You may not realise it but you are signing legal agreements all the time. Think of all those “terms and conditions” boxes you tick when you buy new software or travel insurance. How many times have you tried to read these and not really understood that they were saying?
You’d not be alone in that. Our research outlines how insurance policies are incredibly difficult to understand. So difficult that you need a PhD to understand them.
Yet consumer law requires legal agreements to be transparent. This means that they need to be in plain, easy to understand language. If legal documents are not written in plain English, then they may not be legally binding – or at least the parts that are not transparent won’t be. So does this apply to those contracts you sign on a regular basis? And, if so, how do we measure how readable something is?
When it comes to assessing whether terms and conditions in insurance policies are written in a transparent way, we analysed seven policies using an array of different readability measures. Since the 1920s a lot of work has been done developing formulae to evaluate the difficulty of a text. These measures generally gauge readability based on the difficulty of a text’s words and the length of its sentences. Over the years, hundreds of different readability measures have been created for English and other languages.
Finding the right measure
The choice of measures is complicated by the fact that different measures calculate readability in different ways. But broadly speaking the measures that we used are based on the number of words in a sentence, as well as the number of syllables or number of letters in the words. If a document has lots of words with multiple syllables and sentences with more words, then its readability will be less than one that has words with fewer syllables and shorter sentences.
There are lots of online tools to calculate readability. Even tools that try to produce the same measure, like the Flesch-Kincaid reading scores, can have significant discrepancies. One of the reasons for this is how the software counts number of words and syllables. For example, does the program count hi-tech as one or two words? Different programs treat these differently. Also, the syllable structure of English is quite complex, which makes it hard for computers to count syllables and there are differences between calculators.
To deal with this variability, and not prioritise one method over another, we used six freely available tools that automatically calculated five common readability measures. To accommodate for the differences, we pooled the various estimates and took an average to assess how readable the exclusion clauses – or “small print” – for typical insurance policies are. Here’s what we found:
All of the policies we tested needed a very high level of education to be fully understood. The most readable policy required almost 14 years of education (high school plus one year of university), while the least readable needed 19 years (PhD level). This suggests that at least some parts of these policies could be challenged on grounds of their transparency – no matter whether they are fair or not. This has important implications for consumers.
Based on these common measures, a reasonable conclusion might be that these policies are not written in clear and plain English. Not so in the UK. Readability scores do not have any legal effect in the UK – there isn’t a target score to beat.
This is because European case law requires courts to consider whether the contract clearly communicates its effects. A reading score may be good evidence that the effects cannot be understood by a consumer, but are not seen as the determining factor. Even a contract with a low reading score might not be clear. Courts are therefore required to make an assessment based on all elements of the contract, and consider whether it is effective in communicating its contents to a reader.
In the US, however, there is a trend towards using reading scores to assess contracts. In Texas, for example, consumer banking contracts have to meet prescribed Flesch-Kincaid reading scores calculated by Microsoft Word. Similarly, in South Carolina loan contracts have a Flesch-Kincaid score of no higher than seventh grade.
With the broadened coverage of health insurance under the Affordable Care Act, states have given increased consideration to the readability of these policies, often using reading scores as a tool to assess contractual clarity. Certainly the insurance policies we looked at would not meet the thresholds we see in these laws.
But while the trend in the US is towards greater use of reading scores, the role of these scores in the UK remains unclear. The chancellor of the exchequer announced a crackdown on confusing small print in his 2017 budget – reading scores could play a role in implementing this.
Meanwhile, businesses could use readability scores to self assess their contracts, and go back to the drawing board if the reading score is too high. Courts and regulators could use them to screen contracts, identify those that are unlikely to be understood by the average consumer and take action to ensure changes are made.
Of course, there should also be vigilance to ensure unclear contracts do not simply game the reading score system. But some use of reading scores could produce real benefits to consumers by helping to make the contracts that they enter into easier to understand.