Expert insights: Major Lester, founder of SPSS UK talks about fifty years of SPSS

This year SPSS is 50 years old. Development started in 1965 by a team of political scientists, frustrated at how much time they had to spent on manual data cleaning before they could start their analysis. The first release was in 1968 and by the end of the 1960s SPSS was in use in over 60 universities. SPSS expanded into the UK in 1985 and Major Lester was its first employee. We spoke to him about the history of SPSS, why it’s been so successful and how he sees its future.

Lorna: Could you start off just by talking to me a bit about your background and how you came to be involved in SPSS in the first place?

Major: Well, I started doing data analysis in 1970 at university. In those days we didn’t have any analytics packages so we had to write our own stuff. That was really the only part of my university degree that I enjoyed. I was doing statistics and it involved going up to the maths block and messing around with the computers. It was great fun and it also got me my first job as a consultant. I was a statistical consultant for 10 years after I left university, and I started using SPSS in about 1974.

I became an SPSS expert. It was the most popular package at the time because it was the easiest to use. Consultants liked it because it would get the job done quickly, and clients liked it because it was reasonably easy to learn to use.

In 1985 SPSS decided to set up an office in the UK and I got the job, mostly because of my experience as a consultant. I had enough experience of SPSS to be able to cover any work that needed to be done in the new office in terms of training and technical support, but really what SPSS wanted was for me to build a marketing machine in the UK and to come up with a method of selling the new PC version.

There was a wonderful wave of growth in the industry because the IBM PC had just come out and was very popular with users. They loved the ability to control it themselves, to be able to do it themselves and not worry about the cost, because of course in the old days when someone else had to do your analysis for you, every time you made a mistake you had to pay for it. We’re all so familiar with that concept of computing these days that we forget how expensive it was in the old days before the PC came in. Once you’d got your own PC you could just slash away at it and you didn’t have to worry about whether you made mistakes or anything like that.

The thing about statistics is that it’s often a bit nerve-racking for people. For most people at university, the statistics course was the most difficult part of their degree. Most people aren’t really interested in having to learn how to use some complex statistical package. They want to do some data analysis of some sort. They want to get the results but they don’t know how to do it. I developed a seminar system for marketing and selling SPSS, whereby you invite people to seminars and then show them that it’s easy to use. It’s not frightening, it’s not difficult.

The trick is not to sell SPSS as a powerful statistics package, but as an easy to use solution to the data analysis problems. That’s the system I developed, and after 10 years, our office had grown rather better than the other international offices of SPSS. In 1995 when the company reorganised, they wanted me to work to take my system to the other offices. I then worked for 10 years as an international manager and going around the world trying to encourage them to adopt the UK’s sales and marketing system. Then after I’d done that for 10 years, there really wasn’t much more for me to do, so I decided to take early retirement.

Lorna: It sounds like you enjoyed your time at SPSS. What was it that you liked about it?

Major: I enjoyed my job enormously because I’m evangelical data analyst. All my life I have wanted people to solve their problems by looking at data instead of just guessing as how they could solve the problem. I’ve always wanted to take a slightly more methodical approach and use data to do it. When I was a consultant you’d go to people and they would present a problem to you, but they wouldn’t think that they could solve it using data. Instead they’d think they could solve it just by having a meeting and everyone brainstorming or something like that. They wouldn’t think that there might be a better way. I felt that it was my life’s work to quietly try and bring people around to the idea that you’ll get a much better result if you just look at some numbers, and have a careful look at what the data says, rather than try and guess.

Lorna: Why do you think people are resistant to that message?

Major: They get entrenched and trapped in constantly sticking to what they know. Of course you will probably have some success doing that, but people don’t look at the details and realise that they will get much more success if they concentrate on the methods that are actually working rather than the ones that are not. Alternatively, people just look for the data to support their pet theories, but you can’t half do the job. You can’t say, “Well, I’ll just analyse these things because I think they look like they’re working and I’ll forget looking at that other stuff.” You really need to look at the whole picture.

Lorna: How has SPSS developed over the years?

Major: There are three main things that you do when you do statistics: descriptive statistics, testing, such as drug testing for instance, and forecasting. Those are the three main areas of work that you can expect to be involved in and of course you need packages for those. SPSS’s greatest strength, historically, was always in the area of descriptive statistics, so doing survey analysis and that sort of stuff.

SAS was perhaps more successful in drug testing because they had persuaded American drug agencies to recommend SAS as a package for doing testing, but SPSS was used occasionally for that as well. Then there was a whole gaggle of different forecasting packages, some of which I wrote myself. Forecasting is the area which more recently, has become more interesting and SPSS developed into this area with packages like SPSS Modeler.

However, for every person who wants to do forecasting there are ten people who want to do descriptive statistics. There’s many more people who want to just know some simple facts about what people think about their product or how they are planning to vote, just those sort of facts. There’s much work in that area and that’s always where SPSS has got its advantage because it’s always been by far the best package for doing that.

Lorna: Is that focus on descriptive statistics changing at all now, do you think?

Major: People are moving more towards forecasting and trying to do more sophisticated analysis because of the more powerful tools that are available, but I suspect there’s still even more descriptive statistics being done. If you think about how simple it is now to throw up a simple survey on Facebook, there’s much more of that kind of data available to people for analysis now than has ever been the case before.

Lorna: I’m wondering if people are doing quite basic stuff, but actually thinking that they’re really out at the cutting edge, and not really aware of what the possibilities might be?

Major: I think that’s certainly true. I think that was always true. What’s truer now is that the quantity of data that they’ve got to look at is probably somewhat swamping. Instead of just literally thinking about a question that you could ask and then working out what the answer to that is, you’re now in the situation where you’ve collected the data, automatically usually, and you’re saying, “What can I ask the data?” rather than, “I’ve got this question I want to answer.” It is data driven now, rather than management driven.

Tex Hull coding away as usual in 1986

Ironically, that is how SPSS started. Right at the very beginning it was data driven. That’s what drove Norman Nie, Tex Hull and those guys to write SPSS in the first place. They had some data, and it was driving them mad that they couldn’t effectively analyse it. People would come to them with data and they would have to do an awful lot of work managing the data before they could do any analysis. They thought, “Why don’t we just write a system that does all that work for us?” They could then get on with their lives as political scientists.

Lorna: What is it about SPSS that you think accounts for its ongoing popularity for 50 years of use and success?

Major: The reason that it has lasted is that it is an easy to use package that does data analysis. It still is the only package around that has the recode statement. If you try and imagine just doing the simplest piece of work without a recode, no one would ever do any bit of data analysis without having dozens of recodes.

When Norman and Tex were developing SPSS they always said the key moment was when they realised that mostly data analysis is getting the data in the right shape, so that you can apply some sort of statistical techniques.

If you have to write a COBOL program, or FORTRAN program, or something like that in order to do that work, you’re not going to do it. You’re going to have to go and get a programmer to do it. Norman and Text realised right at the very beginning that if you make the manipulation of the data simple, if you make the preparation, and of course the analysis simple, then people will do it themselves. You won’t have to employ other people to do it. People will have a go themselves. People who are not computer experts.

That was the revolution. This is what I’ve been doing all my life, training people who aren’t natural mathematicians, statisticians or computer programmers, to look at their own data. You discover much more if it’s your own data and you’re doing it yourself, than if you go and employ someone else to do it. Someone else will have an interest in it, but won’t have anything like the interest you have.

Lorna: What do you think will be the impact of things like R then, thinking about people doing things for themselves, and not having to code and so on?

Major: I think the key challenge with R is that it’s not very easy to use. I don’t think SPSS has too much to worry about to be honest with you. As long as it doesn’t lose itself up it’s own back side by making itself more difficult, then I suspect it will always be the best solution.

There’s so much stuff coming on now, where the users don’t have to do anything at all. They don’t have to think about the data analysis side of it at all, because it’s all web-based, and it all goes straight from the click response of the user through some tables. I suspect at that end of the market there will be competition that SPSS needs to think about but in the mass market, the middle part where the vast majority of survey analysis happens, SPSS will continue to dominate on that.

Since retiring from SPSS, Major spends his time playing bridge and tennis, wind surfing and working in his pottery studio.