Rigging Academic Articles to be more Progressive

I have previously discussed how articles are altered such that the conclusions appear progressive even though the data says anything but. My article on wikipedia in action is all about this, and my upcoming book Smart and Sexy, which will be published by Arktos, also discusses this with respect to intelligence testing and brain size measurements among many other things. The red pill subreddit recently had a confession of such manipulation by a firm which does team building training (archive in case first link gets lost). Though the source is ultimately 8chan, I have seen enough of this stuff elsewhere that I think that it is very plausible that this person is real and being truthful. The short of it is that males very clearly did better than females in an organized task requiring spontaneous coordination. The order of performance went all male–>mixed gender–>all female. Since that doesn’t work for pushing the narrative, nonsense factors were made to appear to be the most important so that it looked like the mixed teams did best. However, the data is still there and unchanged for those who pick at it and they will be able to see the male teams did have better performance. This is exactly what happened in the research paper I looked at with respect to racial relatedness in the wikipedia in action article. Though the writing seems to say people of different races can be more related than people from the same race, the data says the exact opposite. So, here too we will see another example whenever this “scientific” article is actually released. Keep an eye out for it because there is more than enough detail for us to look at their exercise description then trace it directly back to this confession. Having this in hand would be absolutely delicious.

Below, is the text of the original confession:

Alright /pol/, here is something to reinforce your opinions on women working in teams.

I am working as a team building coach in Germany. I hold courses for a company were teams are being tested and need to work together to fulfil their tasks. The goal is to have a better working team afterwards and to address problems within the team. Now before I get startet none of this is scientific. We use certain tests that need certain skills and are measured by certain factors, such as time needed, number of steps, etc. We record everything but it is not really a scientific test environment(no control groups, no randomization etc.)

To describe one particular exercise:

In a group of (usually) 16 people everyone gets blindfolded and gets an object. 4 people get the very same object. Now it is up to the people themselves to find the other 3 guys with the same object to form a group of 4 people and advance to the next excercise.

Now, the object is basically two dimensional and the key to finding your group is to count the edges. You cant see, but you can feel how many edges your object has. The perfect way would be to put a finger on one edge and then start counting the edges with your other hand until you know the number.

You can either tell everyone your method so time is not wasted(indicator of strong leadership skill) or you try to locate someone else, ask him for his number of edges and so on(poor leadership, no systematic working, you get the idea).

On saturday last week I had to finish a presentation(lll get back to that later, its the reason I post it here on /pol/) that was requested by a study group of the BMBF, the “Bundesministerium für Familie und Forschung”, Ministry of Family and Science here in Germany). We keep track of the performance of every team and have access to quite an amount of data. The exercise described has been done 356 times and I want to talk a little about the results.

All female teams did absolutely terrible. There are only very few instances in which the women figured out to count the edges and utilized the method to achive success, let alone figured out that someone should take the lead. Even with strong female lead a lot of women were unable to figure out how to count the edges without losing count. They were just starting to count the edges without indicating where they started. There were 2 reports of women claiming to have objects with more than 20 edges while the physical maximum is nine.

There is almost no difference between all female teams and female teams with strong female leadership. Strong female leadership does increase performance but only if detailed instructions are given by the female leader. It is necessary to describe the process step by step. The best performing all female team with strong female leadership did the following:

  1. Female leader commands everyone to be quiet several times while female are already discussing subjects not related to tasks.
  2. Female leader achieves silence, explains that you have to count the edges. She also explains the method.
  3. Female leader asks everyone to find other group members with the same number of edges.
  4. Chaos ensues. Female leader tries to get everyone to be quiet again.
  5. Female leader achieves silence and commands all with 7 edges to move towards her voice.
  6. Female leader appoints a sub leader for another number, asks group member to move towards the voice of the sub leader. Repeats the process several times until all groups are established.

Yet they are still the performing worse than mixed teams with male leader ship and a lot of mixed teams with poor male leadership. This is in stark contrast to an all male team with strong male leadership.

  1. Male leader demands silence right alter the tasks starts. There is no discussion, no period of figuring out who the leader is.
  2. Male leader says everyone should count the edges. There is no explanation of the method, yet there is no documented case in which a males failed to get the right number of edges.
  3. Male leader commands all 43 to move toward his voice, verbally appoints sub leaders for other groups while the other still move.
  4. Subleaders start to command their numbers to come close to their voice, it gets a little louder since 4 people are saying their number constantly.
  5. Groups are established.

This was the fastest documented case. Male teams with no strong leadership came in second. Someone usually yelled the method, everyone else copied it and then everyone just yelled his number until all groups were established. Mixed teams with (strong or poor) male leadership came in third, Mixed teams with strong female leadership didnt exist, it was always a male taking the lead or figuring out the method first, others copied it. Mixed teams with no leadership didnt exist either. Female teams with strong female leadership came in fourth and Female teams with no or poor leadership came in 5th by a long margin.

Now the problem lies within the results itself. They are considered sexist and discriminatory. It is not what the study group wants to hear, alter all it is for our super progressive government that sees women as superior to men and mixed teams as an ideal, which is why I was asked by my boss to make it look like mixed teams performed the best. I didnt want to fix the numbers, l just had to come up with something that made avarage results look good. So the number one indicator that determines whether it was a success or not is not the time needed, the efficency of the method or another metric. It is harmony within the group. display of natural leadership meaning no one forced someone else to listen to his opinion. Strong male leadership tended you yell out commands that addressed everybody and demanded certain actions while leadership in mixed teams usually asked politely. I also turned letting your fellow group members figure out the solution themselves, giving them time into a plus. Oh yeah, and creativity of solution, sehr wichtig.

Average became the new greatness. Mixed teams and female teams had top scores on all these feel good items, performance was ignored. lm about to hold this presentation later this week and hand over all the data. I am excited what they cook up with it but left a stinky trip mine in there. The numbers have not been changed and if they use this for any paper or recommendation in their proposals for new policies the compromising data is still in there.

So if you see someone claiming bullshit of women being superior or some shit you should take a closer look at the numbers. What was measured, how it was measured etc. lm pretty sure I am not the onyl one who riggs his data in a way that it looks better for the intended purpose.

Share Button

Wikipedia in Action on Race

I like to refer to Lewontin’s fallacy frequently when debating people who deny the biological basis of race. Wikipedia, while clearly not perfect, did have a reasonable article (at least for quick referral of lay-people) on the paper written by W.F. Edwards which coined “Lewontin’s fallacy.”(1) A brief overview is that in the 1970’s an academic social justice advocate published a paper(2) in which he claimed that there is more variation within individuals from one race than there is variation between different racial populations. So much that you can regularly find people of different races who are more similar to each other than they are to members of their own race. However, the first paper linked to above shows that the problem mainly stems from the fact that very few loci were studied by Lewontin. Allele frequencies differ between populations and with enough loci studied, the ability to distinguish between racial groups based purely on genetic information is quite high. Virtually 100%.

As is typical for pretty much all articles on Wikipedia, anything that isn’t politically correct can be expected to drift over time such that claims that are not PC are deleted, diluted, and placed next to a larger number of criticisms than is warranted such that it implies that the non-PC claims seem unsupported or only supported by very few outliers. Sometimes, like in this article, a paper which can be seen to support one conclusion actually supports the opposite on more careful inspection. All of this is the wikipedia version of death by 1000 cuts. I once tried editing the page on gender differences in intelligence and was basically run out and banned by marxist feminists. I assume this happens to anyone who objectively tries to include factual and balanced information into potentially politically incorrect articles. These same people got that article deleted or subsumed into gender differences in psychology for awhile, but it looks like it has been resurrected now. Honestly, the constant battle over these sorts of articles is just beyond all reason and I will never bother editing wikipedia again. Chances are your work is just going to get deleted and there are other platforms where that won’t happen.

Subjectively, it seems like this sort of thing has been happening to the Lewontin’s fallacy article, but I will let you be the judge:

Here is an old archived version of this article.

Here is an archived version of the current article.

Here is a direct link to the article. (It shouldn’t look different than the above link at the time of this post, but who knows what future changes will be made. In a year or two it could be interesting to compare these three versions)

The thing that is most obvious in my mind is that a paper discussed in an earlier version of the article which supported the concept of Lewontin’s fallacy has had any reference to it completely deleted. Here is the now deleted content:

Studies of human genetic clustering have shown that people can be accurately classified into racial groups using correlations between alleles from multiple loci. For instance, a 2001 paper by Wilson et al. reported that an analysis of 39 microsatellite loci divided their sample of 354 individuals into four natural clusters, which broadly correspond to four geographical areas (Western Eurasia, Sub-Saharan Africa, China, and New Guinea)

In addition, a paper which purports to undermine the concept that Lewontin’s thinking is fallacious is present at the end in both versions, but is quoted more (and very selectively) in the most recent version. In my opinion, the findings in both wikipedia versions are misrepresented.

In the old article this:

The paper claims that this masks a great deal of genetic similarity between individuals belonging to different clusters. Or in other words, two individuals from different clusters can be more similar to each other than to a member of their own cluster, while still both being more similar to the typical genotype of their own cluster than to the typical genotype of a different cluster. When differences between individual pairs of people are tested, Witherspoon et al. found that the answer to the question “How often is a pair of individuals from one population genetically more dissimilar than two individuals chosen from two different populations?” is not adequately addressed by multi locus clustering analyses. They found that even for just three population groups separated by large geographic ranges (European, African and East Asian) the inclusion of many thousands of loci is required before the answer can become “never”

On the other hand, the accurate classification of the global population must include more closely related and admixed populations, which will increase this above zero, so they state “In a similar vein, Romualdi et al. (2002) and Serre and Paabo (2004) have suggested that highly accurate classification of individuals from continuously sampled (and therefore closely related) populations may be impossible”. Witherspoon et al. conclude “The fact that, given enough genetic data, individuals can be correctly assigned to their populations of origin is compatible with the observation that most human genetic variation is found within populations, not between them. It is also compatible with our finding that, even when the most distinct populations are considered and hundreds of loci are used, individuals are frequently more similar to members of other populations than to members of their own population”

expanded into this:

In the 2007 paper “Genetic Similarities Within and Between Human Populations”,[20] Witherspoon et al. attempt to answer the question, “How often is a pair of individuals from one population genetically more dissimilar than two individuals chosen from two different populations?”. The answer depends on the number of polymorphisms used to define that dissimilarity, and the populations being compared. When they analysed three geographically distinct populations (European, African and East Asian) and measured genetic similarity over many thousands of loci, the answer to their question was “never”. However, measuring similarity using smaller numbers of loci yielded substantial overlap between these populations. Rates of between-population similarity also increased when geographically intermediate and admixed populations were included in the analysis

Witherspoon et al. conclude that, “Since an individual’s geographic ancestry can often be inferred from his or her genetic makeup, knowledge of one’s population of origin should allow some inferences about individual genotypes. To the extent that phenotypically important genetic variation resembles the variation studied here, we may extrapolate from genotypic to phenotypic patterns. […] However, the typical frequencies of alleles responsible for common complex diseases remain unknown. The fact that, given enough genetic data, individuals can be correctly assigned to their populations of origin is compatible with the observation that most human genetic variation is found within populations, not between them. It is also compatible with our finding that, even when the most distinct populations are considered and hundreds of loci are used, individuals are frequently more similar to members of other populations than to members of their own population. Thus, caution should be used when using geographic or genetic ancestry to make inferences about individual phenotypes”,[20] and warn that, “A final complication arises when racial classifications are used as proxies for geographic ancestry. Although many concepts of race are correlated with geographic ancestry, the two are not interchangeable, and relying on racial classifications will reduce predictive power still further.”

This paper… It had decent data and methodology actually. But as is almost always the case with these sorts of things, interpretations and framing of the results are key. It is clear that the people who wrote this are deliberately softballing their wording either to cover their ass (my guess) or to promote a more progressive narrative.

ω in the following quotes is defined as given a certain number of loci considered, the probability of individuals originating from two distinct geographical areas will be more similar to each other than to someone originating closer to them. I.E., the probability that two randomly selected individuals from different races will be more similar to each other than each is similar to a randomly selected member of their own race. Keep in mind that ω is not the same as determining what race a person is based on genetic data. Even with small numbers of loci and a high ω, there is very low probability of misclassifying the race of an individual person. From the very same paper used to undermine the Edwards’ paper:

[A relatively large ω is found with low numbers of loci] It breaks down, however, with data sets comprising thousands of loci genotyped in geographically distinct populations: In such cases, ω becomes zero.

With the large and diverse data sets now available, we have been able to evaluate these contrasts quantitatively. Even the pairwise relatedness measure, ω, can show clear distinctions between populations if enough polymorphic loci are used. Observations of high ω and low classification errors are the norm with intermediate numbers of loci (up to several hundred)

Thus the answer to the question “How often is a pair of individuals from one population genetically more dissimilar than two individuals chosen from two different populations?” depends on the number of polymorphisms used to define that dissimilarity and the populations being compared. The answer, ω, can be read from Figure 2. Given 10 loci, three distinct populations, and the full spectrum of polymorphisms (Figure 2E), the answer is ω ≅ 0.3, or nearly one-third of the time. With 100 loci, the answer is ∼20% of the time and even using 1000 loci, ω ≅ 10%. However, if genetic similarity is measured over many thousands of loci, the answer becomes “never” when individuals are sampled from geographically separated populations.

Molecular biologists and geneticists use a little bit different definition of polymorphism than some other branches in biology. In this case, they are referring to single nucleotide differences in the genome. This is equivalent to having one letter different in spelling a word. Prog and prig mean almost the same thing, but there is one letter difference which slightly changes the meaning. This is a reasonable analogy to the differences in the genetic code.

What this paper says (and it should be said with less tip-toeing) is that if you only consider a small number of these single nucleotide polymorphisms, there is a high degree of error and you can often erroneously conclude that two people from different races are more similar to each other than they are to individuals of their own race. The key word here is erroneously. This is a statistical problem, not biological fact. If you consider thousands of SNPS at once, then you have virtually no chance of encountering this problem. The authors of this paper found that Edwards was right and Lewontin was wrong. Individuals from two different races are never more similarly related than people from the same race, and the genetics supports this when you consider enough loci. It is pretty unambiguous. The quotes in the Wikipedia article and in the paper don’t really represent what the researchers actually found. The researchers had to dress this language up the way they did because of progressive influence in academia. Chances are they wouldn’t have gotten published if they were straight forward about what they found, and even if they could have published political heresy they may have had their careers ruined by SJWs in academia. See what happens when you don’t toe the line with the progressive narrative by reading what happened to a University of Texas researcher who didn’t find the “right” conclusions with regards to gay couples raising children. Though there is a huge problem with how Wikipedia articles are written and “maintained,” they wouldn’t have been able to misconstrue these results so badly if it weren’t from the same sorts of SJWs in academia malevolently influencing researchers. Though it shouldn’t be understated that the wikipedia editors did in fact selectively quote from this already bludgeoned paper. Two layers of SJW influence changed the findings of this paper to mean the exact opposite of what it actually found. Unbelievable. It is truly amazing that this sort of shenanigans is allowed to go on.

You might object that “thousands” is a huge number and that this demonstration of statistical problems convincingly shows that races don’t differ if it takes that many to reduce error to zero. However, the human genome is about 3 billion base pairs long. If you were to use 3000 base pair SNPs, which is consistent with the minimum in the paper, then you need to utilize only .0001% of the whole genome to reduce this error to zero. Or, if you want to consider SNPs only, there are about 10 million SNPs in the human genome. A sample of 3000 SNPs is only .003% of the total number of SNPs that could be used. This is a conservative estimate because their figure 2 indicates it only takes about 1000 SNPS to minimize this error. In other words, it only takes a vanishingly small fraction of the genome to relieve you of this statistical error that can find that humans from two different races are more similar to each other than either is to their own race.

Yet this paper, which so conclusively shows that human races are different from each other on the genetic level, is used to debunk the original Edwards’ paper. The author’s of the paper attempt to debunk themselves or at least pretend like they found the opposite of what they actually did. This paper is absolutely one of the worst instances of doublethink I have ever come across. It literally blows my mind. As a society, we seem to have a real hatred for truth when it comes to biological realities and the uninformed are clearly being purposefully told lies.

Sidenote: I know there was another article on cathedral entryism on Wikipedia in the alt-right in the last year or so, but for the life of me I can’t find it. If anyone can provide a link I would appreciate it. Edit: Found it.

(1) Bioessays. 2003 Aug;25(8):798-801. Human genetic diversity: Lewontin’s fallacy. Edwards

(2) The Apportionment of Human Diversity. R. C. Lewontin. 1972

(3) Genetics. 2007 May; 176(1): 351–359. doi:  10.1534/genetics.106.067355 PMCID: PMC1893020 Genetic Similarities Within and Between Human Populations J. Witherspoon, S. Wooding, A. R. Rogers, E. E. Marchani, W. S. Watkins, M. A. Batzer, and L. B. Jorde

Share Button