Blog

Nov
2019

Do School Performance Tables Really Measure School Performance?

Research published recently by the University of Bristol shows that analysis of school performance results change when pupils' background characteristics are taken into account. Commenting on the study, the Northern Powerhouse says that "league tables punish and reward the wrong schools". This study supports our findings from a previous blog, but also illustrates the importance of carefully considering what is included in a statistical model.

Jun
2019

Novel Educational Research using Longitudinal Survey Data

A recent article in the Guardian newspaper reports on a study published last month that looks at parents' responses to news from the Ofsted inspection of their child's school. Aside from the eye-catching headline, we found a number of aspects of this study interesting.

Mar
2019

Restaurant Ratings Model: Visualising the Key Findings

Recently we were inspired by an article in The Economist that looked at the demographic predictors of voting intention in the 2018 US congressional election. The article used one of our workhorse models, a logistic regression, and it contained a great visualisation of the results. We were keen to use the concept to present our own work.

Jan
2019

Presenting the Results of a Multinomial Logistic Regression Model: Odds or Probabilities?

Understanding the outputs from a model can sometimes be daunting. In this post, we show how we can convert the outputs from a multinomial logistic regression to gain more easily interpretable insights from the model.

Nov
2018

Measuring Human Behaviours and Traits with Item Response Theory

In the fields of education and psychology a frequent ambition is measurement. A statistical technique called Item Response Theory can be used to measure human behaviours and traits which are not directly measurable, such as confidence, ability and resilience.

Aug
2018

Analysing Outcomes with Multiple Categories

In a previous post, we discussed how to analyse categorical data with two outcomes using logistic regression model. Here we extend the logistic regression model to the case of an outcome with more than two categories.

Feb
2018

Analysing Categorical Data Using Logistic Regression Models

In a recent post we introduced some basic techniques for summarising and analysing categorical survey data. Whilst these methods are a great way to start exploring your categorical data, to really investigate them fully, we can apply a more formal approach.

Apr
2017

Key Driver Analysis

It’s important to identify and understand the drivers of key business outcomes. You might want to understand which aspects of your service influence how likely a customer will be to recommend you, for example. A so called key driver analysis can be used to address this sort of question.

May
2016

EU Referendum Polls: Modelling Voting Intention

In our latest blog on the EU referendum we look at fitting a model to the opinion poll results published in the run up to June 23rd. The model highlights how polling companies and practices influence the results and concludes that as of May 24th the probability of the UK remaining in the EU is about 68%.