As a beginner to statistics, navigating the world of mixed effect models can be overwhelming. I recently stumbled upon a Reddit post from someone who’s struggling to decide whether to use dharma with their lmer model. I can totally relate.
The poster is working on an analysis of human brain region volumes, using a mixed effect model that includes region, gender, hemisphere, and age as factors. Initially, they used the lmer model, but after checking the assumptions for normal distribution of residuals and heteroskedasticity, they found some heavy tails and patterns in the residuals. They tried transforming the volumetric values using log, but it didn’t quite work out. Then, they switched to the glmmTMB model and used dharma to check the residuals, which seemed to produce better results. But here’s the thing: you can also use dharma with lmer models, which added to the confusion.
So, what’s a beginner to do? When faced with multiple options, it’s essential to understand the strengths and weaknesses of each tool. In this case, dharma is a great tool for checking residuals, but it’s not the only one. The lmer model is a popular choice for mixed effect models, but it’s not always the best fit.
As someone who’s new to statistics, it’s natural to feel overwhelmed by the sheer number of options available. But the key is to take it one step at a time. Start by understanding the assumptions of each model, and then experiment with different tools to see what works best for your specific problem. And don’t be afraid to ask for help – whether it’s online or in person.
In this case, the poster could try exploring other options, such as using different transformations or weights, or even switching to a different model altogether. The important thing is to stay curious, keep learning, and don’t be afraid to try new things.