Posts
Journal papers and ideas I find interesting, explained for occupational hygienists. All credit to the original authors referenced within — I encourage you to read the source work.
A short self-reflection on the first 12 months of doing my PhD.
Sometimes a chart is so messed up it’s almost beautiful.
An overview of priors in Bayesian data analysis, including a tool for exploring informative priors.
An introduction to hierarchical modelling, and the benefits of partial pooling over assessing data SEG-by-SEG.
How the 95% UCL is driven by sample size and exposure variability.
A SEG’s arithmetic mean also tells you the maximum possible exceedance fraction. A little maths explains why.
Including the variability of each task-based sample gives far more information about the possible TWAs.
Quantifying gravimetric analysis uncertainty under ISO 15767 and limits of detection.
How exposure standards are written in legislation, and the questions raised by the new workplace exposure limits.
A short, personal reflection on the AIOH ’24 conference in Perth.
A basic overview of the statistics used in IHstats — less on calculation, more on the intuition of what they mean.
Clarifying common misconceptions about confidence intervals, and why Land’s exact approximation may be unreliable.
An explanation of upper confidence limits of the arithmetic mean.
Exactly what an MVUE is, and how it compares to other ways of estimating the mean of lognormal data.
Our systems and tools do exactly what they were designed to do — which isn’t always what the user intended.
A review of Rappaport & Kupper’s hygiene-centric statistics textbook — challenging, but a genuinely interesting read.
A deeper look at how QQ plots are made, so you can get the most from them when checking your data’s distribution.
There’s no consensus yet on treating censored data, but one thing is clear: substitution is fabrication.
The algorithm that explores the parameter space of Bayesian calculations — the engine behind otherwise impossible solutions.
On the idealistic principles of occupational hygiene in tension with the priorities of business.
Estimating exposures without sampling — user-friendly modelling tools for preliminary assessment, with a note of caution.
A feel for how Bayes’ theorem works, and why you should be using it in occupational hygiene.
A reflection on my experience at the 2023 AIOH conference.
What protections should be in place for the misuse of information collected during real-time monitoring exposure assessments?
A more robust rule of thumb that allows you to compare the largest value with the OEL to determine compliance with high levels of certainty.
How openly should we share the information we collect in exposure assessments? Are there limits? And what metrics should we use to tailor the communicated information to achieve the best outcome?
The OSHA silica table has been practically replicated in Queensland, Australia. Unfortunately, following the control requirements does not guarantee effective management of exposures.
First Year of a Part-Time PhD
A short self-reflection on the first 12 months of doing my PhD.
The Best Worst Graph
Sometimes a chart is so messed up it’s almost beautiful.
What’s a Prior?
An overview of priors in bayesian data analysis, including an introduction to a tool that allows you to explore informative priors.
Hierarchical Modelling
An introduction to hierarchical modelling in occupational hygiene, and the benefits of partial pooling over assessing data SEG-by-SEG.
Why is my 95% UCL so high?
95% UCLs is related to sample size and exposure variability.
The Mean Caps Exceedance Fraction
Surprising at first, know what a SEG’s arithmetic mean is also tells you what the maximum possible exceedance fraction is. This post explains why this is with a little bit of maths.
Task to TWA Estimation
Why do TWAs have to be calculated from single measures of concentration and time? If we include the variability for each task-based sample we can get much more information about the possible TWAs.
Gravimetric Uncertainty
All measurements have uncertainty. Without investigating, you won’t know if it’s insignificant or a major concern. This post goes through quantifying gravimetric anaylsis uncertainty under ISO15767 and limits of detection. What could be more fun?
Interpreting Workplace Exposure Limits
A discussion on how workplace exposure standards are written in legislation and how the new workplace exposure limits are described. I share my confusion and the questions I have on the matter.
AIOH Conference ‘24 Reflection
A short, personal reflection on my experiences at the AIOH ‘24 conference in Perth.
Basic Hygiene Statistics
A basic overview of the statistics used in IHstats. There is little focus on the calculation, and more about the intuition of what they mean.
Confidence Interval Fallacies
An extension to the previous post on confidence intervals. This clarifies more explicitly some of the misconceptions, and interpretations. I also suggest that Land’s Exact approximation is unreliable (hopefully I’m wrong!).
What’s a UCL?
An explanation of upper confidence limits of the arithmetic mean
What’s an MVUE?
The MVUE is just “the best mean”, right? Not necessarily. Read about exactly what an MVUE is and how it compares to to the other ways of estimating the arithmetic mean of lognormal data when performing statistics reviews in occupational hygiene.
The Golems of Occupational Hygiene
Systems and tools in occupational hygiene do exactly what they were designed to do, though this may not always be what the user intended. These are a Golems of Occupational Hygiene.
Quantitative Exposure Assessment (Book Review)
A book review on Rappaport & Kupper’s “Quantitative Exposure Assessment”. A 2005 occupational hygiene centric statistics textbook. While challenging to follow at times for those without statistical background, it is a very interesting read to shape how you think about exposure assessments.
QQ Plots
QQ plots are incredibly useful in learning about the distribution of your sampled data. You may have used these plots in your analysis before, but this post will dive a little deeper into how they are made so you can get the most out of them.
Substitution is Fabrication
Censored data is an annoyance but a common occurrence in hygiene. While there is no consensus yet on the best method to treat censored data, one thing is clear: Substitution is Fabrication.
Markov Chain Monte Carlo
Markov Chain Monte Carlo is an algorithm that explores the parameter space of Bayesian calculations. It’s the engine tht makes otherwise impossible solutions reveal themselves step-by-step.
Reflection on ‘Worker Health First’
A reflection on how I think about idealistic principles of occupational hygiene in conflict with the priorities of business.