Hi, I’m Saskia.
I’m a recent graduate with an MMath in Statistics from the University of St Andrews. I’m currently embarking on a PhD at Cardiff Univeristy looking at Bayesian ordinal probit models for Likert-type item data.
I’ll just breakdown some of that jargon for you, in case it’s unfamiliar:
- Bayesian: a branch of statistics different to the default frequentist methods (sometimes referred to as “classical”). Frequentist is what most people are taught and what I learned in my undergraduate degree (I also did the one Bayesian module offered!).
- Ordinal: ordered categorical data. So “types of pet” is categorical data which is distinct from numerical data (e.g. height in metres) because basically we’re working with labels rather than numbers. If these labels have an order (e.g “small”, “medium”, “large”) then it’s ordinal data. The models I have ordinal data as the response variable.
- Probit: the cumulative distribution function (CDF) of a standard normal distribution. Here it’s just a mathematical function that’s used in the model.
- Likert-type item: Likert scales are a big deal in psychology and its related fields. You ask a participant a bunch of questions (“items”) and they can respond with “strongly agree”, “agree”, “neither agree nor disagree”, “disagree” or “strongly disagree”. We recode these answers 1-5, based on wether you want higher numbers to mean “disagree” or “agree”. With a true Likert scale the responses for the whole set of items are scored and summed to put them on the scale, a debatable practice. If the items fit the format but aren’t part of a scale, they’re called Likert-type items. Likert(-type) item responses are ordinal data.
This blog will contain academic book reviews, explanations of my work and general PhD advice. As I happen to suffer from anxiety, depression and dyspraxia I will also post about their associated issues from time to time. I believe they are relevant both to how I engage with my research and also with how readers may benefit from this blog.
I’m looking forward to sharing this journey with you all, so grab a cuppa and settle in for 3+ years of tea, equations and statistical musings.