Populism and p-values

Time for The Guardian to get the bad data-journalism award for this headline (25 February 2019).

Vaccine scepticism grows in line with rise of populism – study

Surges in measles cases map tightly to countries where populism is on the march

The report was a journalistic account of a paper by Jonathan Kennedy of the Global Health Unit, Centre for Primary Care and Public Health, Barts and the London School of Medicine and Dentistry, Queen Mary University of London, entitled Populist politics and vaccine hesitancy in Western Europe: an analysis of national-level data.1

Studies show a strong correlation between votes for populist parties and doubts that vaccines work, declared the newspaper, relying on support from the following single chart redrawn by The Guardian‘s own journalists.
PV Fig 1
It seemed to me there was more to the chart than the newspaper report. Is it possible this was all based on an uncritical regression line? Like this (hand drawn line but, believe me, it barely matters – you get the idea).
PV Fig 2
Perhaps there was a p-value too but I shall come back to that. However, looking back at the raw chart, I wondered if it wouldn’t bear a different analysis. The 10 countries, Portugal, …, Denmark and the UK all have “vaccine hesitancy” rates around 6%. That does not vary much with populist support varying between 0% and 27% between them. Again, France, Greece and Germany all have “hesitancy” rates of around 17%, such rate not varying much with populist support varying from 25% to 45%. In fact the Guardian journalist seems to have had that though too. The two groups are colour coded on the chart. So much for the relationship between “populism” and “vaccine hesitancy”. Austria seems not to fit into either group but perhaps that makes it interesting. France has three times the hesitancy of Denmark but is less “populist”.

So what about this picture?
PV Fig 3
Perhaps there are two groups, one with “hesitancy” around 6% and one, around 17%. Austria is an interesting exception. What differences are there between the two groups, aside from populist sentiment? I don’t know because it’s not my study or my domain of knowledge. But, whenever there is a contrast between two groups we ought to explore all the differences we can think of before putting forward an, even tentative, explanation. That’s what Ignaz Semmelweis did when he observed signal differences in post-natal mortality between two wards of the Vienna General Hospital in the 1840s.2 Austria again, coincidentally. He investigated all the differences that he could think of between the wards before advancing and testing his theory of infection. As to the vaccine analysis, we already suspect that there are particular dynamics in Italy around trust in bureaucracy. That’s where the food scare over hormone-treated beef seems to have started so there may be forces at work that make it atypical of the other countries.3, 4, 5

Slightly frustrated, I decided that I wanted to look at the original publication. This was available free of charge on the publisher’s website at the time I read The Guardian article. But now it isn’t. You will have to pay EUR 36, GBP 28 or USD 45 for 24 hour access. Perhaps you feel you already paid.

The academic research

The “populism” data comes from votes cast in the 2014 elections to the European Parliament. That means that the sampling frame was voters in that election. Turnout in the election was 42%. That is not the whole of the population of the respective countries and voters at EU elections are, I think I can safely say, are not representative of the population at large. The “hesitancy” data came from something called the Vaccine Confidence Project (“the VCP”) for which citations are given. It turns out that 65,819 individuals were sampled across 67 countries in 2015. I have not looked at details of the frame, sampling, handling of non-responses, adjustments and so on, but I start off by noting that the two variables are sampled from, inevitably, different frames and that is not really discussed in the paper. Of course here, We make no mockery of honest ad-hockery.6

The VCP put a number of questions to the respondents. It is not clear from the paper whether there were more questions than analysed here. Each question was answered “strongly agree”, “tend to agree”, “do not know”, “tend to disagree”, and “strongly disagree”. The “hesitancy” variable comes from the aggregate of the latter two categories. I would like to have seen the raw data.

The three questions are set out below, along with the associated R2s from the regression analysis.

Question R2
(1) Vaccines are important for children to have 63%
(2) Overall I think vaccines are effective 52%
(3) Overall I think vaccines are safe. 25%

Well the individual questions raise issues about variation in interpreting the respective meanings, notwithstanding translation issues between languages, fidelity and felicity.

As I guessed, there were p-values but, as usual, they add nil to the analysis.

We now see that the plot reproduced in The Guardian is for question (2) alone and has  R2 = 52%. I would have been interested in seeing R2 for my 2-level analysis. The plotted response for question (1) (not reproduced here) actually looks a bit more like a straight line and has better fit. However, in both cases, I am worried by how much leverage the Italy group has. Not discussed in the paper. No regression diagnostics.

So how about this picture, from Kennedy’s paper, for the response to question (3)?

PV Fig 5

Now, the variation in perceptions of vaccine safety, between France, Greece and Italy, is greater than between the remainder countries. Moreover, if anything, among that group, there is evidence that “hesitancy” falls as “populism” increases. There is certainly no evidence that it increases. In my opinion, that figure is powerful evidence that there are other important factors at work here. That is confirmed by the lousy R2 = 25% for the regression. And this is about perceptions of vaccine safety specifically.

I think that the paper also suffers from a failure to honour John Tukey’s trenchant distinction between exploratory data analysis and confirmatory data analysis. Such a failure always leads to trouble.

Confirmation bias

On the basis of his analysis, Kennedy felt confident to conclude as follows.

Vaccine hesitancy and political populism are driven by similar dynamics: a profound distrust in elites and experts. It is necessary for public health scholars and actors to work to build trust with parents that are reluctant to vaccinate their children, but there are limits to this strategy. The more general popular distrust of elites and experts which informs vaccine hesitancy will be difficult to resolve unless its underlying causes—the political disenfranchisement and economic marginalisation of large parts of the Western European population—are also addressed.

Well, in my opinion that goes a long way from what the data reveal. The data are far from from being conclusive as to association between “vaccine hesitancy” and “populism”. Then there is the unsupported assertion of a common causation in “political disenfranchisement and economic marginalisation”. While the focus remains there, the diligent search for other important factors is ignored and devalued.

We all suffer from a confirmation bias in favour of our own cherished narratives.7 We tend to believe and share evidence that we feel supports the narrative and ignore and criticise that which doesn’t. That has been particularly apparent over recent months in the energetic, even enthusiastic, reporting as fact of the scandalous accusations made against Nathan Phillips and dubious allegations made by Jussie Smollett. They fitted a narrative.

I am as bad. I hold to the narrative that people aren’t very good with statistics and constantly look for examples that I can dissect to “prove” that. Please tell me when you think I get it wrong.

Yet, it does seem to me the that the author here, and The Guardian journalist, ought to have been much more critical of the data and much more curious as to the factors at work. In my view, The Guardian had a particular duty of candour as the original research is not generally available to the public.

This sort of selective analysis does not build trust in “elites and experts”.

References

  1. Kennedy, J (2019) Populist politics and vaccine hesitancy in Western Europe: an analysis of national-level data, Journal of Public Health, ckz004, https://doi.org/10.1093/eurpub/ckz004
  2. Semmelweis, I (1860) The Etiology, Concept, and Prophylaxis of Childbed Fever, trans. K Codell Carter [1983] University of Wisconsin Press: Madison, Wisconsin
  3.  Kerr, W A & Hobbs, J E (2005). “9. Consumers, Cows and Carousels: Why the Dispute over Beef Hormones is Far More Important than its Commercial Value”, in Perdikis, N & Read, R, The WTO and the Regulation of International Trade. Edward Elgar Publishing, pp 191–214
  4. Caduff, L (August 2002). “Growth Hormones and Beyond” (PDF). ETH Zentrum. Archived from the original (PDF) on 25 May 2005. Retrieved 11 December 2007.
  5. Gandhi, R & Snedeker, S M (June 2000). “Consumer Concerns About Hormones in Food“. Program on Breast Cancer and Environmental Risk Factors. Cornell University. Archived from the original on 19 July 2011.
  6. I J Good
  7. Kahneman, D (2011) Thinking, Fast and Slow, London: Allen Lane, pp80-81

Regression done right: Part 1: Can I predict the future?

I recently saw an article in the Harvard Business Review called “Refresher on Regression Analysis”. I thought it was horrible so I wanted to set the record straight.

Linear regression from the viewpoint of machine learning

Linear regression is important, not only because it is a useful tool in itself, but because it is (almost) the simplest statistical model. The issues that arise in a relatively straightforward form are issues that beset the whole of statistical modelling and predictive analytics. Anyone who understands linear regression properly is able to ask probing questions about more complicated models. The complex internal algorithms of Kalman filters, ARIMA processes and artificial neural networks are accessible only to the specialist mathematician. However, each has several general features in common with simple linear regression. A thorough understanding of linear regression enables a due diligence of the claims made by the machine learning advocate. Linear regression is the paradigmatic exemplar of machine learning.

There are two principal questions that I want to talk about that are the big takeaways of linear regression. They are always the first two questions to ask in looking at any statistical modelling or machine learning scenario.

  1. What predictions can I make (if any)?
  2. Is it worth the trouble?

I am going to start looking at (1) in this blog and complete it in a future Part 2. I will then look at (2) in a further Part 3.

Variation, variation, variation

Variation is a major problem for business, the tendency of key measures to fluctuate irregularly. Variation leads to uncertainty. Will the next report be high or low? Or in the middle? Because of the uncertainty we have to allow safety margins or swallow some non-conformancies. We have good days and bad days, good products and not so good. We have to carry costly working capital because of variation in cash flow. And so on.

We learned in our high school statistics class to characterise variation in a key process measure, call it the Big Y, by an histogram of observations. Perhaps we are bothered by the fluctuating level of monthly sales.

RegressionHistogram

The variation arises from a whole ecology of competing and interacting effects and factors that we call the cause-system of the outcome. In general, it is very difficult to single out individual factors as having been the cause of a particular observation, so entangled are they. It is still useful to capture them for reference on a cause and effect diagram.

RegressionIshikawa

One of the strengths of the cause and effect diagram is that it may prompt the thought that one of the factors is particularly important, call it Big X, perhaps it is “hours of TV advertising” (my age is showing). Motivated by that we can generate a sample of corresponding measurements data of both the Y and X and plot them on a scatter plot.

RegressionScatter1

Well what else is there to say? The scatter plot shows us all the information in the sample. Scatter plots are an important part of what statistician John Tukey called Exploratory Data Analysis (EDA). We have some hunches and ideas, or perhaps hardly any idea at all, and we attack the problem by plotting the data in any way we can think of. So much easier now than when W Edwards Deming wrote:1

[Statistical practice] means tedious work, such as studying the data in various forms, making tables and charts and re-making them, trying to use and preserve the evidence in the results and to be clear enough to the reader: to endure disappointment and discouragement.

Or as Chicago economist Ronald Coase put it.

If you torture the data enough, nature will always confess.

The scatter plot is a fearsome instrument of data torture. It tells me everything. It might even tempt me to think that I have a basis on which to make predictions.

Prediction

In machine learning terms, we can think of the sample used for the scatter plot as a training set of data. It can be used to set up, “train”, a numerical model that we will then fix and use to predict future outcomes. The scatter plot strongly suggests that if we know a future X alone we can have a go at predicting the corresponding future Y. To see that more clearly we can draw a straight line by hand on the scatter plot, just as we did in high school before anybody suggested anything more sophisticated.

RegressionScatter2

Given any particular X we can read off the corresponding Y.

RegressionScatter3

The immediate insight that comes from drawing in the line is that not all the observations lie on the line. There is variation about the line so that there is actually a range of values of Y that seem plausible and consistent for any specified X. More on that in Parts 2 and 3.

In understanding machine learning it makes sense to start by thinking about human learning. Psychologists Gary Klein and Daniel Kahneman investigated how firefighters were able to perform so successfully in assessing a fire scene and making rapid, safety critical decisions. Lives of the public and of other firefighters were at stake. This is the sort of human learning situation that machines, or rather their expert engineers, aspire to emulate. Together, Klein and Kahneman set out to describe how the brain could build up reliable memories that would be activated in the future, even in the agony of the moment. They came to the conclusion that there are two fundamental conditions for a human to acquire a skill.2

  • An environment that is sufficiently regular to be predictable.
  • An opportunity to learn these regularities through prolonged practice

The first bullet point is pretty much the most important idea in the whole of statistics. Before we can make any prediction from the regression, we have to be confident that the data has been sampled from “an environment that is sufficiently regular to be predictable”. The regression “learns” from those regularities, where they exist. The “learning” turns out to be the rather prosaic mechanics of matrix algebra as set out in all the standard texts.3 But that, after all, is what all machine “learning” is really about.

Statisticians capture the psychologists’ “sufficiently regular” through the mathematical concept of exchangeability. If a process is exchangeable then we can assume that the distribution of events in the future will be like the past. We can project our historic histogram forward. With regression we can do better than that.

Residuals analysis

Formally, the linear regression calculations calculate the characteristics of the model:

Y = mX + c + “stuff”

The “mX+c” bit is the familiar high school mathematics equation for a straight line. The “stuff” is variation about the straight line. What the linear regression mathematics does is (objectively) to calculate the m and c and then also tell us something about the “stuff”. It splits the variation in Y into two components:

  • What can be explained by the variation in X; and
  • The, as yet unexplained, variation in the “stuff”.

The first thing to learn about regression is that it is the “stuff” that is the interesting bit. In 1849 British astronomer Sir John Herschel observed that:

Almost all the greatest discoveries in astronomy have resulted from the consideration of what we have elsewhere termed RESIDUAL PHENOMENA, of a quantitative or numerical kind, that is to say, of such portions of the numerical or quantitative results of observation as remain outstanding and unaccounted for after subducting and allowing for all that would result from the strict application of known principles.

The straight line represents what we guessed about the causes of variation in Y and which the scatter plot confirmed. The “stuff” represents the causes of variation that we failed to identify and that continue to limit our ability to predict and manage. We call the predicted Ys that correspond to the measured Xs, and lie on the fitted straight line, the fits.

fiti = mXic

The residual values, or residuals, are obtained by subtracting the fits from the respective observed Y values. The residuals represent the “stuff”. Statistical software does this for us routinely. If yours doesn’t then bin it.

residuali = Yi – fiti

RegressionScatter4

There are a number of properties that the residuals need to satisfy for the regression to work. Investigating those properties is called residuals analysis.4 As far as use for prediction in concerned, it is sufficient that the “stuff”, the variation about the straight line, be exchangeable.5 That means that the “stuff” so far must appear from the data to be exchangeable and further that we have a rational belief that such a cause system will continue unchanged into the future. Shewhart charts are the best heuristics for checking the requirement for exchangeability, certainly as far as the historical data is concerned. Our first and, be under no illusion, mandatory check on the ability of the linear regression, or any statistical model, to make predictions is to plot the residuals against time on a Shewhart chart.

RegressionPBC

If there are any signals of special causes then the model cannot be used for prediction. It just can’t. For prediction we need residuals that are all noise and no signal. However, like all signals of special causes, such will provide an opportunity to explore and understand more about the cause system. The signal that prevents us from using this regression for prediction may be the very thing that enables an investigation leading to a superior model, able to predict more exactly than we ever hoped the failed model could. And even if there is sufficient evidence of exchangeability from the training data, we still need to continue vigilance and scrutiny of all future residuals to look out for any novel signals of special causes. Special causes that arise post-training provide fresh information about the cause system while at the same time compromising the reliability of the predictions.

Thorough regression diagnostics will also be able to identify issues such as serial correlation, lack of fit, leverage and heteroscedasticity. It is essential to regression and its ommision is intolerable. Residuals analysis is one of Stephen Stigler’s Seven Pillars of Statistical Wisdom.6 As Tukey said:

The greatest value of a picture is when it forces us to notice what we never expected to see.

To come:

Part 2: Is my regression significant? … is a dumb question.
Part 3: Quantifying predictions with statistical intervals.

References

  1. Deming, W E (‎1975) “On probability as a basis for action”, The American Statistician 29(4) pp146-152
  2. Kahneman, D (2011) Thinking, Fast and Slow, Allen Lane, p240
  3. Draper, N R & Smith, H (1998) Applied Regression Analysis, 3rd ed., Wiley, p44
  4. Draper & Smith (1998) Chs 2, 8
  5. I have to admit that weaker conditions may be adequate in some cases but these are far beyond any other than a specialist mathematician.
  6. Stigler, S M (2016) The Seven Pillars of Statistical Wisdom, Harvard University Press, Chapter 7

Does noise make you fat?

“A new study has unearthed some eye-opening facts about the effects of noise pollution on obesity,” proclaimed The Huffington Post recently in another piece or poorly uncritical data journalism.

Journalistic standards notwithstanding, in Exposure to traffic noise and markers of obesity (BMJ Occupational and environmental medicine, May 2015) Andrei Pyko and eight (sic) collaborators found “evidence of a link between traffic noise and metabolic outcomes, especially central obesity.” The particular conclusion picked up by the press was that each 5 dB increase in traffic noise could add 2 mm to the waistline.

Not trusting the press I decided I wanted to have a look at this research myself. I was fortunate that the paper was available for free download for a brief period after the press release. It took some finding though. The BMJ insists that you will now have to pay. I do find that objectionable as I see that the research was funded in part by the European Union. Us European citizens have all paid once. Why should we have to pay again?

On reading …

I was though shocked reading Pyko’s paper as the Huffington Post journalists obviously hadn’t. They state “Lack of sleep causes reduced energy levels, which can then lead to a more sedentary lifestyle and make residents less willing to exercise.” Pyko’s paper says no such thing. The researchers had, in particular, conditioned on level of exercise so that effect had been taken out. It cannot stand as an explanation of the results. Pyko’s narrative concerned noise-induced stress and cortisol production, not lack of exercise.

In any event, the paper is densely written and not at all easy to analyse and understand. I have tried to pick out the points that I found most bothering but first a statistics lesson.

Prediction 101

Frame(Almost) the first thing to learn in statistics is the relationship between population, frame and sample. We are concerned about the population. The frame is the enumerable and accessible set of things that approximate the population. The sample is a subset of the frame, selected in an economic, systematic and well characterised manner.

In Some Theory of Sampling (1950), W Edwards Deming drew a distinction between two broad types of statistical studies, enumerative and analytic.

  • Enumerative: Action will be taken on the frame.
  • Analytic: Action will be on the cause-system that produced the frame.

It is explicit in Pyko’s work that the sampling frame was metropolitan Stockholm, Sweden between the years 2002 and 2006. It was a cross-sectional study. I take it from the institutional funding that the study intended to advise policy makers as to future health interventions. Concern was beyond the population of Stockholm, or even Sweden. This was an analytic study. It aspired to draw generalised lessons about the causal mechanisms whereby traffic noise aggravated obesity so as to support future society-wide health improvement.

How representative was the frame of global urban areas stretching over future decades? I have not the knowledge to make a judgment. The issue is mentioned in the paper but, I think, with insufficient weight.

There are further issues as to the sampling from the frame. Data was taken from participants in a pre-existing study into diabetes that had itself specific criteria for recruitment. These are set out in the paper but intensify the questions of whether the sample is representative of the population of interest.

The study

The researchers chose three measures of obesity, waist circumference, waist-hip ratio and BMI. Each has been put forwards, from time to time, as a measure of health risk.

There were 5,075 individual participants in the study, a sample of 5,075 observations. The researchers performed both a linear regression simpliciter and a logistic regression. For want of time and space I am only going to comment on the former. It is the origin of the headline 2 mm per 5 dB claim.

The researchers have quoted p-values but they haven’t committed the worst of sins as they have shown the size of the effects with confidence intervals. It’s not surprising that they found so many soi-disant significant effects given the sample size.

However, there was little assistance in judging how much of the observed variation in obesity was down to traffic noise. I would have liked to see a good old fashioned analysis of variance table. I could then at least have had a go at comparing variation from the measurement process, traffic noise and other effects. I could also have calculated myself an adjusted R2.

Measurement Systems Analysis

Understanding variation from the measurement process is critical to any analysis. I have looked at the World Health Organisation’s definitive 2011 report on the effects of waist circumference on health. Such Measurement Systems Analysis as there is occurs at p7. They report a “technical error” (me neither) of 1.31 cm from intrameasurer error (I’m guessing repeatability) and 1.56 cm from intermeasurer error (I’m guessing reproducibility). They remark that “Even when the same protocol is used, there may be variability within and between measurers when more than one measurement is made.” They recommend further research but I have found none. There is no way of knowing from what is published by Pyko whether the reported effects are real or flow from confounding between traffic noise and intermeasurer variation.

When it comes to waist-hip ratio I presume that there are similar issues in measuring hip circumference. When the two dimensions are divided then the individual measurement uncertainties aggregate. More problems, not addressed.

Noise data

The key predictor of obesity was supposed to be noise. The noise data used were not in situ measurements in the participants’ respective homes. The road traffic noise data were themselves predicted from a mathematical model using “terrain data, ground surface, building height, traffic data, including 24 h yearly average traffic flow, diurnal distribution and speed limits, as well as information on noise barriers”. The model output provided 5 dB contours. The authors then applied some further ad hoc treatments to the data.

The authors recognise that there is likely to be some error in the actual noise levels, not least from the granularity. However, they then seem to assume that this is simply an errors in variables situation. That would do no more than (conservatively) bias any observed effect towards zero. However, it does seem to me that there is potential for much more structured systematic effects to be introduced here and I think this should have been explored further.

Model criticism

The authors state that they carried out a residuals analysis but they give no details and there are no charts, even in the supplementary material. I would like to have had a look myself as the residuals are actually the interesting bit. Residuals analysis is essential in establishing stability.

In fact, in the current study there is so much data that I would have expected the authors to have saved some of the data for cross-validation. That would have provided some powerful material for model criticism and validation.

Given that this is an analytic study these are all very serious failings. With nine researchers on the job I would have expected some effort on these matters and some attention from whoever was the statistical referee.

Results

Separate results are presented for road, rail and air traffic noise. Again, for brevity I am looking at the headline 2 mm / 5 dB quoted for road traffic noise. Now, waist circumference is dependent on gross body size. Men are bigger than women and have larger waists. Similarly, the tall are larger-waisted than the short. Pyko’s regression does not condition on height (as a gross characterisation of body size).

BMI is a factor that attempts to allow for body size. Pyko found no significant influence on BMI from road traffic noise.

Waist-hip ration is another parameter that attempts to allow for body size. It is often now cited as a better predictor of morbidity than BMI. That of course is irrelevant to the question of whether noise makes you fat. As far as I can tell from Pyko’s published results, a 5 dB increase in road traffic noise accounted for a 0.16 increase in waist-hip ratio. Now, let us look at this broadly. Consider a woman with waist circumference 85 cm, hip 100 cm, hence waist-hip ratio, 0.85. All pretty typical for the study. Predictively the study is suggesting that a 5 dB increase in road traffic noise might unremarkably take her waist-hip ratio up over 1.0. That seems barely consistent with the results from waist circumference alone where there would not only be millimetres of growth. It is incredible physically.

I must certainly have misunderstood what the waist-hip result means but I could find no elucidation in Pyko’s paper.

Policy

Research such as this has to be aimed at advising future interventions to control traffic noise in urban environments. Broadly speaking, 5 dB is a level of noise change that is noticeable to human hearing but no more. All the same, achieving such a reduction in an urban environment is something that requires considerable economic resources. Yet, taking the research at its highest, it only delivers 2 mm on the waistline.

I had many criticisms other than those above and I do not, in any event, consider this study adequate for making any prediction about a future intervention. Nothing in it makes me feel the subject deserves further study. Or that I need to avoid noise to stay slim.

Anecdotes and p-values

JellyBellyBeans.jpgI have been feeling guilty ever since I recently published a p-value. It led me to sit down and think hard about why I could not resist doing so and what I really think it told me, if anything. I suppose that a collateral question is to ask why I didn’t keep it to myself. To be honest, I quite often calculate p-values though I seldom let on.

It occurred to me that there was something in common between p-values and the anecdotes that I have blogged about here and here. Hence more jellybeans.

What is a p-value?

My starting data was the conversion rates of 10 elite soccer penalty takers. Each of their conversion rates was different. Leighton Baines had the best figures having converted 11 out of 11. Peter Beardsley and Emmanuel Adebayor had the superficially weakest, having converted 18 out of 20 and 9 out of 10 respectively. To an analyst that raises a natural question. Was the variation between the performance signal or was it noise?

In his rather discursive book The Signal and the Noise: The Art and Science of Prediction, Nate Silver observes:

The signal is the truth. The noise is what distracts us from the truth.

In the penalties data the signal, the truth, that we are looking for is Who is the best penalty taker and how good are they? The noise is the sampling variation inherent in a short sequence of penalty kicks. Take a coin and toss it 10 times. Count the number of heads. Make another 10 tosses. And a third 10. It is unlikely that you got the same number of heads but that was not because anything changed in the coin. The variation between the three counts is all down to the short sequence of tosses, the sampling variation.

In Understanding Variation: The Key to Managing ChaosDon Wheeler observes:

All data has noise. Some data has signal.

We first want to know whether the penalty statistics display nothing more than sampling variation or whether there is also a signal that some penalty takers are better than others, some extra variation arising from that cause.

The p-value told me the probability that we could have observed the data we did had the variation been solely down to noise, 0.8%. Unlikely.

p-Values do not answer the exam question

The first problem is that p-values do not give me anything near what I really want. I want to know, given the observed data, what it the probability that penalty conversion rates are just noise. The p-value tells me the probability that, were penalty conversion rates just noise, I would have observed the data I did.

The distinction is between the probability of data given a theory and the probability of a theory give then data. It is usually the latter that is interesting. Now this may seem like a fine distinction without a difference. However, consider the probability that somebody with measles has spots. It is, I think, pretty close to one. Now consider the probability that somebody with spots has measles. Many things other than measles cause spots so that probability is going to be very much less than one. I would need a lot of information to come to an exact assessment.

In general, Bayes’ theorem governs the relationship between the two probabilities. However, practical use requires more information than I have or am likely to get. The p-values consider all the possible data that you might have got if the theory were true. It seems more rational to consider all the different theories that the actual data might support or imply. However, that is not so simple.

A dumb question

In any event, I know the answer to the question of whether some penalty takers are better than others. Of course they are. In that sense p-values fail to answer a question to which I already know the answer. Further, collecting more and more data increases the power of the procedure (the probability that it dodges a false negative). Thus, by doing no more than collecting enough data I can make the p-value as small as I like. A small p-value may have more to do with the number of observations than it has with anything interesting in penalty kicks.

That said, what I was trying to do in the blog was to set a benchmark for elite penalty taking. As such this was an enumerative study. Of course, had I been trying to select a penalty taker for my team, that would have been an analytic study and I would have to have worried additionally about stability.

Problems, problems

There is a further question about whether the data variation arose from happenstance such as one or more players having had the advantage of weather or ineffective goalkeepers. This is an observational study not a designed experiment.

And even if I observe a signal, the p-value does not tell me how big it is. And it doesn’t tell me who is the best or worst penalty taker. As R A Fisher observed, just because we know there had been a murder we do not necessarily know who was the murderer.

E pur si muove

It seems then that individuals will have different ways of interpreting p-values. They do reveal something about the data but it is not easy to say what it is. It is suggestive of a signal but no more. There will be very many cases where there are better alternative analytics about which there is less ambiguity, for example Bayes factors.

However, in the limited case of what I might call alternative-free model criticism I think that the p-value does provide me with some insight. Just to ask the question of whether the data is consistent with the simplest of models. However, it is a similar insight to that of an anecdote: of vague weight with little hope of forming a consensus round its interpretation. I will continue to calculate them but I think it better if I keep quiet about it.

R A Fisher often comes in for censure as having done more than anyone to advance the cult of p-values. I think that is unfair. Fisher only saw p-values as part of the evidence that a researcher would have to hand in reaching a decision. He saw the intelligent use of p-values and significance tests as very different from the, as he saw it, mechanistic practices of hypothesis testing and acceptance procedures on the Neyman-Pearson model.

In an acceptance procedure, on the other hand, acceptance is irreversible, whether the evidence for it was strong or weak. It is the result of applying mechanically rules laid down in advance; no thought is given to the particular case, and the tester’s state of mind, or his capacity for learning is inoperative. By contrast, the conclusions drawn by a scientific worker from a test of significance are provisional, and involve an intelligent attempt to understand the experimental situation.

“Statistical methods and scientific induction”
Journal of the Royal Statistical Society Series B 17: 69–78. 1955, at 74-75

Fisher was well known for his robust, sometimes spiteful, views on other people’s work. However, it was Maurice Kendall in his obituary of Fisher who observed that:

… a man’s attitude toward inference, like his attitude towards religion, is determined by his emotional make-up, not by reason or mathematics.

Data and anecdote revisited – the case of the lime jellybean

JellyBellyBeans.jpgI have already blogged about the question of whether data is the plural of anecdote. Then I recently came across the following problem in the late Richard Jeffrey’s marvellous little book Subjective Probability: The Real Thing (2004, Cambridge) and it struck me as a useful template for thinking about data and anecdotes.

The problem looks like a staple of elementary statistics practice exercises.

You are drawing a jellybean from a bag in which you know half the beans are green, all the lime flavoured ones are green and the green ones are equally divided between lime and mint flavours.

You draw a green bean. Before you taste it, what is the probability that it is lime flavoured?

A mathematically neat answer would be 50%. But what if, asked Jeffrey, when you drew the green bean you caught a whiff of mint? Or the bean was a particular shade of green that you had come to associate with “mint”. Would your probability still be 50%?

The given proportions of beans in the bag are our data. The whiff of mint or subtle colouration is the anecdote.

What use is the anecdote?

It would certainly be open to a participant in the bean problem to maintain the 50% probability derived from the data and ignore the inferential power of the anecdote. However, the anecdote is evidence that we have and, if we choose to ignore it simply because it is difficult to deal with, then we base our assessment of risk on a more restricted picture than that actually available to us.

The difficulty with the anecdote is that it does not lead to any compelling inference in the same way as do the mathematical proportions. It is easy to see how the bean proportions would give rise to a quite extensive consensus about the probability of “lime”. There would be more variety in individual responses to the anecdote, in what weight to give the evidence and in what it tended to imply.

That illustrates the tension between data and anecdote. Data tends to consensus. If there is disagreement as to its weight and relevance then the community is likely to divide into camps rather than exhibit a spectrum of views. Anecdote does not lead to such a consensus. Individuals interpret anecdotes in diverse ways and invest them with varying degrees of credence.

Yet, the person who is best at weighing and interpreting the anecdotal evidence has the advantage over the broad community who are in agreement about what the proportion data tells them. It will often be the discipline specialist who is in the best position to interpret an anecdote.

From anecdote to data

One of the things that the “mint” anecdote might do is encourage us to start collecting future data on what we smelled when a bean was drawn. A sequence of such observations, along with the actual “lime/ mint” outcome, potentially provides a potent decision support mechanism for future draws. At this point the anecdote has been developed into data.

This may be a difficult process. The whiff of mint or subtle colouration could be difficult to articulate but recognising its significance (sic) is the beginning of operationalising and sharing.

Statistician John Tukey advocated the practice of exploratory data analysis (EDA) to identify such anecdotal evidence before settling on a premature model. As he observed:

The greatest value of a picture is when it forces us to notice what we never expected to see.

Of course, the person who was able to use the single anecdote on its own has the advantage over those who had to wait until they had compelling data. Data that they share with everybody else who has the same idea.

Data or anecdote

When I previously blogged about this I had trouble in coming to any definition that distinguished data and anecdote. Having reflected, I have a modest proposal. Data is the output of some reasonably well-defined process. Anecdote isn’t. It’s not clear how it was generated.

We are not told by what process the proportion of beans was established but I am willing to wager that it was some form of counting.

If we know the process generating evidence then we can examine its biases, non-responses, precision, stability, repeatability and reproducibility. Anecdote we cannot. It is because we can characterise the measurement process, through measurement systems analysis, that we can assess its reliability and make appropriate allowances and adjustments for its limitations. An assessment that most people will agree with most of the time. Because the most potent tools for assessing the reliability of evidence are absent in the case of anecdote, there are inherent difficulties in its interpretation and there will be a spectrum of attitudes from the community.

However, having had our interest pricked by the anecdote, we can set up a process to generate data.

Borrowing strength again

Using an anecdote as the basis for further data generation is one approach to turning anecdote into reliable knowledge. There is another way.

Today in the UK, a jury of 12 found nurse Victorino Chua, beyond reasonable doubt, guilty of poisoning 21 of his patients with insulin. Two died. There was no single compelling piece of evidence put before the jury. It was all largely circumstantial. The prosecution had sought to persuade the jury that those various items of circumstantial evidence reinforced each other and led to a compelling inference.

This is a common situation in litigation where there is no single conclusive piece of data but various pieces of circumstantial evidence that have to be put together. Where these reinforce, they inherit borrowing strength from each other.

Anecdotal evidence is not really the sort of evidence we want to have. But those who know how to use it are way ahead of those embarrassed by it.

Data is the plural of anecdote, either through repetition or through borrowing.

Is data the plural of anecdote?

I seem to hear this intriguing quote everywhere these days.

The plural of anecdote is not data.

There is certainly one website that traces it back to Raymond Wolfinger, a political scientist from Berkeley, who claims to have said sometime around 1969 to 1970:

The plural of anecdote is data.

So, which is it?

Anecdote

My Concise Oxford English Dictionary (“COED”) defines “anecdote” as:

Narrative … of amusing or interesting incident.

Wiktionary gives a further alternative definition.

An account which supports an argument, but which is not supported by scientific or statistical analysis.

Edward Jenner by James Northcote.jpg

It’s clear that anecdote itself is a concept without a very exact meaning. It’s a story, not usually reported through an objective channel such as a journalism, or scientific or historical research, that carries some implication of its own unreliability. Perhaps it is inherently implausible when read against objective background evidence. Perhaps it is hearsay or multiple hearsay.

The anecdote’s suspect reliability is offset by the evidential weight it promises, either as a counter example to a cherished theory or as compelling support for a controversial hypothesis. Lyall Watson’s hundredth monkey story is an anecdote. So, in eighteenth century England, was the folk wisdom, recounted to Edward Jenner (pictured), that milkmaids were generally immune to smallpox.

Data

My COED defines “data” as:

Facts or impormation, esp[ecially] as basis for inference.

Wiktionary gives a further alternative definition.

Pieces of information.

Again, not much help. But the principal definition in the COED is:

Thing[s] known or granted, assumption or premise from which inferences may be drawn.

The suggestion in the word “data” is that what is given is the reliable starting point from which we can start making deductions or even inductive inferences. Data carries the suggestion of reliability, soundness and objectivity captured in the familiar Arthur Koestler quote.

Without the little hard bits of marble which are called “facts” or “data” one cannot compose a mosaic …

Yet it is common knowledge that “data” cannot always be trusted. Trust in data is a recurring theme in this blog. Cyril Burt’s purported data on the heritability of IQ is a famous case. There are legions of others.

Smart investigators know that the provenance, reliability and quality of data cannot be taken for granted but must be subject to appropriate scrutiny. The modern science of Measurement Systems Analysis (“MSA”) has developed to satisfy this need. The defining characteristic of anecdote is that it has been subject to no such scrutiny.

Evidence

Anecdote and data, as broadly defined above, are both forms of evidence. All evidence is surrounded by a penumbra of doubt and unreliability. Even the most exacting engineering measurement is accompanied by a recognition of its uncertainty and the limitations that places on its use and the inferences that can be drawn from it. In fact, it is exactly because such a measurement comes accompanied by a numerical characterisation of its precision and accuracy, that  its reliability and usefulness are validated.

It seems inherent in the definition of anecdote that it should not be taken at face value. Happenstance or wishful fabrication, it may not be a reliable basis for inference or, still less, action. However, it was Jenner’s attention to the smallpox story that led him to develop vaccination against smallpox. No mean outcome. Against that, the hundredth monkey storey is mere fantastical fiction.

Anecdotes about dogs sniffing out cancer stand at the beginning of the journey of confirmation and exploitation.

Two types of analysis

Part of the answer to the dilemma comes from statistician John Tukey’s observation that there are two kinds of data analysis: Exploratory Data Analysis (“EDA”) and Confirmatory Data Analysis (“CDA”).

EDA concerns the exploration of all the available data in order to suggest some interesting theories. As economist Ronald Coase put it:

If you torture the data long enough, it will confess.

Once a concrete theory or hypothesis is to mind, a rigorous process of data generation allows formal statistical techniques to be brought to bear (“CDA”) in separating the signal in the data from the noise and in testing the theory. People who muddle up EDA and CDA tend to get into difficulties. It is a foundation of statistical practice to understand the distinction and its implications.

Anecdote may be well suited to EDA. That’s how Jenner successfully proceeded though his CDA of testing his vaccine on live human subjects wouldn’t get past many ethics committees today.

However, absent that confirmatory CDA phase, the beguiling anecdote may be no more than the wrecker’s false light.

A basis for action

Tukey’s analysis is useful for the academic or the researcher in an R&D department where the environment is not dynamic and time not of the essence. Real life is more problematic. There is not always the opportunity to carry out CDA. The past does not typically repeat itself so that we can investigate outcomes with alternative factor settings. As economist Paul Samuelson observed:

We have but one sample of history.

History is the only thing that we have any data from. There is no data on the future. Tukey himself recognised the problem and coined the phrase uncomfortable science for inferences from observations whose repetition was not feasible or practical.

In his recent book Strategy: A History (Oxford University Press, 2013), Lawrence Freedman points out the risks of managing by anecdote “The Trouble with Stories” (pp615-618). As Nobel laureate psychologist Daniel Kahneman has investigated at length, our interpretation of anecdote is beset by all manner of cognitive biases such as the availability heuristic and base rate fallacy. The traps for the statistically naïve are perilous.

But it would be a fool who would ignore all evidence that could not be subjected to formal validation. With a background knowledge of statistical theory and psychological biases, it is possible to manage trenchantly. Bayes’ theorem suggests that all evidence has its value.

I think that the rather prosaic answer to the question posed at the head of this blog is that data is the plural of anecdote, as it is the singular, but anecdotes are not the best form of data. They may be all you have in the real world. It would be wise to have the sophistication to exploit them.

Bad Statistics I – the phantom line

I came across this chart on the web recently.

BadScatter01

This really is one of my pet hates: a perfectly informative scatter chart with a meaningless straight line drawn on it.

The scatter chart is interesting. Each individual blot represents a nation state. Its vertical position represents national average life expectancy. I take that to be mean life expectancy at birth, though it is not explained in terms. The horizontal axis represents annual per capita health spending, though there is no indication as to whether that is adjusted for purchasing power. The whole thing is a snapshot from 2011. The message I take from the chart is that Hungary and Mexico, and I think two smaller blots, represent special causes, they are outside the experience base represented by the balance of the nations. As to the other nations the chart suggests that average life expectancy doesn’t depend very strongly on health spending.

Of course, there is much more to a thorough investigation of the impact of health spending on outcomes. The chart doesn’t reveal differential performance as to morbidity, or lost hours, or a host of important economic indicators. But it does put forward that one, slightly surprising, message that longevity is not enhanced by health spending. Or at least it wasn’t in 2011 and there is no explanation as to why that year was isolated.

The question is then as to why the author decided to put the straight line through it. As the chart “helpfully” tells me it is a “Linear Trend line”. I guess (sic) that this is a linear regression through the blots, possibly with some weighting as to national population. I originally thought that the size of the blot was related to population but there doesn’t seem to be enough variation in the blot sizes. It looks like there are only two sizes of blot and the USA (population 318.5 million) is the same size as Norway (5.1 million).

The difficulty here is that I can see that the two special cause nations, Hungary and Mexico, have very high leverage. That means that they have a large impact on where the straight lines goes, because they are so unusual as observations. The impact of those two atypical countries drags the straight line down to the left and exaggerates the impact that spending appears to have on longevity. It really is an unhelpful straight line.

These lines seem to appear a lot. I think that is because of the ease with which they can be generated in Excel. They are an example of what statistician Edward Tufte called chartjunk. They simply clutter the message of the data.

Of course, the chart here is a snapshot, not a video. If you do want to know how to use scatter charts to explain life expectancy then you need to learn here from the master, Hans Rosling.

There are no lines in nature, only areas of colour, one against another.

Edouard Manet