Twitter in the time of coronavirus

25th March 2020 by Timo Hannay [link]

One way of finding out what people are saying about schools is to log-in to your favourite social media app and scroll – a lot. Reclusive data nerds that we are, we prefer to crunch the data instead. This is something we typically do at the end of each academic year (see our previous posts in 2018 and 2019), but we also do it whenever we want to take the temperature of the public conversation about schools. Our current febrile times seem like another important moment to do just that, so this post will look at recent trends on Twitter to gauge exactly what has been concerning us and when.

Going viral

Figure 1 shows daily numbers of tweets from the UK that have mentioned schools. (It uses a relative scale in which 100 is set to the mean daily activity during January; we don't know the absolute numbers because this analysis is based on a subset of all tweets.¹)

Emerging from the New Year slump in early January, there were clear weekly cycles with dips in activity at the weekends and a slightly more sustained reduction during the half-term break in late February. Then activity ramped up rapidly during March, reaching a crescendo on 18th March, when the government announced that all schools would close. It has since subsided, but as of 24th March was still running at almost double the normal level of activity. (Hover over the graph below to see corresponding data values.)

Gathering clouds

What have people been so actively tweeting about? Figure 2 shows word clouds of terms that have characterised each month so far this year (relative to the same months in 2018 and 2019, thus downplaying perennial seasonal topics).

In that distant halcyon period known as January 2020, as a new year and the reality of a strong Conservative government dawned, there was much talk of petitions: to make the teaching of sign language compulsory, to provide free sanitary products for girls, or to ban halal meat. By February, however, amid talk of Greta Thunberg and Pancake Day, coronavirus was already looming large on Twitter, if not yet in the country. Then, come March, there has really been only one topic of conversation – the words in that dark cloud speak for themselves.

(Use the menu below to view these different months.)

Topics of conversation

Figure 3 shows these temporal trends more clearly, focusing on topics related to coronavirus². In tweets about schools, there has actually been relatively little talk of the disease itself and even less about government's advice on what and what not to do. Discussion of mitigation strategies such as hand washing and social separation have been a bit more prevalent. Perhaps unsurprisingly, though, school closures has been the most common topic of conversation, peaking at around 50% of all tweets about schools. But notice that the rise in mentions of school closures preceded the government's 18th March announcement by about a week: it seems that we collectively knew what was coming. This has since subsided and in the last few days the subjects of key workers and especially home schooling have become much more prevalent.

Looking at all of these topics together provides some insight into our shifting concerns about the education of our children – a glimpse into the collective mind of a society in shock but doing its best to cope.

(Click on the figure legend to turn individual data sets on or off; double-click to show one data set on its own. Hover over the graph to see corresponding data values.)

As ever, we welcome your comments: [email protected]. Stay safe.

Footnotes:

1. Actually a subset of a subset. We used the Twitter API, which provides a sample of relevant tweets, and also limit ourselves to tweets that can be geolocated within or close to England.
2. We used a list of just over 100 terms that have been prevalent in recent tweets and seemed to be reasonably reliably associated with particular coronavirus-related topics. However, some terms are potentially ambiguous. For example, 'distance' can refer to social distancing as well as distance learning; 'homework' might concern home schooling, but can also refer to the more routine kind set by schools during normal times. Note also that whilst these different topics are shown stacked on top of each other, they are not strictly additive because individual tweets can potentially mention more than one of them.

Developments in staff development

19th March 2020 by Timo Hannay [link]

Update added 19/3/2020: Here is a blog post from our collaborators at the Teacher Development Trust to complement what you can read below.

Following on the heels of our last post about trends in teacher recruitment, this one is about staff development. The two topics are clearly connected: as we said last year, what is the point of hiring teachers if you don't motivate and develop them? Here we update our previous analysis of staff development spend at state-funded schools in England, adding 2018 data to the trends we saw for 2012 to 2017, as well as presenting some new perspectives on the data.

Once again, we are delighted to have conducted this analysis in collaboration with our friends at the Teacher Development Trust (TDT).

Our main findings are:

• Overall spending on staff development increased in 2018 compared to the previous year, though in nominal terms it barely moved above the level reached during the last peak in 2016, indicating a real-terms drop since then.
• Secondary schools spent an average of just over £520 per teacher in 2018, a substantial increase on previous years, though this still only accounts for 0.54% of total spending. 65% of secondary-school teachers (about 132,000 full-time equivalents) worked at schools that spent less than £500 per teacher in 2018.
• Primary schools spent an average of nearly £710 per teacher, or 0.66% of their total spending. This was down from a high of 0.75% in 2016. 41% of all primary teachers (about 90,300 full-time equivalents) worked in schools that spent less than £500 per teacher in 2018.
• There is considerable regional variation. In general, London, the South East and the East of England showed the highest levels of staff development spend in 2018. The highest-spending region among primary schools (the South East) spent 21% more per teacher than the lowest-spending region (the South West). Among secondary schools, London spent 26% more per teacher than the East and West Midlands.
• School trusts – especially large multi-academy trusts (MATs) – spent more per teacher than local authority-maintained schools. Among primary schools, large MATs spent 37% more than LA-maintained schools; among secondary schools the difference was 74%.

A year older and wiser?

Our previous report analysed years from 2012 to 2017; this one adds 2018. (Here, '2018' refers to the 2017-18 academic year, and similarly for other years. Full analysis of 2019 is not yet possible because financial data for academies have not been released. We expect to issue another update once that happens, probably in early autumn.)

Figure 1 shows spending patterns by year and school type. Total spending on staff development rose to £284m in 2018, an annual increase of 10.7% since 2017, but only 1.6% higher than 2016. (Note that all numbers presented here are in nominal terms, so make no allowance for inflation.) The rise in 2018 was principally due to an increase among secondary schools (blue line), with primary schools (red) and special schools (purple) remaining flat.

As a percentage of total spend, primary and special schools actually showed a slight reduction, with secondary schools displaying a substantial increase, albeit from a lower base. This had the overall effect of reducing the gap in spending rates between primary and secondary schools. On a per-teacher basis¹, primary schools spent an average of £709 in 2018 compared to £522 among secondary schools. The comparable figures per pupil were £32 and £29, respectively.

(Use the menu below to explore these different measures. Click on the figure legend to turn individual lines on or off. Hover over the graph to see corresponding data values and sample sizes.)

Averages of the kind shown above can hide wide variations between individual schools, and sometimes statistical outliers can have disproportionate effects. One way to see through this is to look at median instead of mean values. In 2018, the median primary school reduced per-teacher staff development spend by £6.40 compared to a year earlier; the median secondary school increased it by £7.40 per teacher.

Another way to think about this is presented in Figure 2, which shows the proportions of schools that displayed increases (green) or decreases (red) in staff development spend compared to the previous year. Among all schools, a slight majority (51% versus 47%) decreased spend in 2018 compared to 2017. This was mainly because 53% of primary schools showed a decrease. In contrast, a majority of secondary schools (56%) showed a year-on-year increase, bucking the trend of preceding years.

(Use the menu below to explore the different school types, including special schools. Hover over the graph to see corresponding data values and sample sizes.)

Uneven money

To illustrate further the wide range of spending patterns between schools, Figure 3 shows the distribution of staff development spending as a proportion of total spending. In 2018, the modal (most common) value among primary schools (red columns) was 0.5% while for secondary schools (blue) it was 0.3%. These patterns were broadly the same in 2017 – but note how many more secondary schools were spending little or nothing on staff development (left-hand blue columns) and how few were spending higher proportions (right-hand blue columns) in 2017 compared to 2018.

Nevertheless, even in 2018, less than 18% of primary schools and only just over 12% of secondary schools spent at least 1% of their budgets on staff development. If, for the sake of argument, we take this to be a desirable level of spending then it represents an aggregate shortfall of over £220m.

To use another arbitrary but perhaps informative threshold, in 2018, 41% of all primary teachers (about 90,300 full-time equivalents) were in schools that spent less than £500 per teacher per year on staff development; the same was true of 65% of secondary teachers (about 132,000 full-time equivalents).

(Use the menu below to explore these and other years. Click on the legend to turn different school types on or off. Hover over the graph to see corresponding data values.)

Geography lessons

So much for the national picture. What, if any, differences are there between schools in various parts of the country? Map 1 shows regional variations in staff development spend during 2018, which were substantial. Among all schools, London, the South East and the East of England showed the highest levels of spending. In particular, London, at £610 per teacher, spent 22% more than the South West. Among primary schools, the North West showed relatively high spending, but the South East had the highest level of all (£714 per teacher), 21% greater than the South West (£591 per teacher). Among secondary schools, the south-eastern regions predominated, with London (£465 per teacher) spending 26% more than the East and West Midlands (£368).

(Use the menu below to explore the different school types, including special schools. Hover over the map to see corresponding data values.)

There may be good reasons for at least some of these disparities (eg, regional differences in the cost of training or average teacher experience), but on the face of it, school staff development appears to present yet another opportunity for the country to 'level up'.

A question of trust

Finally, Figure 4 shows average staff development spend by type and size of school trust. Across all schools, local authority-maintained schools spent about as much as single-academy trusts (SATs) – just under £600 per teacher – while small multi-academy trusts (up to 10 schools) spent over £650 and large multi-academy trusts (more than 10 schools) spent over £750 per teacher.

However, this is potentially misleading because the mix of schools in each group is different – it is probably fairer to look at the trends by school phase. When we do, the differences are even starker. Among primary schools, those in large MATs spent 37% more than those that were LA-maintained (£885 versus £645). Secondary schools show a similar but even more extreme pattern, with large MATs spending 74% more per teacher than LA-maintained schools (£637 versus £366).

(Click on the figure legend to turn individual phases on or off. Hover over the map to see corresponding data values and teacher numbers.)

This is interesting in part because a recent survey of teacher autonomy conducted by the NFER (National Foundation for Educational Research) and the TDT showed the reverse pattern. Taking both sets of results together, it appears that school trusts – especially large multi-academy trusts – offer less autonomy but more professional development. Insofar as this reflects genuine differences in management philosophy, teachers will presumably vote with their feet during the current recruiting season.

As ever, we welcome your thoughts to [email protected]. In addition, users of our premium service, SchoolDash Insights, can access further detail on a wide variety of income and expenditure patterns at national, local and school levels – just go to the Finances section.

Footnotes:

1. The use of per-teacher values should not be taken to imply that all staff development spend is necessarily devoted to teachers alone, though presumbly the vast majority of it is. Note also that per-teacher and per-pupil figures use full-time equivalents (FTEs), not headcounts.

Teacher recruitment: A mid-term report

4th March 2020 by Timo Hannay [link]

Peak teacher-hunting season for schools runs roughly from January to May. Now that we're two months into the 2020 recruiting drive, this post takes a quick look at the trends so far by examining year-on-year changes by subject area. For an overview of what happened during the 2018-2019 academic year, see our previous analysis, conducted in collaboration with the Gatsby Foundation; for the very latest stats and underlying job listings, see our Jobs section.

The data used here have been gathered by us from the websites of secondary schools, sixth-form college and further education colleges in England. We index these every night looking for vacancies. Of course, that doesn't mean we find every new teaching position – some are not advertised online, while some school websites are not amenable to indexing or are occasionally unresponsive. But we find at least some positions for the vast majority of schools (well over 90%) and consider our data set comprehensive enough to faithfully reflect national trends.

Hire and higher

Figure 1 shows cumulative year-on-year changes in the number of teacher vacancies over the last 12 months. More specifically, it compares the numbers of vacancies found each week between March 2019 and February 2020 with those found in the corresponding weeks between March 2018 and February 2019. So far we have seen nearly 1,800 more vacancies in the last 12 months than we did a year earlier. (The big spike in April is an artefact caused by the different timing of Easter in 2018 versus 2019 and can be safely ignored.)

Roughly speaking, half of this year-on-year increase occurred during May and June 2019 and the other half has occurred during January and February 2020, though there were also more modest rises during the autumn of 2019.

As well as these changes in total teacher vacancies, it's also informative to look at individual subject areas. The largest increases, at least in terms of the absolute numbers shown here, were seen in Maths and Music/Drama (though note that that latter includes a large proportion of peripatetic positions). There were more modest rises in Design and Technology, Languages, Religious Education, Technology (ie, computing) and English, while Geography and History have remained broadly flat. Science is currently down year on year. These subject-specific trends are easier to see if you toggle off the line for all subjects.

(Click on any of the links above to toggle individual subjects on or off; clicking on the legend at the top of Figure 1 has the same effect and provides some additional subject areas to explore. You can also use the menu above the chart to switch between weekly and cumulative values. Hover over the graph to see corresponding data values.)

It is important to recognise that these numbers provide only a snapshot of recent changes. For example, the recent increase in language teaching positions follows previous declines, while the decrease in science positions comes off the back of earlier increases (see our previous analysis for further details).

We plan to report again more fully in April or May, as we enter the final phase of the current teacher recruiting season. In the meantime, keep an eye on our Jobs section, which provides summary stats updated every week as well as underlying jobs listings refreshed every day across all subject areas – from Art to Technology and everything in between. Users of our premium service, SchoolDash Insights, can also access maps and tables showing hiring patterns by subject, location and school.

Outliers: On geographically isolated schools

3rd March 2020 by Timo Hannay [link]

Update: Here is the BBC News coverage of this analysis.

We have written before about the impact of poverty on schools and also studied geographical trends in the education system. This post will combine those two aspects by looking at the effects of physical isolation and economic deprivation on school effectiveness, as judged by Ofsted ratings. In summary, we find that:

• Among primary schools, geographical isolation has relatively little impact on school effectiveness; poverty is a much bigger driver.
• For secondary schools, however, poverty and isolation seem to act in concert. Among the relatively small number of secondary schools with high levels of both, two-thirds are rated 'Requires Improvement' or 'Inadequate'.
• The areas in which these schools are located also showed high levels of support for Brexit and large swings towards the Conservatives in the 2019 general election, suggesting that their improvement may be a matter not only of education policy but also national politics.

Disadvantage and disconnection

How does geographical integration or isolation affect school quality? To explore this question, Figure 1 combines data from the Department for Transport, Ofsted and the Department for Education to show how Ofsted ratings vary by travel time from each school to the nearest major employment hub (essentially, a town or city)¹.

Looking first at primary schools², those with low proportions of poor pupils³ show little difference between integrated and isolated areas, though more isolated schools are slightly less likely to be 'Outstanding'.

Among primary schools with medium or high levels of deprivation, we see larger overall proportions of under-performing schools as well as slightly more pronounced trends by travel time – though the latter effects are still relatively weak. In short, primary schools with larger numbers of poor pupils tend to be judged less effective than those with more affluent pupil populations, but those in isolated areas don't perform much worse than those in more integrated areas. The main effect of geographical isolation is at the top end, where we see lower proportions of 'Outstanding' schools.

The story for secondary schools is rather different. Even at low deprivation levels, more isolated schools are substantially more likely to under-perform and less likely to be judged 'Outstanding'. At medium and high deprivation levels, the overall proportions of under-performing schools rise substantially and the effects of geographical isolation are also pronounced, creating a proverbial double whammy. Among secondary schools with high proportions of poor pupils that are are also more than 30 minutes away from the nearest major employment hub, two-thirds (19 out of 29) are rated 'Requires Improvement' (RI) or 'Inadequate'; only one – Gladesmore Community School in Tottenham – is rated 'Outstanding'.

(Use the menu above Figure 1 to explore these various parameters. Hover over the columns to see accompanying data, including numbers of schools.)

It's important to note that travel times of more than 30 minutes are unexceptional: the average across all areas of England is about 33 minutes and the average for all school locations is about 36 minutes. But when combined with high levels of poverty, even relatively modest levels of isolation seem to have substantial effects on quality of education, at least for secondary schools.

On its own, this analysis doesn't reveal the reasons for the differences between primary and secondary phases, so we can only speculate. However, it appears likely that they are connected at least in part to teacher recruitment since proximity to a major employment hub presumably makes it easier to hire suitable staff, especially the kinds of subject specialists required by secondary schools.

School atlas

To make the picture less abstract, let's see where these schools are located. Map 1 uses a Google Heatmap to show the density of schools of various types.

Looking first at schools with any proportion of poorer pupils, we can see that schools within 10 minutes travel of a major employment hub are located in major towns and cities – no surprise there. As we increase the travel time to 10-20 minutes, 20-30 minutes or more than 30 minutes, the locations spread out to fill most of the country – though if you zoom in to any major metropolitan area you can see the city centres hollowing out as the travel time increases. (To zoom, double-click on the map or use the +/- controls at the bottom-right.)

Focusing now on schools with high proportions of poorer pupils, those within 10 minutes travel of major employment centres are very heavily concentrated in a handful of major cities. But when we look at the most isolated – ie, more than 30 minutes from the nearest major employment centre – we see a very different distribution that takes in certain suburbs of major cities such as London, Birmingham, Manchester and Newcastle, as well as a scattering of coastal locations in places like Kent, East Anglia, Lancashire and the North East.

(We have looked here are primary and secondary schools together, but you can use the menus below to view them separately, and to explore other parameters.)

It is also interesting to characterise these different locations in terms of socioeconomic factors beyond education. Figure 2 shows deprivation indicators for the local areas around each group of schools⁴.

On average, the most isolated schools are not located in areas with the highest levels of overall deprivation (as measured by the government's 'Index of Multiple Deprivation') or even child deprivation. As might be expected, they also have relatively low levels of housing deprivation, environmental deprivation and crime. But they have high levels of unemployment and, of course, educational deprivation.

Note also that among isolated schools there is a particularly wide gap in unemployment levels between schools with relatively affluent pupils (blue columns) and those with poorer intakes (red columns). This is because unemployment in wealthier areas tends to fall as you move away from city and town centres, while in poorer areas in tends to rise. Interestingly, educational deprivation shows a similar pattern, which supports our speculation above that lacklustre labour markets may be affecting school performance in poorer, more isolated areas.

(Use the menu in Figure 2 to explore these and other deprivation indicators.)

The overall message is that isolated schools vary greatly in their local circumstances – it is the ones that are in poor areas that should cause concern.

That said, it would be a mistake to think of geography and demography as destiny. The results described above indicate that across England as a whole there is a correlation between poor, isolated schools and underperformance, but that doesn't mean it's true everywhere. To demonstrate this, Maps 2 and 3 show schools colour-code by travel time and Ofsted rating, respectively. They use a simple binary colouring scheme, which makes the patterns easier to see at the expense of losing some nuance: green indicates shorter travel times (<= 20 minutes) and higher Ofsted ratings ('Outstanding' or 'Good'), while red indicates longer travel times and lower Ofsted ratings. Where isolation tends to correlate with underperformance, we would expect to see similar patterns of green and red dots.

Among secondary schools with high levels of deprivation, the distributions of green and red dots in both maps are similar in certain parts of the country such as London, the south and the north-west. But in other areas, such as the midlands and north-east, they often differ, sometimes markedly. Among secondary schools with medium levels of deprivation, and among primary schools, even when they have high levels of deprivation, the correspondence is much less obvious, as we would expect. (Use the menu above Map 2 to explore these and other parameters. Both maps will change automatically. Note that if you look at primary schools and/or schools with low levels of deprivation then you'll get a lot of dots!)

Geopolitics

The impact of these trends potentially extends beyond education into the domain of national politics. To see this, consider Figure 3, which shows the political characteristics of the areas in which these groups of schools are located⁵.

In the 2016 EU referendum, more isolated areas generally showed greater support for Brexit. But notice also the trends by deprivation level: in more integrated parts of the country, Brexit support was lower around schools with more poor pupils, in more isolated areas, the reverse was true. This roughly corresponds to the common perception that Brexit was supported by two distinct groups: well-heeled people in cities and poorer people in suburban and rural areas.

The swing in Conservative support at the 2019 general election shows an even starker pattern. More isolated areas – particularly those with high proportions of poor pupils and, as we have seen, lower-rated schools – showed by far the largest shifts in support.

So this is a story about more than education. It is about parts of the country with high poverty, relatively low connectedness and worse-than-average schools that have swung behind the Tories in the hope that they will make things better. The question now is whether they will.

Footnotes:

1. Journey-time statistics for the lower super output area (LSOA) of each school come from the Department for Transport. In this context, 'major employment hub' refers to an employment centre with at least 5,000 jobs. Travel times of exactly 10, 20 or 30 minutes are included in the lower relevant category; for example, schools with a travel time of 20 minutes are included in the '10-20 minutes' category. Times quoted are for travel by foot and public transport. Travel times by car, while generally shorter, show similar overall patterns.
2. This analysis includes only mainstream state schools in England. The small number of all-through schools have been categorised under 'secondary'.
3. School deprivation categories are based on the proportions of pupils eligible for free school meals and use the following DfE bands: schools below 20% are 'low', those above 35% are 'high', and 'medium' schools are those with proportions in between.
4. Data come from the UK government's 'English indices of deprivation 2019'. Values for each school are calculated by determining the mean for all postcodes within a radius of 2km (primary schools) or 4km (secondary schools).
5. The EU referendum results were reported by local authority area, the general election results by Westminster parliamentary constituency. The numbers presented here are weighted averages that reflect the numbers of relevant schools in each electoral area.