# Opinions on Fb. Part #7. Mostly on statistics applications

Posted December 10, 2022

on:**“Scientific racism”?**

**Pseudo-scientific racism** (because racism is not scientific), was a way in which European colonial governments — and the **statisticians they hired** to do government survey, data collection, and interpretation of the data— justified their racist policies by using **statistical measurement** methods, often in a extremely** biased and incorrect way.**

Using **scientific terminology and measurements** to back their results was a way to sooth their guilty conscience.

An underlying theme of statistics is that **subjectivity is everywhere, **in almost every step in the interpretation of data.

Since the Data (collected measurements) almost never speak for themselves, it has to be interpreted in the surrounding context.

**And the context of how measurement were done is of utmost importance**.

The statisticians actually wants to know the **distribution function** of a set of data, but go ahead of trying to fit data to their wished hypothesis or **Null Hypothesis**, and thus use the flawed statistical equations.

The** p-value **of a hypothesis test is the probability of your test statistic taking the observed value or more extreme, *assuming* your null hypothesis is true.

In mathematical notation, this would be

The vertical line means “given H0”, which means “assuming the null hypothesis is true”. The value *x* would be, say a “z-score”, or a “t-statistic”, or a “chi-squared statistic”— if those are familiar words from your statistics classes.

One of the most common mistakes is assuming that **the p-value tells you the probability of your null** hypothesis given the data-evidence. This is ** wrong**.

The mathematical notation for this would be:

Image source: https://en.wikipedia.org/wiki/P-value#/media/File:P-value_in_statistical_significance_testing.svg

One of these alternatives is called **“Bayes factors”**.

The Bayes factor helps compare whether data under the null hypothesis is * more likely* or not than the alternative one.

It basically tells us how much the data (evidence) support the alternative over the null, which can be very useful.

Bayes probability takes the **context into consideration** by adopting a prior odds in the data:

Mathematically, it’s written like this:

## Leave a Reply