Statistics isn't exactly the most popular subject amongst students of psychology and social science, but it is nonetheless an important one to ensure they're trained to conduct reliable research. For years, statistics lecturers have had to frame their teaching around the limitations of statistics software, but statscloud aims to change all that; it's new interface is designed to work with students and lecturers, and facilitate the teaching of statistics. Here's how:

The current way to teach statistics

A typical lecture in statistics usually looks like this: a statistics lecturer discusses the theory behind a test, with references to formulas and a worked example, and goes on to demonstrate how that test is conducted in a stats package. Usually, this will involve some lecture slides with screenshots of the software running the analysis at key points, or perhaps a live demonstration of the analysis being done. As part of this, the lecturer will show students where they can find the analysis in a menu, which boxes they need to tick to get everything they need, and what to look out for in the output. While all this is going on, students frantically scribble down notes so they can do all of this themselves later.

Great. Well, not really.

This isn't the most elegant way to learn how to run a statistical anlaysis. The first issue is that students typically don’t get the opportunity to try this out in the software until after their lecture. When they do, they'll sit at a computer with the software open in front of them, their lecture slides to the side, their scribbled notes next to those, and perhaps a textbook on how to use the software thrown in for good measure. When it comes to running a new analysis for themselves, they’ll have to flick constantly between these resources, make sure they're following every step to the letter, try to understand exactly what they’re doing along the way - and do all of that without getting flustered about it.

This is the problem with statistics software. It’s not designed for students; it’s designed for professional statisicians who know exactly what they’re doing. Unless you know exactly what to click on, when, and what to look out for in the output, it's not at all easy to use and, crucically, doesn't really teach you anything as you're using it. A great deal of students will just click around until they see something that looks like the thing they want and then scan through the output to find something vaguely familiar to them. Once they've identified the numbers they think they need to write down, they'll write them in the report, but without an explanation from the software on what it's done for them or the theory behind what's happened.

Some universities have recognised this problem and have abandoned point-and-click stats software altogether, instead teaching R (via R Studio). The advantage of using R's command-line interface is that it allows students to start off with some very simple commands and then, once they're comfortable with those, build up to running some advanced analyses. R isn't really something you can use half-heartedly; you really have to throw yourself into it and understand what you're doing all the time. This gives it a clear edge over the point-and-click hit-and-miss nature of many statistics apps; it ensures students are doing everything properly and professionally.

The main drawback of teaching R in undergraduate psychology courses though is that not every student will take to it - and that's being polite. Students who want to specialise in qualitative or observational research will hardly be interested in spending hours on end typing out code. They haven't signed up to become computer programmers, and the thought of being consistently punished for typing a semi-colon instead of a comma is likely to put them off the field altogether. We don't want that.

A new way of teaching statistics

So, we need a compromise - and this is where statscloud comes in. Firstly, it's user-interface is quite different from other stats apps, and that's because it's designed for people new to statitsics and teach them what is happening as they use it. The app is full of information bars that break everything down for them in plain English, tooltips to help give clarity in critical places, and an in-built help system accessible at any time within the app.

Secondly, it's really easy to use inside lectures. Because it's a web app, and works on any device, students can have the app open on their device and follow the statistics lecturer live during an interactive session. The app doesn't need to be downloaded or installed either and, because it can be opened with a single link, lecturers can put a short URL up on screen which students can open on whatever device they have on them; laptops, Chromebooks, tablets or smartphones. The analysis can then be done together as a class where students play an active role in the learning process.

It's very easy for universities to use statscloud in teaching too. There is no licensing, nothing to download and nothing to install. If students can click on a link, they can open the app. Even though it's a web app, statscloud doesn't require a server to work either, so university departments don't need to worry about getting a dedicated server up and running for it, or setting up logins for all their students. Opening a project in statscloud is literally as easy as clicking on a link.

There are also some unique features in statscloud that make it an ideal stats package for teaching the theory behind statistical tests. One of these is 'live formulas' feature. Typically, when you run a test, your statistics software will give you a test statistic, but nothing much else; it won't show you the formula it used to calculate it, or any interim values it calculated in the process. The failure to show a worked formula in an analysis is a huge omission in stats software and something I was determined to include in statscloud right from the start.

Now, for instance, you can give students a worked example, then put that same example in a statscloud project. For analyses that require new variables to be calculated (e.g. ranked data for non-parametric tests), you can also save those variables in your dataset. The ability to save variables and see the working for all analyses is a fantastic educational tool.

Finally, for all you R enthusiasts, statscloud was designed to help transition people to R. Because many people will struggle to dive straight into a command-line-interface, statscloud offers a graphical representation of their data and analysis on screen first, and then shows what the code looks like to replicate it in R. All of the code is annotated too, so it offers a clear, informative introduction to the programming language.

Summary

As someone who once struggled with statistics, I constantly asked myself throughout statscloud's development: "what would have made it easier for me to learn statistics?" Pretty much every unique feature in statscloud was developed under that philosophy and, as odd as it may sound, I genuinely believe that's the best qualification I had for building the app. Hopefully, my own incompetence with statistics has made the app a stronger teaching tool for students and lecturers alike. Try it yourself here and see what you think!