Calling All Brandwatch Customers: Want to Try the Latest Product Features Before Anyone Else?
By Mercedes Lois BullJun 16
Published February 18th 2020
At Brandwatch Qriously, transparency has always been a priority.
As members of the British Polling Council, we release in-depth information about our methodology once survey results are released to the public (you can see our 2019 UK General Election release here and see how close we got to the final result in this article). And we’ve worked to exceed these requirements, for example by releasing an in-depth guide to our sampling methods.
This time, we’ve decided to go a step further.
We’re releasing the code and raw data behind our successful 2019 UK General Election prediction to Github.
This means that anyone can create their own splits in the data, learn how scientific polling is done, and draw their own conclusions about why the public voted as they did.
The 2019 General Election was historic not only in giving Boris Johnson the majority he needed to ensure the UK’s departure from the EU, but also in the scale of the Conservative win.
While Brandwatch Qriously and other pollsters have released analysis about why people voted as they did, access to this data allows anyone to come to their own conclusions, based on their own research.
Crucially, our poll included questions about a wide range of different issues, such as how the respondent voted in the past, what issues are important to them, and leadership preferences.
We also asked a number of demographic questions so, for example, you could look at perceptions about the importance of immigration by employment status, or political interest by party support.
What’s more, the code is released with an open source licence, and the data with a highly permissive licence, giving you the freedom to share and adapt the data, as well as produce your own work with it.
We also hope that even people without a specific interest in understanding the 2019 General Election results could use this as a way of learning about survey methods.
An important thing to note is that, while we provide code, you don’t have to use it. If you prefer, you can open up our data in a program of your choice (such as SPSS, Stata, or Excel). The repo contains a brief README.md file with further information.
The Python code provided is a very stripped down version of our actual prediction, so someone with basic Python experience should be able to run it, and play around themselves quite easily.
We hope you enjoy playing with the data, and we’d love to know how you get on in the comment section below.