3. STATISTICS IN GOVERNMENT Paul Altobell, Ministry of Defence

3.1 Introduction

Government has come to provide more and more services to more and more sectors of our society and economy. Carrying out these executive functions means that its need for numerate staff has much in common with that of Industry and Commerce. The pleas we hear from industry for graduates with statistical skills to bolster the growing Total Quality culture and for a numerate work force capable of being empowered to run and monitor processes at the point of delivery are strongly echoed in the Civil Service.

I would add a further reason why we need a general increase in numeracy. That is to enhance the democratic process. It is constantly surprising how difficult the man or woman in the street finds it to use straightforward official figures to judge how our public services are doing in (say) delivering health care or prison services, or how, as a nation we are doing relative to our competitors. As a corollary to this we also need those who present official figures to have the basic skills and commonsense to make sure the figures are easily understood.

I concede that we could have an interesting debate on whether the purpose of education was to meet the needs of the labour market. Let us leave that for another time. The only assumption I make here is that the labour market has at least some bearing on how pupils need to be educated.

3.2 Needs of the Civil Service

What kind of people do we need from schools and universities in the Civil Service, and what kind are we actually getting? Some are good, some are breathtakingly good, but many have gaps in their knowledge and abilities which might surprise you.

Let us take graduates first. My worry is that they know about statistics, but they don't know statistics. For instance, ask them, "If a MORI opinion poll on voting intentions takes a sample of 500 in a constituency of 30,000 voters how many should they take in a constituency of 60,000 voters?" Some say 1,000, some try to use a square root relationship, but very few say 500. Yet they can trot out the theory and formulae which show that the precision of the result relies broadly on the sample size and not on the population size. In other words they do not really connect what they have learned with its implications for real life. Why is this and has it any relevance to the way statistics teaching is approached in schools?

3.3 What is wrong

As a layman I feel the answer to the second part of the last question is "probably", and with some trepidation I suggest the graduates face three main problems which are also present in schools:

(i)  Statistics is not sold as a useful life skill;
(ii)  Statistics teaching contains too many mental wrist slaps;
(iii)  Statistics is taught the wrong way round.

Let me take things in the reverse order and deal with the last point first. There is a tendency to try to give the student a sound theoretical grounding and then mention one or two applications, or perhaps get students to do one or two synthetic exercises to check that they understand the theory.

So they are faced with formulae which float in some abstract area of their mind, not anchored to reality or experience. Where is the sense of exploration and discovery in that? Is this really the way to bring home the reality and implication of statistics? I think not. Contrast this with the approach proposed by the Sciences Education Board of the National Research Council in the USA. They recommend something like:

Now we come to the mental wrist slaps. We are too prone to seize on slight deviations from rigorous statistical rectitude (such as dividing by "n" instead of "n-1") so that the pupils, instead of feeling pleasure from the insight they have gained into some aspect of life represented by the data, are put off by their failure to follow (to them) arcane and complicated rules and procedures - a perfect example of aversion therapy.

So, to them, instead of statistics being a useful life skill it is something to avoid because some 'clever dick' will point out all the mistakes instead of giving credit for a good try at getting some insight into a problem. Just imagine how difficult it would be for pupils to learn French if they were rapped over the knuckles every time they made the slightest slip. The pleasure of successfully ordering a baguette in a Parisian bakery would be eclipsed by the carping criticism that an adjective did not strictly agree with the verb.

3.4 Conclusion

It is not my intention to tar all teachers with the same brush. Obviously many teachers are concerned about how statistics are taught. And from the conversations I have had with many teachers, it is clear that many schools are doing excellent work. However I strongly suspect that is not the case with all schools.

Top

Contents