Posted by Ray Poynter, 17 December 2022
NewMR and The Future Place have recently launched the first iteration of Word Cloud Plus – a better way to create word clouds. In this post, I show how I am using Word Cloud Plus to look at open-ended comments from surveys.
The data for this example comes from a project run by NewMR a few years ago and relate to the elements that people felt contributed to a good presentation. You can access a copy of the open-ended responses by clicking here.
Pre-processing the text
Before pasting the text file into Word Cloud Plus, I did a very quick tidying of the file. I loaded the file into a text processor with a spelling/grammar checker. In the text processor, I standardised the spelling in terms of the difference between UK and USA preferences (e.g. humor and humour), some obvious typos, and I identified some cases where the survey platform had not put a space between two words. I also removed any specific references to named individuals. This was a very light touch edit, not good enough to be sure I caught everything, but good enough to help Word Cloud Plus to find the key words and phrases.
A first approximation
I ran Word Cloud Plus with its default setting, for example, 50 terms, using the Combination Counting algorithm, with a max phrase length of 4 words per phrase.
The big two elements are the content and the slides, closely followed by the presentation, emphasizing the deliverable itself. The next two elements are voice and audience.
If I click on ‘audience’, I can see the sorts of phrases that use the term, for example: “the ability to engage the audience”, “a sense of humour is a great help as is an appreciation of the practical applications that the audience may have for their topic”, “ability to engage members of the audience and ‘make sense’”, “ability to hold the attention of the audience through their voice”, “the ppts should be crisp and to the point need to consider the type of audience and their levels of familiarity with the subject”.
To make the cloud clearer, I can make all the audience and engage phrases the same colour and all of the voice-related phrases another, as in the image below.
The coherence can be improved by moving the elements to make similar items closer together. As in the updated image below.
Of course, there is no limit to how far one could keep digging into the data, both in terms of refining messages, such as those about visual content, but also in terms of removing things that seem less meaningful.
Looking at the Summary
As well as the visual representation, Word Cloud Plus offers you to the chance to look at the underlying counts and styles, as in the table below.
Tweaking the Stop Words
By telling Word Cloud Plus to treat the word ‘ability’ as a stop word, I think we get a better picture.
By moving a few more terms, by replacing the word ‘slide’ with ‘slides’ and by adding ‘good’, ‘great’, and ‘make’ to the list of stop words, I think we get a better sense of the breadth of things that different people lead to a presenter being great. And a reminder that different people value different things.