de Rubens, G., Xu, X., & Li, J. (2020). Coronavirus comes home? Energy use, home energy management, and the social–psychological factors of COVID-19. Energy Research and Social Science, 68, 101688). DOI: https://doi.org/10.1016/j.erss.2020.101688
Clark, N. (2004). Damp indoor spaces and healthy indoor spaces news conference. https://www8.nationalacademies.org/onpinews/newsitem.aspx?RecordID=s11011
Copernicus Climate Change. (2020). ECMWF reanalysis v5 (ERA5). European Centre for Medium-Range Weather Forecasts (ECMWF). https://www.ecmwf.int/en/forecasts/dataset/ecmwf-reanalysis-v5
Core R Team. (2019). A language and environment for statistical computing. R Foundation for Statistical Computing, 2. http://www.r-project.org
DECC. (2013). Smart metering implementation programme. March. Department of Energy and Climate Change (DECC). www.nationalarchives.gov.uk/doc/open-government-licence/
Delmas, M. A., Fischlein, M., & Asensio, O. I. (2013). Information strategies and energy conservation behavior: A meta-analysis of experimental studies from 1975 to 2012. Energy Policy, 61, 729–739. DOI: https://doi.org/10.1016/j.enpol.2013.05.109
Elam, S., Webborn, E., McKenna, E. J., & Oreszczyn, T. (2021). Smart energy research lab observatory data, 2019–2020: Secure access. [Data collection]. 3rd Edition. DOI: https://doi.org/10.5255/UKDA-SN-8666-3
García, S., Parejo, A., Personal, E., Ignacio Guerrero, J., Biscarri, F., & León, C. (2021). A retrospective analysis of the impact of the COVID-19 restrictions on energy consumption at a disaggregated level. Applied Energy, 287, 116547. DOI: https://doi.org/10.1016/j.apenergy.2021.116547
Hern, A. (2020). Covid-19 could cause permanent shift towards home working. The Guardian. https://www.theguardian.com/technology/2020/mar/13/covid-19-could-cause-permanent-shift-towards-home-working
Honoré, A. (2020). Natural gas demand in Europe: The impacts of COVID-19 and other influences in 2020. https://www.oxfordenergy.org/wpcms/wp-content/uploads/2020/05/Quarterly-Gas-Review-Issue-
Huebner, G., Oreszczyn, T., McKenna, E. J., Webborn, E., Elam, S., Direk, K., & Watson, N. (2020). The impact of Covid-19 on self-reported energy use in UK homes. DOI: https://doi.org/10.17605/OSF.IO/DWPG7
Huebner, G. M., Cooper, J., & Jones, K. (2013). Domestic energy consumption—What role do comfort, habit, and knowledge about the heating system play? Energy and Buildings, 66, 626–636. DOI: https://doi.org/10.1016/j.enbuild.2013.07.043
Huebner, G. M., Fell, M. J., & Watson, N. E. (2021). Improving energy research practices: Guidance for transparency, reproducibility and quality. Buildings and Cities, 2(1), 1–20. DOI: https://doi.org/10.5334/bc.67
IEA. (2020). Covid-19 impact on electricity—Analysis—IEA. May. International Energy Agency (IEA). https://www.iea.org/reports/covid-19-impact-on-electricity
Jephcote, C., Hansell, A. L., Adams, K., & Gulliver, J. (2021). Changes in air quality during COVID-19 ‘lockdown’ in the United Kingdom. Environmental Pollution, 272, 116011. DOI: https://doi.org/10.1016/j.envpol.2020.116011
Koskinen, O. M., Husman, T. M., Meklin, T. M., & Nevalainen, A. I. (1999). The relationship between moisture or mould observations in houses and the state of health of their occupants. European Respiratory Journal, 14(6), 1363–1367. PMID: 10624768. DOI: https://doi.org/10.1183/09031936.99.14613639
Krarti, M., & Aldubyan, M. (2021). Review analysis of COVID-19 impact on electricity demand for residential buildings. Renewable and Sustainable Energy Reviews, 143, 110888. DOI: https://doi.org/10.1016/j.rser.2021.110888
MHCLG. (2021). English Housing Survey. Ministry of Housing Communities & Local Government (MHCLG). https://www.gov.uk/government/collections/english-housing-survey
Mihalakakou, G., Santamouris, M., & Tsangrassoulis, A. (2002). On the energy consumption in residential buildings. Energy and Buildings, 34(7), 727–736. DOI: https://doi.org/10.1016/S0378-7788(01)00137-2
Navigator. (2012). Smart meters: Research into public attitudes. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/48381/5424-smart-meters-research-public-attitudes.pdf
ONS. (2011). 2011 Census. Office for National Statistics (ONS). https://www.ons.gov.uk/census/2011census
ONS. (2019). Living longer: Is age 70 the new age 65? Office for National Statistics (ONS). https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/ageing/articles/livinglongerisage70thenewage65/2019-11-19
ONS. (2020a). Coronavirus and homeworking in the UK. Office for National Statistics (ONS). https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/bulletins/coronavirusandhomeworkingintheuk/april2020
ONS. (2020b). Employment in the UK. Office for National Statistics (ONS). https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/bulletins/employmentintheuk/june2020
ONS. (2020c). Key workers: Population and characteristics, 2019. Office for National Statistics (ONS). https://www.ons.gov.uk/releases/keyworkerspopulationandcharacteristics2019
ONS. (2020d). Coronavirus and key workers in the UK. Office for National Statistics (ONS). https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/articles/coronavirusandkeyworkersintheuk/2020-05-15
ONS. (2020e). Personal and economic well-being in Great Britain: May 2020. Office for National Statistics (ONS). https://www.ons.gov.uk/peoplepopulationandcommunity/wellbeing/bulletins/personalandeconomicwellbeingintheuk/may2020
ONS. (2020f). Provisional births in England and Wales. Office for National Statistics (ONS). https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/livebirths/articles/provisionalbirthsinenglandandwales/2020
ONS. (2021). Economic labour market status of individuals aged 50 and over, trends over time. Office for National Statistics (ONS). https://www.gov.uk/government/statistics/economic-labour-market-status-of-individuals-aged-50-and-over-trends-over-time-september-2021/economic-labour-market-status-of-individuals-aged-50-and-over-trends-over-time-september-2021#average-age-of-exit-from-the-labour-market
OS. (n.d.). Address base. Ordnance Survey. https://www.ordnancesurvey.co.uk/business-government/products/addressbase
Santiago, I., Moreno-Munoz, A., Quintero-Jiménez, P., Garcia-Torres, F., & Gonzalez-Redondo, M. J. (2021). Electricity demand during pandemic times: The case of the COVID-19 in Spain. Energy Policy, 148, 111964. DOI: https://doi.org/10.1016/j.enpol.2020.111964
SEAI. (2020). Tracking effect of COVID-19 on energy supply and demand. May. https://www.nratrafficdata.ie/c2/gmapbasic.asp?sgid=ZvyVmXU8jBt9PJE$c7UXt6
SERL. (2021). Accessing observatory data: Information for researchers. https://serl.ac.uk/wp-content/uploads/2021/05/Accessing-SERL-Observatory-data-info-for-researchers-v04a.pdf
Snow, S., Bean, R., Glencross, M., & Horrocks, N. (2020). Drivers behind residential electricity demand fluctuations due to COVID-19 restrictions. Energies, 13(21), 5738. DOI: https://doi.org/10.3390/en13215738
Uglow, C. E. (1981). The calculation of energy use in dwellings. Building Services Engineering Research and Technology, 2(1), 1–14. DOI: https://doi.org/10.1177/014362448100200101
Webborn, E., Few, J., McKenna, E., Elam, S., Pullinger, M., Anderson, B., Shipworth, D., & Oreszczyn, T. (2021). The SERL Observatory Dataset: Longitudinal smart meter electricity and gas data, survey, EPC and climate data for over 13,000 households in Great Britain. Energies, 14(21), 6934. DOI: https://doi.org/10.3390/EN14216934/S1. DOI: https://doi.org/10.3390/en14216934
Wickham, H. (2016). ggplot2 Elegant Graphics for Data Analysis (Use R!). Springer. DOI: https://doi.org/10.1007/978-3-319-24277-4
Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L., François, R., Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T., Miller, E., Bache, S., Müller, K., Ooms, J., Robinson, D., Seidel, D., Spinu, V., … Yutani, H. (2019). Welcome to the tidyverse. Journal of Open Source Software, 4(43). DOI: https://doi.org/10.21105/joss.01686