The dog that didn’t bark – a great way to find insight in information
Posted by Ray Poynter, 10 May 2018
In the Sherlock Holmes story ‘The Adventure of Silver Blaze’ the key to finding the story (i.e. the person who committed the crime) is the curious incident of the dog in the night. When the horse (the Silver Blaze) was being stolen, the dog in the stable yard did not bark, and that was what was curious. This clue led Holmes to deduce that the theft was an inside job; conducted by somebody the dog already knew.
Over the years, I have found the ‘dog that did not bark’ idea to be a useful tool when trying to find key messages and stories from a set of information. Thinking about what has not been said by research participants can be as revealing as what has been said.
When analyzing the data, ask yourself what is not there? Consider the following examples:
- You read some references for a new employee, and they all stress team player, effort, innovative and cheerfulness. But none of the references address quality of work or accuracy – so perhaps the person’s work is not great – something you need to check.
- You look at a set of NPS data for banks, comprising scores and open-ends. You notice that for most brands about 20% of the positive comments relate to clarity of communications – but for your client, this topic is hardly mentioned – a topic to be investigated.
- You run a cluster analysis and produce four neat clusters. Try looking at how many people are not well suited to the clusters they are in and how many have been flagged as outliers – there might be something interesting going on.
- You show a set of concepts to people and 10% select ‘None of these’ as their preferred option. Explore this group, maybe they are saying something important, or maybe they should not have been recruited in the first place.
- With a set of Agree/Disagree data, look at how many pick ‘Neither Agree Nor Disagree’, how does this change for the different scales you have used? Are these neutral scores (people who are not barking) spread evenly amongst the sample, or are some people systematically less likely express a position?
Jon Puleston, Vice President of Innovation at Lightspeed Research has published widely on the topic of improving questionnaires and research. Jon has suggested using this approach in questionnaire design exercises to get a clearer view about who has an opinion (positive or negative) and who does not. One of his examples relates to tidying, where he explores ‘I am untidy’ (something that only a minority select) and ‘I enjoy keeping my place tidy’ (again only a minority select this) – which identifies that most people do not bark, in this case because they do not feel strongly either way. Jon has also shown the power of the unasked/unanswered question with topics like driving over the speed limit. If you ask people if they exceed the speed limit when driving, many people are reluctant to answer truthfully. However, Jon has found that by asking “I get anxious about travelling over the speed limit.” there is a large overlap between people who do not agree with this statement and who drive over the speed limit.
If you enjoyed this post, why not sign up to the NewMR newsletter to make sure you do not miss out on future articles?