May 4, 2021
This report reviews the coronavirus infections of adolescents between ages 16 and 18 and studies the link between coronavirus infections and distance education in secondary and vocational education. The aim is to assess whether distance education has had an impact on the number of infections among the 16-18-year-old adolescents and their family members. The study focuses on municipalities that moved to distance education at the end of November or beginning of December 2020.
Preliminary results suggest that no strong conclusions can be drawn from the regional comparison of the effects of distance learning. However, the incidence rate decreased significantly among adolescents and their family members after stricter measures were taken, including distance education. The report does not find that the number of infections among 16–18-year-olds decreased more than infections among 13–15-year-olds, who were in contact teaching during the corresponding time period. In addition, the incidence of infections among parents of 16-18-year-old adolescents has decreased in municipalities that introduced distance education at the same rate as the incidence of parents of 13-15-year-olds.
Assessing the impact of distance learning will require further research, which is why the Situation Room will continue to explore the issue during this spring.
Finland closed its primary and secondary schools for two moths during spring 2020 due to coronavirus. The objective was to prevent the virus from spreading, but it remains unclear whether the measures taken were effective. During fall 2020 several municipalities moved their secondary and vocational education to distance learning after experiencing an increase in infections. Unfortunately, no proper research data is available of these regional measures. The connection between coronavirus infections and distance learning is relevant again as all lower and upper secondary education was moved to distance learning for at least three weeks in March 2021. In the past weeks several experts and politicians have discussed the possible effects of the closure of schools in media. Surprisingly, there remains little research evidence on the subject.
This report analyses the connection between coronavirus infections and the transition to distance education using Finnish register data. The review period of the study is the end of the year 2020. Estimating the effects of the beginning of the year 2020 is more challenging, as the number of infections among primary school aged children and the rate of testing were low, and the transition to distance learning was made simultaneously everywhere. In addition, almost all primary schools have remained in contact teaching during the fall, unlike the upper secondary and vocational schools. Estimating the effects of moving to distance learning can be done more reliably for the upper secondary and vocational schools for the fall, as there has been regional and temporal variation in making the transition during this time period. For these reasons, this report focuses on studying the connection between distance education of upper secondary and vocational education and coronavirus infections.
Two methods of estimation were used to study the connection. The first method was based on using regional and temporal differences in the execution of the transition to distance learning. The second method was based on differences between different population groups in allocation of constraints. The research utilizes data from the Finnish National Infectious Diseases Register maintained by the Finnish institute for health and welfare (THL), individual level register data (i.e., FOLK) from Statistic Finland and information gathered by Helsinki GSE about municipal level contact, hybrid, and distance learning decisions.
The main point of interest is the rate of infections among students in upper secondary and vocational schools or 16-18-year-old adolescents and their family members. With the material available we can monitor the infections of teachers as well, but due to restrictions in data, we cannot make the distinction between lower and upper secondary and vocational school teachers. The number of infections among teachers has remained relatively low in the focus areas, which is why the centre of attention in this report is at analysing the infections among students in upper secondary and vocational schools or 16-18-year-olds and their family members.
Coronavirus infections of children, adolescents, and teachers
Before analysing the connection between distance learning and coronavirus infections, we will present some key figures of the infections of teachers and students. These include comparison of infections rates at different levels of education in both the Helsinki Metropolitan Area and whole Finland. In the metropolitan area the infection rate has been higher than elsewhere in Finland.
The number of infections relative to the size of the age group are presented in the figure below for children under school age (0-6-years of age), elementary aged children (7-12-year-olds), lower secondary school aged children (13-15-year-olds) and adolescents aged 16-18. The relative number of infections has been high especially among adolescents aged 16-18 during the fall (July-December).
It is important to note that the relative number of infections was low in all age groups during spring. The reasons for this might be the low level of testing of children and youth and effective distance learning. Even though researching the infections from spring is encountered with difficulties it would be important to further review the infections of this time period.
The next figure shows the infections relative to the size of each age group in the metropolitan area. In the fall the share of infections was notably larger in the age group of 16-18-year-olds compared to younger age groups. From figures 1 and 2 we can see that the share of coronavirus infections has been about twice as large among lower secondary school aged children and adolescents of age 16-18 in the metropolitan area compared to whole Finland.
The relative numbers of infections of teachers at different levels of education are presented in the figure below. The differences between groups are fairly small, though primary school, kindergarten and vocational training teachers have the highest relative infection rates. Unfortunately, with the data and material in use, we cannot distinguish between lower and upper secondary school teachers. In addition, the presentation of more detailed information is limited by the small absolute number of infections for teachers in individual municipalities and regions examined. For this reason, we focus in our report on 16-18-year-old adolescents and their family members.
Research setting: Distance learning and coronavirus infections
The impact of any single measure is difficult to estimate because several other measures are often taken simultaneously and the regional state of the epidemy varies. For this reason, a reliable estimation of single measures requires a research setting, were the set restriction only affects regionally or a limited part of the population. The effects of distance learning of upper secondary and vocational schools can be measured more reliably for fall rather than spring 2020. This is because there were differences between municipalities in moving to distance learning and there were cases were upper secondary and vocational schools moved to distance learning while lower secondary schools did not. In addition, there was more testing for infections in fall than during spring, meaning the probability of some groups being systematically left untested was lower than in spring 2020.
This report will analyse the municipalities of the metropolitan area and several other municipalities that moved their upper secondary education to distance learning in the first week of December (week 49). In the metropolitan area distance learning was implemented 3.12. prior to which (20.11.) secondary schools had introduced partial distance education or rotation in teaching, meaning only some of the students were in contact teaching at a time. Aside from the metropolitan area, many other municipalities in Uusimaa (Järvenpää, Kauniainen, Kerava, Kirkkonummi, Nurmijärvi, Mäntsälä, Pornainen, Porvoo, Tuusula) and in Finland (Jyväskylä, Kotka, Kouvola, Oulu, Salo) switched to upper secondary school distance education in the week 49 (30.11.-2.12.). This report will study these municipalities, and regional changes in infection rates in the age group of 16-18-year-old adolescents in November and December.
The majority of secondary and vocational schools in the municipalities listed above had contact teaching during the time period of October to December apart from singular cases of quarantines or classes being taught online. It is important to note that in the metropolitan area the recreational activities of children and adolescents were restricted from 30.11. onwards. That is why the connection between distance learning and infections cannot be directly evaluated in the metropolitan area. The analysis will instead be focused on the combined effect of restricting recreational activities and moving to distance learning. In addition, other restrictions may have affected the number of infections detected and complemented the impact of individual restrictions.
The figures and analyses presented in the next chapter are based on a setting, where incidences of infections are monitored in two distinct groups. The treatment group consists of municipalities where distance learning was introduced and in use during the review period. In the regional comparison the treatment group includes municipalities in the metropolitan area and the municipalities, where distance learning was introduced for upper secondary and vocational schools at the turn from November to December. We will additionally present a separate comparison including only the municipalities in the metropolitan area.
The control group includes municipalities where the upper secondary and vocational schools were in contact teaching except for single schools or campuses (i.e., Tampere) during the review period. In our research the municipalities included were Joensuu, Kangasala, Kajaani, Kokkola, Kuopio, Nokia, Pori, Rauma, Rovaniemi, Savonlinna, Sastamala, Seinäjoki, Tampere, and Ylöjärvi. The difference-in-differences setting allows making a distinct separation between the group that received the treatment and the one that did not. However, the impact of restrictions other than distance learning that were executed simultaneously cannot be controlled. An additional challenge is posed by possible changes in infection rates in the treatment and control group that took place before education was moved online. A scenario like this might take place if education is moved online because of a worsening infection situation.
In addition to regional comparison, we compare the infection situation of families in municipalities where upper secondary and vocational teaching was moved online. The families were chosen by the age of the youngest child. Families, whose youngest child was on the last grade of lower secondary or first grade of upper secondary or vocational school (or the same age) were included. This comparison is similar to the study of Vlachos et al. (2021), which studied the impacts of distance learning in Sweden. We take advantage of differences in targeting distance learning on different families with children (youngest child aged 13-15 vs. youngest child aged 16-18).
 Vlachos et al. (2021) studied the impact of distance education on coronavirus infections among parents, teachers, and teachers’ spouses of upper secondary school students in Sweden. Parents, teachers, and spouses of teachers of lower secondary school students were used as a control group. Upper secondary schools were closed in spring 2020 while lower secondary schools remained open, which allowed for comparison between the groups. Researchers have continued to further improve comparability of groups by looking only at the families of student in the last grade of lower secondary school and the first grade of upper secondary or vocational school.
A remarkable advantage of the setting based on the comparison of cohorts is the similarity of the control and treatment groups in regards of their backgrounds and infection levels prior to the treatment (figures below). In addition, any other possible restrictions affects both groups in a similar way. A potential weakness is that the restrictive measures taken may indirectly affect the treatment and control groups for example through the spreading of infections. Another concern is that there might be behavioural differences between the groups on seeking testing. In practice, the ones in contact teaching might have sought testing for the virus more easily based on early symptoms. In the light of our analysis, the problem is relatively small for December 2020, since testing for the virus had increased. In addition, the issue is probably smaller with the parents than children or adolescents themselves.
In the figures 4 and 5 are presented the incidence of infections in 16-18-year-olds in the treatment and control group on a weekly level. The y-axis measures incidence of infections for the last seven days. The incidence of infections is calculated by dividing the number of infections by population and multiplying by 100,000. For example, the infections of 16-18-year-olds in the metropolitan area on a certain day are calculated by summing the infections of the previous seven days, then dividing by the population in the group of 16-18-year-olds and finally multiplying by 100,000. If the incidence of infections on a certain day fell below five, the incidence of infections was replaced by five for data protection reasons.
 The Finnish Institute for Health and Welfare (THL) uses a corresponding calculation method to estimate the changes in coronavirus infections.
In the figure 4 the treatment group includes all municipalities in the metropolitan area and the municipalities that moved their upper secondary and vocational education online. In figure 5 are included only the municipalities in the metropolitan area. The moment of intervention (week 49) is marked by the red line on the left-hand side. On the same week recreational activities in the metropolitan area were restricted. The red line on the right-hand side indicates the time one week later, because the restriction can be expected to affect infections with a delay of approximately 5-7 days (Vlachos et al. 2021).
From figure 4 we can see that the incidence of infections (per 100,00) increased noticeably among 16-18-year-olds of the treatment group before moving to remote learning. Similar development is not observed in the control group, which includes municipalities that did not introduce distance learning. It is therefore possible that the treatment groups introduced remote learning as a response to a worsening disease situation. This makes the analysis of results more difficult in our research setting.
Based on figure 4 the incidence of infections among 16-18-year-olds seems to have decreased after the introduction of distance learning. Even though the absolute change has been large in the treatment group, we cannot directly draw the conclusion that the percentage of decrease has been greater in the treatment group than in the control group. Interpretation of the results is complicated by low levels of infections in the control group. In addition, for data protection reasons, 5 observations have been entered into the control group if the number of infections was lower than this.
 Occasional infections may also explain the variability in the incidence of infections in the control group.
It should be noted that the declining of infections may also be affected by other restrictions, in particular the simultaneous restriction of children’s recreational activities in the Helsinki Metropolitan Area. Based on this research design, we are not able to properly distinguish the potential impact of other restrictions. Regardless, after the transition to distance education, the incidence of infections can be said to have significantly decreased in the municipalities that made the transition.
Figures 6 and 7 show the incidence of coronavirus infections in family members of 16-18-year-olds in the treatment and control groups by week. The figure 6 shows the treatment group with municipalities of Helsinki Metropolitan Area and the municipalities that moved their upper secondary and vocational education online. The treatment group in figure 7 only includes municipalities in the Helsinki Metropolitan Area. Infections in family members also show a clear increase in the treatment group before the switch from contact teaching to distance education was done. There is a clear decline in the incidence of infections examined in the treatment group after switching to distance learning. No evident decrease can be observed before this. The results are similar for both cases, but changes in the incidence of infections are slightly lower when the control group includes only the municipalities of the Helsinki Metropolitan Area instead of the whole treatment group.
 Due to the low number of infections in the control group, we do not present similar figures for the parents of 16-18-year-old adolescents.
In addition to comparisons that take advantage of regional variability, we compare the incidence of infections among 13-15-year-olds (lower secondary school students) and 16-18-year-olds (upper secondary or vocational school students or adolescents their age) in municipalities that transitioned to distance learning of upper secondary and vocational schools. We can assume that the distance education of upper secondary and vocational schools had a lower impact on the incidence of infections in 13-15-year-olds than in 16-18-year-olds. The municipalities examined in the graph below include municipalities in the Helsinki Metropolitan Area and other municipalities that introduced distance learning (listen in chapter 3).
Based on our results the incidence of coronavirus infections has developed in a fairly similar manner among 13-15-year-olds and 16-18-year-olds before introducing distance learning. After the switch to distance education, infections decrease with a delay among both 16-18-year-olds and 13-15-year-olds. In addition to distance learning, other restrictions may contribute to the reduction of infections. Even though the changes are similar between the comparison groups, it cannot be concluded that distance learning could not have an effect. Rather, the comparison provides a lower bound for the potential impact of distance learning, as the restrictions may also have had an indirect effect on the comparison group. However, the results do not provide clear evidence that the transition to distance learning was the reason for reduced infections among young people.
In figure 9 we compare the incidence of corona infections in parents of 13-15-year-olds and 16-18-year-olds in municipalities that introduced distance learning in upper secondary and vocational education. Based on the results, infections have developed in a similar manner among parents of 13-15-year-olds and 16-18-year-olds before transitioning to distance education. From the graph we can see that the coronavirus infections in parents of 16-18-year-olds have significantly decreased with distance learning, but infections in parents of 13-15-year-olds have similarly decreased.
It is difficult to estimate the average change in comparison groups based on the graphs above. In the graphs below we show the incidence rates of 14 days before and after the transition to distance learning. Figure 10 presents the incidence rates for 13-15-year-olds and 16-18-year-olds and figure 11 for their parents.
The figure below shows the incidence rates of 13-15-year-olds and 16-18-year-olds before and after the transition to distance education. The figure describing the incidence before the transition to distance education has been calculated from the 14-day time period before switching to distance education. Similarly, the figure describing the incidence after the transition to distance learning was made has been calculated from the 14-day time period, starting one week after the switch was made. The incidence rate is calculated with a one-week delay, so that the transition will have time to affect the statistical infection rates.
Two things may be noted from the figure. The incidence rates of 13-15- and 16-18-year-olds are very similar before introducing distance learning and the incidence rates of both groups decline after the change. It is good to keep in mind that both groups are simultaneously affected by other restrictive measures.
Pictured in the figure below are the incidence rates of parents of 13-15- and 16-18-year-olds before and after the transition to distance education. The incidence rates are calculated in a similar manner as in the figures before. The incidence rates of parents are close to each other before the switch to distance education was made, similarly to the figures presented earlier. The incidence rates of parents of both lower and upper secondary school aged youth are lower in the latter time period. Distance education and/or other restrictions appear to have had an infection-reducing effect.
This report examined the coronavirus infections in adolescents aged 16-18 and studied the connection between distance education of upper secondary and vocational schools on coronavirus incidence. We compared the incidences of infections in 16-18-year-olds and their parents in municipalities, where distance education was introduced in the turn of November to December. Lower secondary education was kept in contact teaching during this time period. In addition, we compared the incidence rates of 13-15-year-olds (lower secondary school students) and 16-18-year-olds (upper secondary or vocational school students) in the municipalities that switched to distance learning.
Preliminary results suggest that strong conclusions cannot be made based on a regional comparison of the effects of distance education. Regardless, the incidence rates appear to have significantly decreased after distance education and other restrictions were introduced. However, we do not find that in municipalities that introduced distance learning the number of infections among 16-18-year-olds decreased more than infections among 13-15-year-olds who were not in distance education. Additionally, the incidence rate of parents declined similarly in both age groups. Distance education can have an indirect effect on the infections among 13-15-year-olds, which is why the effect of distance education cannot be directly deducted by comparing the groups.
This report focused on infections at the end of the year 2020. It is important to continue researching the effects of distance education in spring 2021 and extend the research to a longer time span as well as to a greater number of municipalities. In addition, it is essential to assess the magnitude and statistical significance of potential impacts using regression models. In the future, it is important to study the effects of distance learning more broadly on various outcomes describing well-being and learning.
Vlachos, J., E. Hertegård and H. Svaleryd (2021) School closures and SARS-CoV-2. Evidence from Sweden’s partial school closure. PNAS, forthcoming. https://doi.org/10.1101/2020.10.13.20211359.
This report was written by Mika Kortelainen (University of Turku, VATT), Jussipekka Salo (University of Helsinki), Tanja Saxell (VATT), Lauri Sääksvuori (THL, University of Turku), Antti Valkonen (Aalto University) and translated by Heidi Koponen (Aalto University)