Things have changed (II)

25 May, 2018 at 19:00 | Posted in Varia | Comments Off on Things have changed (II)

 
1968

The evidential sine qua non

24 May, 2018 at 18:10 | Posted in Theory of Science & Methodology | Comments Off on The evidential sine qua non

612090-W-Edwards-Deming-Quote-In-God-we-trust-all-others-bring-data

Things have changed …

24 May, 2018 at 15:41 | Posted in Varia | Comments Off on Things have changed …

 

Why the p-value is a poor substitute for scientific​ reasoning

24 May, 2018 at 15:07 | Posted in Statistics & Econometrics | Comments Off on Why the p-value is a poor substitute for scientific​ reasoning

 

A non-trivial part of teaching statistics is made up of learning students to perform significance testing. A problem I have noticed repeatedly over the years, however, is that no matter how careful you try to be in explicating what the probabilities generated by these statistical tests really are, still most students misinterpret them.

This is not to blame on students’ ignorance, but rather on significance testing not being particularly transparent (conditional probability inference is difficult even to those of us who teach and practice it). A lot of researchers fall prey​ to the same mistakes.

If anything, the above video underlines how important it is not to equate science with statistical calculation. All science entail human judgement, and using statistical models doesn’t relieve us of that necessity. Working with misspecified models, the scientific value of significance testing is actually zero —  even though you’re making valid statistical inferences! Statistical models and concomitant significance tests are no substitutes for doing real science.

In its standard form, a significance test is not the kind of ‘severe test’ that we are looking for in our search for being able to confirm or disconfirm empirical scientific hypotheses. This is problematic for many reasons, one being that there is a strong tendency to accept the null hypothesis since they can’t be rejected at the standard 5% significance level. In their standard form, significance tests bias against new hypotheses by making it hard to disconfirm the null hypothesis.

And as shown over and over again when it is applied, people have a tendency to read “not disconfirmed” as “probably confirmed.” Standard scientific methodology tells us that when there is only say a 10 % probability that pure sampling error could account for the observed difference between the data and the null hypothesis, it would be more “reasonable” to conclude that we have a case of disconfirmation. Especially if we perform many independent tests of our hypothesis and they all give ​the same 10% result as our reported one, I guess most researchers would count the hypothesis as even more disconfirmed.

Most importantly — we should never forget that the underlying parameters we use when performing significance tests are model constructions. Our p-values mean next to nothing if the model is wrong. Statistical​ significance tests DO NOT validate models!

statistical-models-theory-and-practice-2-e-original-imaeahk3hfzrmxz9In journal articles a typical regression equation will have an intercept and several explanatory variables. The regression output will usually include an F-test, with p-1 degrees of freedom in the numerator and n-p in the denominator. The null hypothesis will not be stated. The missing null hypothesis is that all the coefficients vanish, except the intercept.

If F is significant, that is often thought to validate the model. Mistake. The F-test takes the model as given. Significance only means this: if the model is right and the coefficients are 0, it is very unlikely to get such a big F-statistic. Logically, there are three possibilities on the table:
i) An unlikely event occurred.
ii) Or the model is right and some of the coefficients differ from 0.
iii) Or the model is wrong.
So?

The truth about what money really is

24 May, 2018 at 12:44 | Posted in Economics | 4 Comments

 

The ugly truth about gangsters of finance

24 May, 2018 at 12:26 | Posted in Economics | Comments Off on The ugly truth about gangsters of finance

 

Soziale Spaltung wird durch Privatschulen zementiert

23 May, 2018 at 21:45 | Posted in Education & School | Comments Off on Soziale Spaltung wird durch Privatschulen zementiert

inklusion_by_thomas_plassmann_6Der Bildungswissenschaftler Marcel Helbig hat für das Wissenschaftszentrum in Berlin die soziale Spaltung in Deutschland messbar gemacht. In seiner aktuellen Studie … hat er zusammen mit seiner Kollegin Stefanie Jähnen gewaltige Datenmengen aus 74 deutschen Städten ausgewertet. Er wollte wissen, wie sich einkommensschwache Menschen auf die Stadtgebiete verteilen. Das Ergebnis: Bis auf wenige Ausnahmen ist die Segregation in Deutschland gestiegen. Arme und Reiche, aber auch Akademiker und Arbeiter bleiben immer mehr unter sich. “In Berlin leben etwa 25 Prozent der armen Kinder in Gebieten, in denen sich Armut massiv ballt”, sagt Helbig. Umgekehrt gebe es Viertel, in denen Kinder gar keine armen Menschen mehr kennen. “Diese soziale Spaltung wird durch Privatschulen noch zementiert”, sagt der Forscher. Denn die boomen: In den vergangenen 20 Jahren hat sich die Zahl der Privatschüler um 30 Prozent erhöht, die Anzahl der privaten Grundschulen sogar vervierfacht. “Das ist ein ernsthafter Hinweis darauf, dass Privatschulen genutzt werden, um sich bestimmten sozialen Gruppen zu entziehen”, sagt Helbig …

“Die sozialen Unterschiede sind da am größten, wo ein gewisses Abgrenzungsbedürfnis gegenüber Hartz-IV- oder Migrantenfamilien vorhanden ist”, sagt der Bildungsforscher Helbig. “Ich habe absolutes Verständnis dafür, dass die Eltern das Beste für ihre Kinder wollen – aber gesamtgesellschaftlich betrachtet wird so Ungleichheit noch verstärkt.”

Fabienne Hurst/Die Zeit

Was uns Marx heute noch zu sagen hat

23 May, 2018 at 20:02 | Posted in Economics | Comments Off on Was uns Marx heute noch zu sagen hat

 

Phishing for Phools

23 May, 2018 at 18:13 | Posted in Economics | Comments Off on Phishing for Phools


As always a pleasure listening to this great and sympathetic economist!
[The book is an interesting and enjoyable read, but perhaps not as ‘innovative’ as the authors seem to think (especially vis-a-vis ‘behavioural economics’ they​ — unconvincingly​ — maintain their approach is “much more general” and in the spirit of “general equilibrium analysis.” That kind of contrived positioning is rather unnecessary​ since even if you’ve heard much of the story before from people like Tversky, Kahneman, or Thaler, it does not make the message less convincing or important!)]

Regression to the mean

23 May, 2018 at 08:52 | Posted in Statistics & Econometrics | Comments Off on Regression to the mean

Regression to the men is nothing but the universal truth of the fact that whenever we have an imperfect correlation between two scores, we have regression to the mean.

Getsemane

23 May, 2018 at 08:23 | Posted in Varia | Comments Off on Getsemane

 

Man vänjer sig …

23 May, 2018 at 00:01 | Posted in Varia | Comments Off on Man vänjer sig …

 

När Ekonomistas är som bäst

22 May, 2018 at 16:02 | Posted in Education & School | Comments Off on När Ekonomistas är som bäst

VlachosFör att utröna vilka effekter skolval har är det naturligtvis centralt att känna till vilken typ av skolor familjer föredrar. Det finns i princip två sätt att få kunskap kring detta: man kan fråga familjer vilken typ av skolor de uppskattar och man kan observera vilka skolor familjer faktiskt väljer. Vilken metod man väljer kan ge väldigt olika svar, vilket manifesteras i en debattartikel av Almega och Friskolornas Riksförbund som hävdar att obligatoriskt skolval är ett sätt att minska den sociala skolsegregationen.

Artikeln är baserad på en enkätundersökning som bland annat kommer fram till att föräldrar med svaga socioekonomiska förutsättningar värdesätter skolor med bra akademiska resultat särskilt högt. Socioekonomiskt starka familjer värderar däremot trivsel högre. Detta resultat påstås i rapportsammanfattningen vara i linje med vad den internationella forskningen visar, men att socioekonomiskt starka familjer skulle värdesätta skolors akademiska nivå lägre än socioekonomiskt svaga är minst sagt kontraintuitivt. Det kan därför vara bra att klargöra vad den internationella forskningen kring skolval egentligen kommit fram till.

En färsk studie från New York, som är under revidering för American Economic Review, finner att familjer söker sig till skolor (i detta fall high-schools) med högpresterande elever. Detta gäller för samtliga undersökta elevgrupper men resultaten visar att elever som själva är högpresterande är särskilt benägna att välja skolor med högpresterande elever. Däremot finns det inget som tyder på familjer skulle vara särskilt benägna att välja skolor som förvaltar sitt elevmaterial väl. Att en skola har ett högt ”förädlingsvärde” lockar alltså inte fler att välja den. Inte heller är familjer mer benägna att välja skolor som enligt studiens statistiska analys skulle passa just deras barn särskilt väl …

Liknande resultat går igen i andra studier. En välkänd sådan från North Carolina finner att familjer med stark socioekonomisk bakgrund är särskilt benägna att välja skolor med goda resultat, vilket alltså i hög grad fångar skolans elevsammansättning. Studien finner också att familjer från olika minoritetsgrupper har en tydlig preferens för skolor där respektive minoritetsgrupp är välrepresenterad. Lika söker alltså i hög grad lika.

Det finns även en svensk rapport som undersöker familjers faktiska skolval (lågstadiet). Hur en skola presterar på de nationella proven är inte särskilt viktigt medan avstånd till hemmet har betydelse. Vidare gör hög andel elever med utländsk bakgrund gör en skola mindre attraktiv, särskilt då för lågutbildade familjer med svensk bakgrund och högutbildade familjer med utländsk bakgrund. Skolor med många nyanlända ratas dock främst av svenska högutbildade familjer. Familjer med svensk bakgrund, speciellt de högutbildade, verkar även angelägna att välja skolor med socioekonomiskt starka elever, något som familjer med utländsk bakgrund verkar undvika. Generellt verkar familjer söka sig till skolor där eleverna tillhör samma socioekonomisk grupp som en själv. Lika söker återigen lika …

Snarare än att förlita sig på vad familjer uppger i enkäter är det önskvärt att studier av skolval baseras på hur familjer faktiskt väljer skolor. Det är trots allt vad familjer gör och inte vad de säger som i slutändan spelar roll för vilka konsekvenser skolvalet får. Dessutom spelar skolvalets utformning roll och ordningen med urval baserat på kötid till populära friskolor är med största sannolikhet något som förstärker skolvalets segregerande effekter. Oavsett skolvalets utformning måste dock slutsatsen från forskningen kring familjers preferenser vara att skolval knappast kan vara ett särskilt verksamt verktyg för att bryta den sociala skolsegregationen.

Jonas Vlachos

Som alltid initierat, relevant och intressant från en av våra duktiga yngre skolforskare.

Ekonomistas blandar allt från etablissemangsekonomers färdigtuggade skräpmat till synnerligen läsvärda artiklar från yngre forskare som Jonas Vlachos.

Sovra och läs!

Krugman’s Gadget Keynesianism

22 May, 2018 at 14:04 | Posted in Economics | 1 Comment

Paul Krugman has often been criticized by people like yours truly for getting things pretty wrong on the economics of John Maynard Keynes.

krugmanWhen Krugman has responded to the critique, by himself rather gratuitously portrayed as about “What Keynes Really Meant,” the overall conclusion is — “Krugman Doesn’t Care.”

Responding to an earlier post of mine questioning his IS-LM-Keynesianism, Krugman write:

Surely we don’t want to do economics via textual analysis of the masters. The questions one should ask about any economic approach are whether it helps us understand what’s going on​ and whether it provides useful guidance for decisions.

So I don’t care whether Hicksian IS-LM is Keynesian in the sense that Keynes himself would have approved of it, and neither should you.

The reason for this rather debonair attitude seems to be that history of economic thought may be OK, but what really counts is if reading Keynes gives birth to new and interesting insights and ideas.

No serious economist would question that explaining and understanding “what’s going on” in our economies is the most important task economists can set themselves — but it is not the only task.  And to compare one’s favourite economic gadget model to what madmen from Chicago have conjured up, well, that’s like playing tennis with the nets down, and we have to have higher aspirations as scientists.

There is also a somewhat disturbing and unbecoming coquetry in his attitude towards great forerunners. It smacks not so little of hubris to simply say “if where you take the idea is very different from what the great man said somewhere else in his book, so what?” Physicists arguing like that when discussing Newton, Einstein, Bohr or Feynman would not be taken seriously.

The absolute all-time low in Krugman’s response is this remarkable passage:

Has declaring uncertainty to be unquantifiable, and mathematical modelling in any form foolish, been productive? Remember, that’s what the Austrians say too.

I won’t comment on the shameful guilt-by-association part of the quote, but re uncertainty it’s absolutely gobsmacking how Krugman manages to mix up the ontological question — is the economy permeated by calculable risk or by genuine and often incalculable uncertainty  — with the epistemological question — how do we manage to analyze and model such an economy. Here Krugman seems to say — much in the spirit of Robert Lucas — that if reality is uncertain and non-ergodic, well then let’s just pretend it’s ergodic and susceptible to standard probabilistic analysis so that we can go on with our mathematical modelling! In other areas of science that would rightfully be considered fraud, but in modern mainstream economics, it’s obviously thought of as an unproblematic and justified procedure.

Being able to model a ‘gadget world’ — a world that somehow could be considered real or similar to the real world — is not the same as investigating the real world. Even though all theories in a strict sense are false, they cannot be unrealistic or false in any way. The falsehood or unrealisticness has to be qualified.

Where does all this leave us? Well, I for one, is not the least impressed by Krugman’s gadget interpretation of economics. And if labels are as uninteresting as he says — well, then I suggest Krugman and other ‘New Keynesians’ stop calling themselves Keynesians at all. I’m pretty sure Keynes would have appreciated not having his theories and thoughts being referred to by people having preciously little to do with those theories and thoughts.

People having bad luck when trying to think

20 May, 2018 at 18:47 | Posted in Economics | 4 Comments

A couple of years ago Andrew Lilico wrote an article in The Telegraph arguing that good economists almost always are mirabile dictu — right:

badluckLord Turner wrote two reports creating important government policies based on his claim that people are often irrational — his pension review of 2005 and his report on the financial crisis in 2009. Indeed, in his famous Turner Review of the financial crisis as chairman of the FSA he claimed to be able to demonstrate the failure of almost the entirety of finance theory as it had developed since the 1950s, albeit without feeling the need to name-check the great men who had produced that theory or consider how they might have defended their ideas.

But rationality is not an assumption of orthodox economic theory in that sense. Instead, it is what is called an “axiom”. No behaviour can prove that people aren’t, in fact, rational, because for an orthodox economist the only kind of explanation of any behaviour that counts as an economic explanation is an explanation that makes sense of that behaviour — that shows why the behaviour is rational.

Wooh! Who would have thought anything like that? Build your science on whatever foolish axioms you like — and since they are axioms no one can really criticize them and show that you are wrong.

Impressive indeed … 

« Previous PageNext Page »

Blog at WordPress.com.
Entries and Comments feeds.

%d bloggers like this: