This post from David Robinson explains very well the point that statistical tests depend crucially on the assumptions you make before starting. The message is you need to look at your data and think about it a bit before doing a statistical test.
And, it also includes code to run the examples in R, my favourite, open source, statistical environment. Look at the post, learn some R and do better stats!
BMJ Open published a paper I wrote in August.
In preparation for a talk about the paper for my department, I had a look at the paper on the website. BMJ Open has a page of usage statistics for each paper, which shows that there are no citations of the paper. This is not unsurprising as its only four months since it was published. But then I saw that on the main papers page there is a section at the bottom called ‘Altmetric’.
I’ve reproduced the diagram here (by looking at the altmetric instructions for embedding badges):
This tells me that the paper has an altemtric score of 32 and, if you click on the diagram, will take you to altmetric’s website with further information. My paper is in the 98th centile of all articles tracked by altmetric. Is this a good thing? Surely being in the 98th centile is good?
Altmetric claims to make article level metrics easy and to track the impact of your papers beyond just citations. This is a good idea, since citations are not the only, or maybe even the main, purpose of research. I did this research in order to influence policy. Most policy makers would read papers, and may be influenced by them, but probably won’t write another research paper citing them. But what does altmetric track? It tracks tweets, newspaper articles, blogs etc. It does this by crawling the web looking for links to articles it tracks. So, the altmetric score can only ever reflect things that are written on the internet, and even then, only if they include a direct hyperlink to the source of the paper. Most policy doesn’t get made on the internet though.
Altmetric have at least done something to help produce actual data about the other impacts of research beyond citations, which is great. There is more to be done though, but I’m not quite sure how. And, I can’t help feeling pretty chuffed to be in the 98th centile of anything.
What do you think about Altmetrics? Are the numbers useful? What other alternative metrics should I look at for my paper?
I’ve been spending a lot of time using open source software recently, and been really impressed by the helpfulness of the open source community and the openness of the software. So, now I’m going to start giving something more substantial back.
This poster was presented at Wessex Society of Ultrasound Regional Anaesthesia annual scientific meeting on Monday. It is a description of our work to improve the pain relief for people with broken hips, but that is not the main purpose for posting it here.
The main reason is so that you can see the finished article here, and then click over to the gist that contains the source code here.
If you want, you can download my gist, and use it to create your own posters. Please be sure to cite appropriate sources, including myself, but mainly the people who made my work possible, such as the people over at
Welcome to my personal blog. This is my first post. Wish me luck in the blogosphere.