Monday, 13 September 2010

Science writing and readability

Science erects high barriers to otherwise well-educated people, some of whom wear “I'm hopeless at science and maths” like a badge of honour. But the level of ignorance among our decision-makers and opinion-formers is far from honourable. It is shameful and dangerous.

The UK went to war in Iraq because members of parliament couldn’t analyse simple science and engineering, in weapons inspectors' reports that were readily available and in the Prime Minister's 24 Sep 2002 speech to Parliament about weapons of mass destruction.

Lacking the basic knowledge that would allow them to distinguish between “significant quantities of uranium from Africa” and “fissile matériel”, the MPs concluded that Saddam would have a nuclear weapon “in a year or two.”

Which was of course the explicit intention of Tony Blair, who used an old lawyer's trick of telling the truth but not the whole truth, and allowing the lie to happen in the listener's mind. By playing on MPs' ignorance of the difference
between natural and fissile uranium – and of the fact that this is the major technical challenge in making nuclear weapons – he frightened Parliament into voting for war.

Honest and effective science communication matters. Without it we launch illegal wars. We are duped by ministers and journalists. We melt the icecaps and pollute the seas.

So there is a huge burden on people who try to explain science to those who don’t yet get it. And especially on writers and scientists who aim to communicate science to children and young people.

There is no point in talking to Sarah Palin or Christopher Monckton about climate change. Or indeed anything else. These mature adults have made their minds up long ago. The same can be said for many other adults, far less irrational, malign or deluded than this perverse pair.

Kids and science
Children are different. All of them have opinions, but most are not yet set in stone. Kids grow, learn, assess evidence and change their minds – and are far more capable of doing so than most of us realise.

If we can find a way to keep the flame of science curiosity that burns brightly in most primary schoolchildren from flickering out when they hit adolescence, the world might hav
e a chance.

But it's not easy to keep teenagers interested in science. Traditional science teaching, aimed exclusively at future scientists, didn’t even try. The new curricula, with their focus on science for citizenship and making connections across the curriculum are much more promising.

But these need to be backed up and kept fresh by communications from the front lines – by direct contact with the people who are out there pushing back boundaries, getting excited, working long hours, making discoveries, doing the science.

Written word
Which brings me back to the barriers. There are
plenty. There are strong emotions stirred by past school science humiliations. There is the difficulty of some science concepts. There’s the maths. There’s the arrogant tone some scientists and writers take when ridiculing unscientific beliefs.

All these repel rather than attract non-scientists.

But I'm not going to talk about any of those here. Instead I want to look at a barrier that has been barely recognised. Which is surprising because it may well be the highest hurdle of them all.

Let me tell you a story – short and relevant. Until recently historians believed the Scots came originally from Northern Ireland. But it’s a myth. It has no foundation in evidence, as Glasgow University's Ewan Campbell has shown. Instead the west of Scotland and north of Ireland were part of a common culture for centuries, united by the sea.

Modern eyes look at big patches of blue on a map and see only a barrier. But the ocean was a highway to our ancestors.

When it comes to the written word, we writers and scientists are like ancient Celts surveying the sea. We see beyond it to old friends, other lands, new excitements. Most children and young people look in the same direction and see salt spray, high waves and hidden danger.

All this was brought home to me gradually but forcefully when trying to get inside the head of my dyslexic son, and when teaching English and science classes in secondary schools after 20 years of R&D in industry.

Scientists and writers are comfortable with words. Much more so than most people. I’m a writer, physicist and teacher, which gives me three overlapping perspectives. I know that what is taught is rarely what is learned, and that the reader seldom hears exactly what a writer is trying to say. Getting stuff out of one mind into another is much harder than many of us realise.

This is why almost everything written by scientists or science writers is unreadable by young people, or indeed most adults.

I am not talking about published papers, whose apparent obscurity masks clarity for colleagues. I’m referring to newspaper articles and blog posts that are supposed to be communicating science to non-scientists.

Most fail. The main reason is that the words and sentence structures in otherwise well-written pieces are too difficult for most readers. You might argue that science is hard and technical words unavoidable. This is true. But it is possible to write about hard science in a way that is accessible to young people. Younger than you might imagine.

I've seen cooperative learning classes with 10- and 11-year olds, who were using readable texts to research, discuss and present sophisticated science concepts, and astonishing even their teachers with the science they were capable of getting to grips with.

So let's take a look at the readability of some recent articles by science writers. Readability statistics are often criticised because good numbers don’t necessarily mean good writing. But this misses the point. We’re not using these stats to pick winners but to eliminate losers - or show where more effort might get us over the line. Good readability stats are a necessary but not sufficient condition for conveying science to young people in writing.

So what does the table three paragraphs down tell us? Well it’s a small sample. But this is an exercise I’ve done many times on many writers. It is not something I’ve only just discovered.

Real Science
I realised long ago that readability was a widespread and widely underestimated problem in science communication. It was one of my motivations in launching the Real Science website.

I'm going to repeat the readability analysis on these 12 writers and maybe a few more in the coming weeks, as they continue to produce articles. This should build a better picture of how consistently readable or otherwise they might be for younger people and non-scientists.

In the meantime it is interesting at least, and maybe significant, that the three most effective writers by readability - on this single exercise - are Alice Bell, a university lecturer in science communication, Alom Shaha, a science teacher and communicator, and Mark Changizi, a research scientist and author.

And that the three least effective – Mark Henderson, Alok Jha and Ben Goldacre – are all science journalists.

TitleWriterSourceReading easeGrade levelFog indexComment
Physicist, chemist & zoologist
Alice BellGuardian677.610.1Easy
King of the UniverseAlom ShahaBlog708.511Easy
Levels of real world wizardryMark ChangiziPsychology Today599.313.2Readable
The shark-toothed dinosaurEd YongBlog5310.313.1Hard
The grand designRobin McKieGuardian4911.815.1Hard
Analysis of PepsiGateMartin RobbinsBlog5211.713.4Hard
Rich-world diseasesDebora MacKenzieNew Scientist4412.114.2Unreadable
Glow in cattle's eyesKatia MoskvitchBBC4912.314.8Unreadable
If the world is going to hellDavid BielloScientific American4512.415.1Unreadable
Blind prejudiceBen GoldacreBad Science4613.216.4Unreadable
Humpback dinosaurAlok JhaGuardian3513.814.9Unreadable
Potential of genomic medicine

Photo credits
Images 2, 3 and 4 are by my highly artistic son, Douglas W. Blane.

Note on readability statistics

“Reading ease” in the table above is Flesch readability. “Grade level” is Flesch-Kincaid. “Fog index” is Gunning-Fog.

I obtain the first two using the built-in readability tool in Word – except when the second comes out at 12.0. In that case, since Word rounds down to 12.0 for this metric, I take the Grade level from this online analysis tool. That also provides me with the Fog index figures.

For the comment column I have crudely assigned one word to sum up readability for a target audience of young teenagers, as follows: Grade level >12 is Unreadable, (10-12] is Hard, (7-10] is Readable, and ≤7 is Easy.

The accepted wisdom is that texts aimed at a wide audience generally need a Fog index less than 12, and that targets for the other two indices should be 60-70 and 6-8. Note that readability increases as Flesch increases, but decreases with Flesch-Kincaid and with Fog.

For comparison, Reader's Digest has a Flesch readability of about 65. Time magazine scores 52, while Harvard Law Review is in the low 30s.

My own observation is that readability stats of one writer in one publication tend to be fairly consistent. But good writers vary their style to suit their audience. When I write for Glasgow University alumni, my Flesch-Kincaid often exceeds 12. For The Guardian it comes out at 10-11, while my articles in Times Educational Supplement and Real Science tend to cluster around 6-8


  1. Note: that Guardian piece of mine was edited, I think by Alok Jha.

  2. Glad you did this exercise. I think a lot of people just aren't aware of what it takes to reach broad audiences.

    From a personal perspective, during my day job, I write for a grade level of 8 or 9, so it's a welcome relief to be able to cut loose a little bit in my blogging. 10's not actually far off what I'm aiming for. For what it's worth the last two articles I've written should fairly span the gulf of easy to hard in terms of topics. If we take that as the range, I write at a grade between 9 and 11, and a reading ease of 49 to 59.

  3. Great article. Glad to see that a blog post I just wrote came out somewhere between Mark Changizi and Ed Yong. Thanks for pointing to the metrics - will be using these in the future!

  4. Sorry should add that I think this is still an interesting piece!

    My note was due to (a) embarrassment (!) and moreover (b) a desire to stress that there's a context behind each piece of writing, one that often involves more than one person. Similarly, individuals write differently in different contexts - I hate to think where my thesis would be on that list! As Frank Swain just tweeted, the effect of these sorts of tests on writers' work often diminishes with more data, because of the variability.

    Further context: you seem to think such stats relevant re your dyslexic son? Is my own dyslexia relevant in my writing style perhaps? I'd say I'm also a fan of Gimpyblog's writing style (and he is dyslexic too).

  5. Interesting post, I would never have considered Goldacre's pieces to be difficult to read.

  6. Very nice piece, and I agree with you. But I'm concerned about the MS Word Flesch readability algorithm; I learned a few years ago when I took a plain-language course that the program's algorithm was flawed, and despite communication with Microsoft, it hadn't been sorted. Do you know if it has, or have you tried comparing the Word result vs. hand-calculating? Tedious, I know...

  7. Really interesting post and exercise! I never realised that Goldacre's pieces were difficult to read, then again I am a scientist and I am familiar with what he is talking about. Very interesting also that the BBC article was difficult to read - surely that would be expected to be the most accessible of the articles. I don’t think that all writing needs to be accessible for everyone; it depends on who you are trying to talk to (I have blogged a little about this myself). Has anyone ever done this exercise with other specialised subjects such as finance/sport? I would predict that a similar range of ‘reading ease’ would be reflected.

  8. Here are some readability sites that let you paste in copy to see its readability score(s). Here are a couple that mash together multiple indices:


    The second one confronts you with the sentences that did the most to run up the score. They don't much like complex sentences.

    Don't know the metrics well so can't guess how these stack up to the best standards.

    My last three Neuron Culture posts (last first) scored 47.890, 47.00, and 55.37 on Flesch Reading Ease scale, and at 11.75, 11.64, and 10.5 (Flesch) grade levels.

    My last Atlantic feature: grade 12.23, reading ease 42.36. The site didn't like it. Readers — and the Best American Science Writing anthology -- did. Just saying.

    last Slate piece: grade 10.37, Reading ease 55.87

    The best piece of online writing I've read in the last few weeks was Michael Lewis's Vanity Fair piece about the Greek credit crisis. I pasted it in. He did better: Grade: 9.01 Ease: 61.62

    He's also funnier than I am. And gets paid more.

  9. This is a really fascinating piece. I am a postdoc slowly trying to transition into a career in science writing. I have been taking an informal science communication course at the University I work at with a member of the English faculty whose research is on science communication for the general public. She has given me some great insights and I hope someday to become as good a writer as some of the bloggers and writers I admire (many of whom are in this list). I am definitely open to learning about the most effective ways of communicating to the widest audience possible and adapting pieces to suit a particular audience and interesting pieces and ideas like this definitely help teach me a few things. I would never of thought about readability (definitely something I will bring up in my next writing class) and it's interesting to note the examples you have given. I definitely think that sometimes bloggers get a little caught up in criticism and witticism and though I still am able to follow as a scientist (though not always I'll admit, particularly when it comes to things outside my realm of expertise) it becomes a little hard for the general public to follow sometimes. Though I do have to say I know a great deal of non-scientists who have read Ben Goldacre's book - is this measurement based on his blog or books, as I think Ed Yong makes an excellent point that he often lets go a little more on his blog since it is mostly science folk that seem to follow them.

    Thanks for an illuminating piece.

  10. Very, very interesting. I had a look at your blog Real Science (I read the gender and math essay) and it's impressive. In particular, including vocabulary words/definitions and, most importantly, the list of questions after the post are extremely important, effective learning tools. It is quite generous of you to put the time and effort into reaching out to young people. Every academic discipline should be doing this. Of course, the other side of the coin is that education reform is badly needed (especially in the US). We need to do away with the top-down, industrial style and replace it with an environment in which students actively participate in the learning process, especially creative problem solving. At the very least, high school graduates should be culturally literate (as it is, they are barely functionally literate) and college graduates should be critically literate. Should be. Until then, you keep doing what you're doing!

  11. This is interesting, although I confess I'm one of those who has always been sceptical of readability scores.

    That to one side, and maybe it's the scientist in me (wink), but the one question that leapt out at me when reading this was "Where's the 'control'? How does this compare to other types of journalism?"

    What happens when you put some economics, current affairs or politics articles through the same filter? Without this comparison, I'm not sure what conclusions we can draw about how science journalism fares in a broader context.

    (Of course, this could open a whole can of worms - e.g. How do left-leaning political commentators compare to more conservative ones? Was unreadable business and economics journalism partly to blame for the financial crisis? I could go on . . . :-) )

  12. Sorry to have missed this until now. This is fascinating and obviously something to work on. But I'm curious. Why did you choose "if the world is going to hell..." for my representation? (Which, since it contains the phrase "ecosystem services" is unreadable even to me. How I loath jargon...)

    It seems to skew the representation here just a bit because it's a blog (yes, we have those at Scientific American). In what might seem a counter-intuitive move, my blogs tend to be more abstract than my "science journalism" because they are aimed at different audiences. A general story on, say, coal is going to be aimed at the general (Scientific American) reader. A blog is aimed at those foolish enough to care what's caught my attention. In my experience that tends to be a more specialized bunch and the writing changes to suit those with a deeper interest in, yes, the aforementioned ecosystem services.

    That said, I haven't done this kind of analysis on "news" vs. "blogs" but I'd be interested to see the results. I'm allowed to get a lot more tangled in my sentences when speaking for myself...