This post is an interview with John MacInnes, the expert instructor and author of SAGE Campus’ upcoming Statistical Significance, See Numbers in Data, Understand Probability, and Know Your Numbers online courses, launching in 2021. The courses are inspired by SAGE’s popular Little Quick Fixes book series.
Q: Hi John, tell us a bit about yourself?
I’m now retired, and Professor emeritus of Sociology and Statistics at the University of Edinburgh. I’m just finishing up a couple of years as Vice President of the Royal Statistical Society, which has been great fun, and doing some work on professional standards in Data Science. That leaves me time to indulge my hobbies, like restoring vintage hi-fi equipment and gazing out my living room window at the sea.
Q: What advice can you give faculty and institutions about the importance of quantitative methods training following your work as Strategic Advisor to the ESRC, British Academy and advisory role for the Nuffield Foundation Q-Step programme?
I think the COVID pandemic has been a reminder both of just how important numbers are, and of how difficult even apparently straightforward measurements like ‘deaths due to COVID’ to produce. British social science has tended to underestimate the importance of numbers and measurement, put scare quotes round mere ‘facts’ and emphasise the importance of theory. I’ve always thought that was mistaken.
In the aftermath of Trump and ‘fake news’ its surely more important than ever to stress that good social science analysis and theory depends upon good evidence and thorough knowledge of the facts. That needs numbers, quantitative methods and statistics. People have been saying this for over seventy years, so we’ve clearly not been saying it right! In my more pessimistic moments I think too much contemporary social analysis is pre-scientific. The argument is from authority rather than evidence and method. Instead of Aristotle or St Thomas Aquinas its Butler, Bordieu or Giddens.
That must change, as the world becomes more data centric. Distinguishing between the abuse or distortion of data and using it well will be central to politics and economics as well as science.
Q: What inspired you to make an online course with SAGE Campus?
Different resources suit different people and learning styles. I’m happy to try anything that might reach a different audience, or reinforce what people get through books, or face to face teaching. Online material can also be especially useful in parts of the world that don’t have the benefit of ready supplies of books and libraries.
Q: You’re the instructor for our Statistical Significance, Know Your Numbers, Understanding Probability and See Numbers in Data online courses. Who do you hope to most benefit from these courses?
I hope the people who benefit most are those who think they don’t like, can’t ‘do’ or won’t need numbers. In the data literacy course I used to teach I claimed that if you could count to ten you knew enough to tackle the course. Most quantitative methods use nothing much more complicated than fractions, and if you can tell the time, you know the basics of fractions. What is challenging, and does demand some effort and concentration is the logic. But the good news is that a little logic goes a surprisingly long way!
I think some social science students struggle with stats because it’s not the same as writing an essay or reading articles or monographs. You can’t skim stuff, go with the flow and gradually absorb the argument in the way you can with the journalistic style that has taken over most of the social sciences. Stats is more like climbing a ladder. You need to be completely secure with the first rung before you can get to the second, and so on. The good news is that the individual steps are not so hard as long as you take them slowly.
Q: Why do you think researchers find these topics so tricky? And why is it so important for institutions to get it right when teaching this?
We know that researchers struggle with these topics because of things like the ‘replication crisis’ in psychology, the arguments about how far scientific journals should rely on ‘p-values’ and significance tests, and the research that has shown that many scientists, and not just in the social sciences, have a poor grasp of the statistical procedures they use. That would be cause enough for concern, but we know that unless we get better at teaching and training on these issues things may get a lot worse because of the virtual explosion of data that is now available for analysis and the much greater ease of access to it.
I think the greatest contribution that institutions could make is also the simplest: provide time. Students find statistics challenging because as well as the details of the procedures (like what is a confidence interval or a correlation coefficient or whatever), they also have to take on board a different view of the world as a stochastic universe.
Most social science deals in grand causal narratives like the rise of industrial capitalism or intersection of ethnicity and class. Stats uses a different view of the world in which the probability of different empirical regularities rises or falls and is captured by variables and their distributions over space and time. No matter how diligent a student you are you just cannot become adept at finding your way around this new world after one 20 credit course.
The abundance of evidence that the data revolution is bringing is a great opportunity for the social sciences, but it is also a risk if people are not sufficiently careful about thinking through what kinds of conclusions different kinds of data can possibly support.
I used to tell my students ‘Never use a number if you don’t know where it has been!’. One of the downsides of the increasing ubiquity of data is that people imagine that good information about everything we might want to know must be just a click or two away. However, good measurement is resource intensive, slow and rare, and in data the devil is always in the detail. I’m always struck by how much we don’t know, don’t have any good data on, and fall back on speculation as a result.
Q: What do you think the benefits are of teaching these topics in an online course format?
The internet makes a lot of quants and statistics teaching easier because of things like video and animations. A lot of quantitative methods are visual, and they’re also dynamic - they’re about change - so online resources work really well.
One of the greatest early statistical thinkers was Ronald Fisher. He was nearly blind, which paradoxically gave him the most vivid geometrical imagination, which I’m sure inspired many of his mathematical ideas. A lot of statistical ideas come across as rather clumsy or convoluted in the written word, when transferred to a diagram they just look obvious!
If you want to learn these skills yourself or are interested in assigning them to your students, researchers or staff, register your interest for a free module of the online courses at the pages below:
Libraries can request a 30 day free trial to SAGE Campus for institution-wide access. Recommend us to your library or request a free trial today if you are an institutional administrator.