In preparation for a huge fMRI dataset — which we’re nearly finished collecting — I’ve been trying to set up some kind of sensible pipeline for doing general statistical modeling / machine learning on functional brain network data. This is … Read more

## Bayes factors are almost impossible to use in practice

I recently came across an exchange in Psyc. Science that perfectly illustrates some of the problems involved with the use of Bayes factors. Scheibehenne, Jamil, and Wagenmakers (2016a) meta-analyze the probability of hotel towel reuse in two conditions, and compare … Read more

## Factor analysis is (probably) better with shrinkage

Extending on my previous *most-reported-summary-statistics-for-factor-analysis-favor-overfitting* comments, which focused on the common summary statistics used to evaluate factor models, I thought it might be interesting to tackle the other side of the problem, which is that covariance estimates tend to be … Read more

## Tuning curves as functional data

I don’t know anything about cellular neuroscience or single cell recording, but I recently came across the problem of estimating the receptive field of a neuron from its spiking frequency in response to movement at various angles. This is directional … Read more

## Bayesian functional linear models pt.1 – Estimating a mean

This post describes step 1 of my quest to build a fully Bayesian general linear model for functional data. I haven’t done it yet, and any solution is likely to be very computationally expensive, but so far I’ve had a … Read more

## Towards efficient IGT model simulation/estimation in R/Stan

As part of my playing around with alternative objective functions for estimating reinforcement learning models of the Iowa Gambling Task (IGT), I needed a way to quickly simulate large numbers of participants. Since base R is slow, I implemented a … Read more

## In which I complain about factor analysis

Factor analysis, like almost all forms of data analysis, is just matrix factorization. Let be a matrix containing observations of variables (or indicators). We suppose that the variables in can be written as linear combinations of unobserved (or latent) factors, … Read more

## Multiple hypothetical comparisons

You don’t actually have to *do* multiple comparisons to have a multiple comparisons problem — comparisons that you might hypothetically have done maybe if the data had been different perhaps will do it. I’ll show you.

Suppose that you’re conducting … Read more