How to make a model

What is a model?

Question \(\rightarrow\) Model

What questions do we ask and why?

Our knowledge systems guide what questions we see as interesting and relevant. Western epistemology is only one (Tuhiwai Smith 2012; Wilson-Hokowhitu 2019)

What questions do we ask and how?

What methods are deemed valid, “objective,” “quantitative” has a history of racism, colonialism, and extraction (Tuhiwai Smith 2012; Clayton 2021)

Moʻokūʻauhau of likelihood

Ronald Fisher
  • proposed likelihood methods starting in 1912
  • racist
  • eugenicist
  • pawn of big tobacco

Moʻokūʻauhau of likelihood

Ronald Fisher

Available scientific knowledge provides a firm basis for believing that the groups of mankind differ in their innate capacity for intellectual and emotional development.

I am sorry that there should be propaganda in favour of miscegenation [bigoted term for marriage between perceived racial groups] *in North America as I am sure it can do nothing but harm.

Unfortunately, considerable propaganda is now being developed to convince the public that cigarette smoking is dangerous [Fisher was funded by the tobacco industry]

What do we do about it?

Consider this: European names are assigned (by some) to organisms the world over (Trisos et al. 2021), but the named people do not own those organism

Statistical methods were developed as tools for extraction and discrimination, but the racists and colonizers don’t own them

Being self-reflective about what questions we ask, how, and why can help us make use of the tools while pushing back against their original uses

Being able to use these tools, and be ethical about it, will help you get a job, and make a positive change with your position

With all that said

Question \(\rightarrow\) Model

With all that said

Question \(\rightarrow\) Predictions \(\rightarrow\) Model

Question to prediction

What is the effect of \(x\) on \(y\)?

  • prediction: if \(x\) goes up \(y\) goes down
  • prediction: \(y\) is greater for group \(x = g_1\) vs. \(x = g_2\)

Making a conceptual graph can help

Question to prediction

What is the effect of \(x\) on \(y\)?

Prediction to model

What is the source of variability?

Prediction to model

What is the source of variability?

Biological

  • lots of little additive or multiplicative processes
    \(\rightarrow\) central limit theorem and normal
  • random birth and death
    \(\rightarrow\) Poisson (technically negative binomial) or Binomial

Sampling and measurement

  • imprecise measurement
    \(\rightarrow\) central limit theorem and normal
  • imperfect detection
    \(\rightarrow\) binomial

How to make a model

For purposes of using maximum likelihood, it’s all about connecting a question to predictions and imagining how those predictions could be different—from one observation to the next, or one experiment to the next—based on random chance.

References

Clayton A. 2021. Bernoulli’s fallacy: Statistical illogic and the crisis of modern science. Columbia University Press.
Trisos CH, Auerbach J, Katti M. 2021. Decoloniality and anti-oppressive practices for a more ethical ecology. Nature Ecology & Evolution 5:1205–1212. Nature Publishing Group UK London.
Tuhiwai Smith L. 2012. Decolonizing methodologies: Research and indigenous peoples. Zed books.
Wilson-Hokowhitu N. 2019. The past before us: Moʻokūʻauhau as methodology. University of Hawaii Press.