Radical Statistics, Issue 127

Cover image, empty auditoriumContents of this Issue

This issue is in two parts: Part I contains papers that were ‘carried over’ from the COVID issue 126, simply because we couldn’t exceed 88 pages even with a reduced font size without having to move to a spined binding which would have been more expensive. Part II contains a diverse set of papers:

  • Dave Byrne, with a critique of the IFS Deaton Review showing how a combination of policy moves (e.g. abolition of schedule A, progressive reduction of tax rates and ability for couples to separate their tax returns) has led to widening inequalities in income and wealth;
  • Danny Dorling, prompted by John Bibby, on gender differences in mortality suggesting that – with reference to Marc Luy’s writing on how much of the sex difference in mortality could be attributed to gender – the gap in life expectancies will narrow substantially by 2050;
  • Anna Powell-Smith, displaying the types of information that are
    missing from the UK government’s extensive compilations.

Prospects for RSN 128

Following on from the discussions at the Conference in February, I asked contributors if they would agree to a student converting their power-point presentations into short texts and two or three speakers have complied. There was also one article for this issue which we collectively decided could do with revision and we have returned to the author who has agreed. So, the issue might be short but not empty!

Another proposal for generating material is the recent publication of the third RadStats compendium, Data in Society, which will be presented by the books’ editors on Saturday 28th. It is a landmark publication, bringing together many of the crucial issues around the production and use of quantitative information.

The contributors to Data in Society summarise many of the concerns around the accessibility and use of statistics in contemporary society. Examples include the lack of data from banking and financial organisations hides the extent of tax evasion of taxation. Government agencies are reducing the number of data series they make available for public scrutiny. The number of healthcare treatments in Britain provided by
private groups is growing steadily.

The book is an eye-opener on the difficulties in holding governments and large organisations to account. Do you agree with the authors’ interpretations? As the editors acknowledge there are data topics the volume does not cover in detail. These include the use of statistics by legal practitioners, housing and homelessness data and climate change data.

The editors of the RadStats journal are planning to devote one journal issue to topics raised by Data in Society, and to topics not discussed in the book. Could you write an article for the journal on any of the topics above? Are there are areas of debate missing from Data in Society?

Administrative Issues

As the Administrator informed those receiving printed copies of the issue that, at the AGM held in London at the end of February 2020, the decision was taken to raise the subscription from £25 to £35 for those wishing to continue to receive printed copies (whilst the membership subscription only – with online access – would remain at £25;  £10 for those on low incomes), otherwise they would be taken off the distribution list which originally includes all 300+ members.

Please make sure you have updated your subscription, or make a donation! – by going to www.radstats.org.uk/membership/ where you can pay by cheque, standing order, PayPal – or by filling in your details in the form on page 54.

Roy Carr-Hill
Radstats Editor

Editorial, Coronavirus Special issue 126

We are flattered by the large number of papers submitted for this [special] issue.  Unfortunately, for reasons of cost, it was decided at the last Annual Conference not to print in colour, so we hope that the presentation of charts and figures has not been too spoilt.  For the same reason, we have had to limit the number of pages that can be staple bound (88) rather than with a spine.  This has meant:

  • after the tragic death of Professor Harvey Goldstein on 9th April from COVID-19-19, we solicited memorial tributes from members and others receiving heartfelt submissions from sixteen people, which we have decided to put on our website under the title of ‘Harvey Goldstein Memoria’;
  • carrying over some papers to the next issue and specifically those by Danny Dorling, Diana Kornbrot, Said Shahtahmasebi and dropping one planned section ’Epilogue’; the choice was made by myself on the basis of being relatively less directly relevant to COVID-19 or less statistical.

We are of course still ‘open for business’ in the sense of welcoming any commentary on the papers included in this issue, any further papers on COVID-19-19; and are particularly interested in receiving papers on countries not covered in section D of this issue.

Another proposal for generating material is the occasion of the publishing of the third RadStats compendium, Data in Society.  It is a landmark publication, bringing together many of the crucial issues around the production and use of quantitative information.

The contributors to Data in Society summarise many of the concerns around the accessibility and use of statistics in contemporary society. Examples include the lack of data from banking and financial organisations hides the extent of tax evasion of taxation. Government agencies are reducing the number of data series they make available for public scrutiny. The number of healthcare treatments in Britain provided privately is growing steadily.

The book is an eye-opener on the difficulties in holding governments and large organisations to account. Do you agree the authors’ interpretations?

As the editors acknowledge there are data topics the volume does not cover in detail. These include the use of statistics by legal practitioners, housing and homelessness data and climate change data.  The editors of the RadStats journal are planning to devote one journal issue to topics raised by Data in Society, and to topics not discussed in the book. Could you write an article for the journal on any of the topics above? Are there are areas of debate missing from Data in Society?

Roy Carr-Hill, Radical Statistics Editor

What models can and cannot do

Guest blog post by David Byrne

Models have been widely deployed in scientific discussion of the likely course of the COVID-19 pandemic to explore the potential impact of different policy interventions. However, any model is a necessary simplification of the system it describes.

Covid-19 is a biological intervention in the complex social systems – the plural is very important – which include human social interactions and policy interventions within existing social relations and institutional structures. These systems have emergent properties. That is not to say that modelling is useless but its use is necessarily limited.

Existing models are basically modifications of traditional epidemiological models of infectious disease transmission with parameters changed to reflect different timings and degrees of social interaction in response to social distancing and lock down regulatory interventions. There is a real problem of scale. Have these models got the scale of the system right before any attempt at description, let alone prediction, is attempted?

Communicable disease public health doctors have consistently made the point that what we have is not one nationwide outbreak but a set of local outbreaks which is why isolation of cases, tracing of contacts and isolation of those contacts is such an important part of the public health armoury. Most modelling seemed to deal only with the national scale and looked at the impact of policies like lockdown at that level although I am aware that local modelling is being attempted. Few models have examined in detail the impact of case isolation, tracing and contact isolation despite this having been a successful strategy in South Korea and elsewhere. An exception is  Kretzschmar, et al (2020). Plainly this is a very important set of interventions to consider.

Although an alternative form of approach has no immediate predictive capacity it is absolutely necessary to develop it in order to learn from this experience, for similar outbreaks will happen again. That approach is case-based process tracing and systematic case-comparison to establish what has worked better. That needs setting up now and whilst data is essential for developing it, and modelling can play a retrospective role if done at the right scale and with full incorporation of structural elements, it is not the only or even the best way to establish wha has worked where.

How might we  learn from this first wave of COVID-19 in order to find out what approaches have worked or not and in what contexts they have worked or not? Note the emphasized plural. Interventions have been interventions in different local complex systems and have themselves been complex. At national or even sub-national scales (where sub-national governments as with provinces in Canada have had appropriate powers) they have combined public health regulatory regimes (again note the plural) – different regimes in different places – with different levels of curative intervention depending on resources and even perhaps (on some limited evidence) different curative approaches, particularly in relation to the diagnostic anticipation and intervening prevention of cytokine storms. On this see the interviews with Chinese Intensive Care Physicians here: https://www.newscientist.com/article/mg24632783-600-wuhans-covid-19-crisis-intensive-care-doctors-share-their-stories/.

The first thing we need to know is just what has been done in different places alongside descriptions of the spatial and temporal contexts in which those things were done. We need careful process tracing and that means that we need good recording of what things were done in reasonable detail. This is a norm of any complex engineering production process but health systems are weak at full case recording other than those insurance based systems which generate financial records for costing. There have been attempts to improve this in non-insurance based systems but at present they are not fully developed. That kind of recording might be useful at the level of the individual patient and might provide the basis of a new wave of learning algorithm-based data mining to guide intervention, but it does not take account of institutional interventions at higher levels. It will be useful, indeed essential, in establishing treatment protocols – the sheer uselessness and inappropriateness of Randomized Controlled Trials other than for vaccine testing in a pandemic is obvious. It will not guide overall health system management.

There are well established tools in evaluation which can deal with the issue of post hoc exploration of what has worked in different contexts this time to guide policy and practice for next time. These are inherently mixed method in that they require the construction of narratives of what has been done, a mix of descriptive quantitative and qualitative specification of the contexts in which things have been done, and the use of data generated from those account to establish the multiple forms of intervention which have worked to different degrees. Equifinality rules OK!  The same outcome – control over the impact of the disease – can be generated in different and multiple ways. We need comparative process tracing based exploration of the multiple and complex ways in which systems have generated different outcomes.

This is precisely the set of problems addressed by  CECAN – a multi research council and UK government department funded investigation into the problems of Evaluating Complex Interventions Across the NEXUS (food, environment, water and energy). A range of approaches have been developed for this purpose:  CECAN’s website provides a full listing, https://www.cecan.ac.uk/.

Developed outside CECAN but interacting with it has been the very interesting approach of Dynamic Pattern Synthesis devised by Phil Haynes (see Haynes 2019).

This combines exploratory cluster analyses with Qualitative Comparative Analysis to explore how policy and practice systems have come to the outcomes they have reached.

Fundamental to this way of finding “what works” is a combination of qualitative materials and quantitative data. QCA – which is one tool but a good one with an established literature of effective use – requires the interpretation of qualitative narrative accounts of process to yield quantitative descriptions of interventions alongside quantitative descriptions of context. For an example of how this can be done see: Blackman et al.(2013).  Note that the level of measurement is often simply binary or at best ordinal specification of the attributes of the systems and of the interventions made within them.

Demanding documentation during crises is a hard thing to do but the construction of narratives, preferably on an ongoing real time basis but if necessary by careful historical investigation is absolutely necessary. We must always be able to say what has been done and if we can’t then we won’t learn what needs to be done.

References

Blackman et al.(2013). “Using Qualitative Comparative Analysis to understand complex policy problems.” Evaluation 19(2):126-140.

Haynes, Phil (2019). Social Synthesis – Finding Dynamic Patterns in Complex Social Systems. Abingdon: Routledge.

Kretzschmar, Mirjam E, Ganna Rozhnova, and Michiel E van Boven (2020). “Isolation and contact tracing can tip the scale to containment of COVID-19 in populations with social distancing.”

 

Picture This!

Guest blog post by R. Allan Reese

From the start, Downing Streets’ daily COVID press conferences have included various graphs slightly amended each day.  In mid April on the Allstat list, I described the presentation and labelling of these graphs as “Boilerplate Excel” and was duly reprimanded for “slagging the people concerned off behind their backs” with “destructive criticism”.  That was not my intention, nor do I accept that criticising a presentation equates with being derogatory about the author. I stand by the assertion that an equivalent lack of attention to spelling, grammar or punctuation would not be condoned in a PR organisation.  The Downing Street presentations were not prepared by hard-pressed, front-line health staff, but by media-savvy folk around the PM.  I wrote to the press office but received no response.

The basis for my criticisms comes from an approach I call Graphical Interpretation of Data (GID), expounded for example in various articles in Significance, freely available online.  Number 10’s daily sequences of graphs and data are available at https://www.gov.uk/government/collections/slides-and-datasets-to-accompany-coronavirus-press-conferences.

It could be argued that changing a style of presentation risks accusations of “spin” if this detracts from the day-to-day comparison.  However, some changes were made mid-stream. Initially, the daily numbers of deaths were plotted on a log scale labelled obscurely “5, 100, 2, 5, 1000, 2, 5, 10k” (Figure 1). The Daily Telegraph published a redrawn version labelling the grid line below the 5 with “0”.  The label “5” actually meant 50 deaths to allow the trajectory for each country to be aligned from the day 50 deaths were reported.  This is all very confusing.

Fig 1: 30 March. Early line graph with enigmatic Y labels and poor linkage to key.

I wrote to the Telegraph about this, and their presentations improved, as did Number 10’s, with labels 50, 100, 200, 500, etc.  The Metro commented on 1 April that the log scale made the growth in number of deaths appear less steep.  They quoted David Spiegelhalter that each presentation has its “advantages and disadvantages” and “there is no ‘right’ way”. However, less mathematically-minded readers would surely see the choice and the changes as spin.

On 8 April these presentations switched to a linear scale with a scale labelled “2K, 4K, 6K …”, thus avoiding showing “real” numbers or a disturbing axis title “Thousands of deaths”.  I described the use of ‘K’ as “nerdy”, especially as K in IT means a power of 2, not 1000.  It is notable that the daily format of the press conferences was a speech by a minister who then handed over to a scientist or medic to describe the graphs, reinforcing the attitude that graphs are for “boffins” – they might be over your head, dear simple reader.

Within GID one often has to guess at the intention of the author: Was the choice of notation accidental or deliberate?  Whom was this graph designed to inform?  I think we have to assume the direct audience are journalists who then interpret the graphs and data for their readership.  On the other hand, some features are so clearly defaults in spreadsheet graph production (e.g. text written horizontally or vertically), that I stand by the assertion that these presentations were handed out without further consideration or editing.

Downing Street’s daily “Global comparison of deaths” compares countries using a line chart. Initially the lines were just colour-coded with a separate key. Then the country names were written at the end of each line. Because each country’s “Day 0” was a different date, all the lines were different lengths.  Because there were ten lines, some were difficult to identify, as some colours were very similar and there was no redundancy (variation in other line characteristics).  The intended message appeared to be that the UK was buried in the middle, on a similar trajectory to the rest of Europe, with the US far worse (nearly three times as many deaths), while China and South Korea had fared much better.

It’s pretty obvious that crude numbers of deaths is a poor comparator, and there is much confusion between numbers of deaths and death rates. BBC’s More or Less (22 April) discussed this and identified the problem that converting using deaths per million population flung San Marino and Andorra to the top.  But you have the same problem calculating rates for many statistics by London boroughs: Westminster may come out top because so few people live there but many people commute.  The GID approach is to draw a graph (of numbers or rates), consider what message you wish to put across, and to revise the graph to clarify and emphasise.

Cristl Donnelly, on More or Less, suggested a better comparison would be to look at excess deaths in each country. This could also be standardised for population size, but might also allow a division into excess deaths from COVID and excess collateral deaths due to non-availability of other health services.

Another graph showed the number of deaths reported daily.  In the first weeks this was for hospitals only, but from mid-April it showed Daily COVID-19 Deaths in All Settings.  Note this was not necessarily the day the person died. Once the “peak” was passed, it was stressed in most presentations that there was a strong weekend effect with greater delays in reporting and hence a jump up each Monday. As a result the bar graph looks quite chaotic. A 7-day rolling average line clarified the general trend, but no visual effect was used to indicate weekends and the dates were labelled at 3-day intervals. Surely a good presentation would demonstrate the periodicity? (Fig 2)

Fig 2: 30 April. The bars show large day-to-day fluctuations while the smoothing line gives a clear, and more comforting, pattern. Which are weekends?

The other graph I draw attention to is the daily “New UK Cases”, based on the number of positive (PCR for antigen) tests reported that day (Fig 3). Initially this was constrained by test availability. By the third week of April a large excess of laboratory capacity over sampling numbers was reported.  According to the rubric, “there are likely many more cases than currently recorded here”, predominantly because sampling was restricted to hospital patients and staff, then extended to wider NHS staff and care workers, but (at the time of writing) not to the wider population.

Fig 3: 19 April.  The numbers written on bars were subsequently dropped as a separate data file was available. Without knowledge of the number of negative tests, it’s hard to evaluate any trend.

Showing the number of positives against an increasing number of daily tests, but not showing the number of tests, disguises any trend in prevalence. It would help if the number of tests or the proportion positive were also reported; these might be split to show the proportions in groups showing symptoms (expected high) and those tested as contacts (hopefully, lower).  Such comparisons were further hindered by gerrymandering the number of tests in late April to claim to have reached the arbitrary target of 100,000 tests “on” 30 April.

Among the problems with this chart are: the dates are written vertically with no indication of weekends or other divisions that might aid interpretation; the actual numbers are written on the bars, again vertically and hard to compare; for two thirds of the period shown the number varied between 4K and 6K and the largely overdrawn grid gives no assistance for comparison; most of the bars are split into two sections, linked to an enigmatic key (Pillar 1 and 2) which requires further recourse to the rubric for an explanation.

The split between “pillars” had me for one puzzled. It derived from the Secretary of State’s plan for five pillars of activity, but at various times the spokesmen distinguished the groups either by the targets for sampling (patients and hospital staff showing symptoms versus wider NHS staff and households) or by the place of testing (PHE versus commercial labs). I failed to find on the website any clear definitions to discriminate between “critical” and “key” workers.  By the end of April this graph had become quite impossible to learn from: the number of cases detected by NHS labs was going down despite PHE opening its Lighthouse labs but the number from other mass testing (private) labs appeared to increase each day. Hence, it appeared to say nothing about the national prevalence and, since there was no effective treatment, offer no assistance to individual patients.

My interpretation is that this is a case of “reporting the data” out of a sense of duty or as a totem to show the approach is “scientific”.  The layout obscures any visible trend except to show Pillar 2 as increasing over its range. High counts on 5 and 8 April are balanced by lows values on 4, 6 and 7 April, so a smoothing line would make the graph far easier to understand.  Having so many written numbers makes this far more of a table than a graph; it’s no good on a screen, especially during a presentation, but you can print it and turn it sideways. Or you could opt for horizontal bars with time running down the screen; this layout fits less well on a landscape screen, but could be split into a row of panes by week.

For screen use, one could easily angle the dates, add grid lines or background shading for weekends and Easter, round the numbers and omit the commas, move the “6K” gridline to label the actual maximum and add a 5000 gridline, change “0K” to “0”, and make the key one-stage.  As the interest is always in the latest figures, one could move the Y labels to the right-hand end.  I would reconsider the colours: the orange is intrusive and a “warning” tint, and the blue is quite dark. Lighter, more neutral colours for the bars would go better with an overlain smoother.

None of this is rocket science or takes much time or resource, but it does show one has thought about the graph and the audience. It shows competence and consideration.

Call for Papers for a Coronavirus special issue

Papers from all disciplines are invited on any relevant topic addressing political aspects of data and statistics.

Joint Editors are John Bibby and Roy Carr-Hill. Please submit an indicative title and brief description. Email: jb43@york.ac.uk

Papers are due by 1st May 2020, for publication the same month (provisional).

Recent issues of the journal are viewable at https://www.radstats.org.uk/journal/.

2020 Conference and Events

2020, London: “Learning from the Past to Build a Better Future
Friday 28th & Saturday 29th February, 2020
Radical Statistics 46th annual conference will take place at
St Luke’s Community Centre, 90 Central St, London EC1V 8AJ.
Register on Eventbrite.
On Saturday, 29th at the same location Radical Statistics will hold its AGM, discussion about Data in Society, and an extended discussion on the future of Radstats, as we approach our 50th year. All welcome.

Register now – Radical Statistics 2020 Conference

“Learning from the Past to Build a Better Future”

London: Friday 28th February 2020, with associated events on 27th and 28th February evenings, and the morning of 29th February.

The 46th annual Radical Statistics Conference will take place at St Luke’s Community Centre, 90 Central St, London EC1V 8AJ.

2020 marks the bicentenary of the birth of Florence Nightingale who was noted as “The Passionate Statistician”. We are proud to mark this with a Keynote Address from Lynn McDonald of Guelph University, world authority on Nightingale and editor of her collected works. Lynne’s talk is entitled “Florence Nightingale and Statistics: What She Did and What She Did Not”.

There will be many other talks and plenty of time for discussion, including:

Danny Dorling on The UK health crisis
Eileen Magnello on Nightingale: A radical and passionate statistician
Andrew Street on Revisiting Nightingale’s vision and hospital outcomes

Dave Byrne on The IFS Deaton review
Paul Marchant on Bad Stats and the public purse
Greg Dropkin on Radiation & A-bomb survivers in Japan

And discussion and sessions on the new Radical Statistics book ‘Data in Society: Challenging Statistics in an Age of Globalisation’

On the following morning, Saturday the 29th, Radical Statistics’ AGM will be held at the same location. In addition all are invited to discuss the future of Radical Statistics as an organisation as we prepare to enter our second half-century. There will be informal social events on the evenings of the 27th and 28th. We hope to end with a guided walk on a FN theme by a professional guide, immediately following the Saturday meeting.

Registration is now open.

Radical Statistics 123 (2019)

Cover pages
Editorial

Submitted Papers

Lies, damned lies, metrics & semantics: Exploring definitions of the end of leprosy (Hansen’s disease) and their implicationsF Houghton, M Winterburn, S Lama & B Cosgrove
Teaching for citizen empowerment and engagementJim Ridgeway & Rosie Ridgeway
Book Reviews


New Book: Data in Society – Challenging Statistics in an age of globalisation – with greatly reduced ‘pre-order’ price for members
Ludi Simpson
News


Minutes of AGM at Liverpool

Commission on ‘Future of Radical Statistics’

Radical Statistics Conference 2020

Revision of Book Reviewing process

New Book: Data in Society – Challenging Statistics in an age of globalisationwith greatly reduced ‘pre-order’ price for members