Selbermach Samstag 272 (11.01.2020)

Welche Themen interessieren euch, welche Studien fandet ihr besonders interessant in der Woche, welche Neuigkeiten gibt es, die interessant für eine Diskussion wären und was beschäftigt euch gerade?

Welche interessanten Artikel gibt es auf euren Blogs? (Schamlose Eigenwerbung ist gerne gesehen!)

Welche Artikel fandet ihr in anderen Blogs besonders lesenswert?

Welches Thema sollte noch im Blog diskutiert werden?

Für das Flüchtlingsthema gibt es andere Blogs

Ich erinnere auch noch mal an Alles Evolution auf Twitter und auf Facebook.

Es wäre nett, wenn ihr Artikel auf den sozialen Netzwerken verbreiten würdet.

Wer mal einen Gastartikel schreiben möchte, der ist dazu herzlich eingeladen

Der univariate Fehlschluss (The Univariate Fallacy)

In Diskussionen zur Biologie wird häufig der „univariante Fehlschluss“ (univariate  Fallacy) begangen:

Dabei handelt es sich um die  Behauptung, dass, wenn es kein einziges, definierendes Merkmal gibt, das zur Trennung von zwei oder mehr Kategorien verwendet werden kann, diese Kategorien nicht existieren.

Einfache Beispiele wären:

Wenn Testosteron nicht der einzige Faktor ist, der sportliche Leistung bestimmt, dann kann Testosteron kein Faktor der sportlichen Leistung sein.

Oder

Weil es eine fließende Grenze der Intersexualität zwischen Männern und Frauen gibt gibt es keine Männer und Frauen

Zwei interessante Bilder dazu:

Einmal der Anfang dieses sehr interessanten Threads dazu:

Hier sieht man, dass bestimmte Eigenschaften im Dreidimensionalen Raum deutliche Cluster bilden, aber eben zusammengefasst auf einen eindimensionalen Raum auf der selben Ebene liegen können.

Diese Grafik aus diesem Artikel macht es noch etwas deutlicher:

Univariate Fallacy

 

Hier sieht man gut die Überlappung in dem eindimensionalen Raum obwohl sie im dreidimensionalen Raum klar getrennt sind.

Der Text hinter dem Link erklärt es auch noch einmal mathematisch, wobei ich das nicht wirklich nachvollziehen kann. Für in dem Gebiet besser aufgestellte aber sicherlich einen Besuch wert.

Ich kopiere daher mal aus dem obigen Tweet-Thread einige Ausführungen, mit Linkanpassung an die Blogumgebung statt an die Twitterumgebung:

1/ The modern far Left has a political agenda to destroy/deconstruct biological realities under the guise of Social Justice. A common way they go about this is by dishonestly applying univariate statistics to multivariate problems. This is called the Univariate Fallacy.

2/ This fallacy, when deployed, is commonly done using a single sentence buried within an article or essay couched around a broader narrative on the history of a particular type of oppression, such as sexism. Let me give you some recent examples of this fallacy in action.

3/ You’ll remember this @nature piece arguing that sex is a spectrum and that perhaps there are more then 2 sexes, even though over 99.98% of humans can be classified at birth as being unambiguously male or female.

4/ In this piece, they hold off deploying the Univariate Fallacy until the second-to-last sentence of a nearly 3500 word essay.

5/ See my  @Quillette essay for a more in-depth treatment of this and similar articles that try to obfuscate about the reality of biological sex.

6/ Then you’ll remember this other @nature piece arguing how the notion that there are or may be differences in the brains of human males and females is „neurosexism.“

7/ In this article they deploy the Univariate Fallacy much sooner, in the 4th paragraph.

8/ For a detailed look at the actual research completely debunking the Univariate Fallacy regarding sex-related brain differences in humans, see my previous thread about this below.

 

9/ And now we have today’s article in the New York Times claiming the importance of testosterone on athletic performance is a „myth.“
10/ In the 7th paragraph is, you guessed it, the Univariate Fallacy. Note the use of the words „single“ and „linear,“ both hallmarks of univariate statistics.
11/ This is, of course, highly misleading nonsense. Testosterone directly guides males through male puberty that results in bigger, faster, and stronger individuals than if they had not done so.
12/ Testosterone supplementation greatly increases muscle growth & protein synth in males & females, which is why it’s used by many bodybuilders & a banned substance in nearly every sport. Peruse the thousands of articles demonstrating this at your leisure.
13/ Furthermore, the sporting agencies that allow trans women to compete with females universally require them to reduce their testosterone levels below a certain threshold, for a certain amount of time, in order to reduce their competitive advantage.
14/ But why would lowering testosterone levels matter if it didn’t create an advantage in the first place? And, it’s not just CURRENT higher testosterone levels that give males an advantage, but PAST higher levels that guided the development of their bodies through male puberty.
15/ This thread is not an attack on univariate statistics. It is an attack on those who deploy them on multivariate problems to push a narrative. Don’t fall for it. It is scientific sophistry; smoke and mirrors designed to confuse you driven by a political agenda.

Realized I never really provided a simple definition of the Univariate Fallacy. Oops, sorry! But basically it is this: The claim that if there is no single, defining trait that can be used to separate two or more categories, then those categories do not exist.

To drive home the dishonesty of the Univariate Fallacy, realize that in the sentence „There is no single, category defining trait that separates male and female brains“, we could replace the word „male“ with „chimp“ and „female“ with „human“ and the statement would still be true.
I am just going to keep adding to this thread as I find more and more examples of the Univariate Fallacy. This article from the Independent critical of the recent ruling against Caster Semenya relies on the UF as the basis for their arguments.
And here it is. Yes, no single factor can 100% determine sex, but that doesn’t mean the categories „male“ & „female“ lack any coherent meaning. There may be extremely rare edge cases where we must use a reasonable heuristic, but this does not obliterate the categories as a whole.
This Guardian article deploys a popular version of the Univariate Fallacy w/ respect to human population genetics known as Lewontin’s Fallacy. This latches onto the fact that most genetic diversity exists within a population than between populations.
While technically true, this fact says almost nothing about the relatedness of any two individuals. The author’s claim that she (an Indian woman) may be more related to her white upstairs neighbor than to another Indian person simply isn’t true.
This faulty conclusion is the result of looking at genes at a SINGLE locus rather than the CORRELATIONAL STRUCTURE among many different loci. This is simply the Univariate Fallacy applied to genetics, otherwise known as Lewontin’s Fallacy. Link to paper
It is important to note that what we colloquially refer to as „races“ are very low resolution descriptors of relatedness, but statistically significant nonetheless.
Also, it is true that no human population is entirely genetically distinct. That would be impossible, but significant genetic structure due to relatedness exists. However, this is certainly no basis whatsoever for truly racist notions of racial essentialism and superiority.
And here’s a @nytimes article by Anne Fausto-Sterling from October of last year using the Univariate Fallacy as the foundation for their entire argument that sex is not binary because bruh, intersex.
„It has long been known that there is no SINGLE biological measure that UNASSAILABLY places each and every human into one of two categories—male or female.“
(…)
It’s always useful to keep in mind that all conversations about what constitutes a „female/male“ all take place within the little white box below. There is no sex spectrum, just a small number of people who’s sex is ambiguous or their sex genotypes don’t match their phenotype.
Ähnliches wurde schon unter anderem Namen hier diskutiert, nämlich unter Begriffen wie „Fuzzy Sets“ und „Unscharfe Mengen“. Eine Fuzzy-Menge (auch unscharfe Menge, englisch fuzzy set) ist eine Menge, deren Elemente nicht notwendig mit Gewissheit, sondern nur graduell zur Menge gehören.
Das scheint mir in die gleiche Richtung zu gehen, auch wenn es nicht per se das Gleiche ist. Vielleicht kann jemand in den Kommentaren die Unterschiede noch etwas aufklären.
Die Lewontin Fallacy könnte hier sicherlich auch für einige Diskussion sorgen, ebenso wie die dort verlinkten Studien.
Aber weil es häufiger als Argument auftaucht scheint es mir eine Figur zu sein, auf die ich noch häufiger zurückkommen werde