## Working with numbers and statistics

### June 3, 2009

Just received my copy of Working with Numbers and Statistics, a “handbook for journalists”, by Charles Livingston and Paul Voakes.

As the subtitle suggests, this – like Math Tools for Journalists – is aimed squarely at the working journalist and the budding reporter, and as a consequence covers much of the same ground: averages, working with data, business reporting, and statistics. Also like Math Tools, it is written from an American perspective, which means some of the content (such as the table summarising the CPI from 1920-2004) is of limited value to a UK audience, although the underlying principles are equally relevant, of course.

It’s a well-ordered and written, if brief (113 pages) work, with plenty of relevant, worked examples. I found the lack of exercices a distinct downside, although in fairness the authors do explain their decision (“This book is not designed as a … textbook”) . Again, a list of further reading or bibliography would have been welcome, but perhaps that’s just me!

One aspect of the book which many will find useful is the conceptual ground covered. Not only does the book examine manipulation of data sets but it goes on to talk about correlation, hypothesis testing, causality and the rules of inference – that is, what conclusions can legimately be drawn from the data. For example, a section on logical implication succinctly demonstrates that the implication “if A then B” only works one-way and does not justify us concluding that “if B then A”.  This may seems an artificial construction, but just count the number of times you’ll hear sports journalists say (imaginary example!): “United haven’t won the title in any season when they failed to beat Fulham by fewer than two goals, so today’s 1-0 win must damage their title chances” – the fact may be odd, but there’s no reason to think the 1-0 scoreline will have any bearing on the outcome of the title race.

This conceptual work is neatly summarised in a chapter on fallacies and cautions, which should be required reading for all journalists (and their editors!).

Perhaps a section on CAR (Computer Assisted Reporting) could be included in subsequent editions, as part of the chapter on spreadsheets, although there is no shortage of textbooks covering this topic.

There is only point where I would quibble with the authors – they claim that in Great Britain, a “billion” is defined as a million million (10¹²), unlike the US definition of a thousand million. While that may be technically true (ie, the dictionary definition), in practice we Brits bowed to the inevitable some time ago and have long accepted the US value.

Working with Numbers and Statistics, A Handbook for Journalists by Charles Livingston and Paul Voakes is published in the UK by Taylor & Francis, £12.99

(Note: for those wanting to use UK CPI data, the Office for National Statistics site contains plenty of helpful spreadsheets)