Samuel Davenport

When can whole-genome SNP heritability be reliably estimated from summary statistics?

This study argues that fixing the intercept in LDSC improves SNP heritability estimation by reducing bias and variance, but LDSC remains biased under population stratification.

Estimating fMRI Timescale Maps

This paper introduces two methods to estimate timescales in resting-state fMRI, showing the time-domain method is more accurate and provides valid statistical inference with Python implementations.

Simultaneous Confidence Regions for Image Excursion Sets: a Validation Study with Applications in fMRI

The paper develops a method to construct simultaneous confidence regions for fMRI data across all activation thresholds and validates it using simulations and real data.

The expected behaviour of random fields in high dimensions: contradictions in the results of Bansal and Peterson (2018)

This paper shows that Bansal and Peterson's (2018) findings about Random Field Theory are incorrect due to coding errors, and that the theory correctly predicts cluster numbers in Gaussian fields.

Selective peak inference: Unbiased estimation of raw and standardized effect size at local maxima

The paper introduces a method to correct selection bias in peak inference for neuroimaging data by estimating raw and standardized effect sizes using resampling techniques.