Data Visualisation

CEO & Founder

nick.jpg

# ...or why I am not a fan of word clouds

One of the questions that occasionally we get asked is how to present data from unstructured free text. And sometimes then we get asked, "Do you do a word cloud?"

To be honest we get asked this less and less these days, and I think as people get more used to data visualisation we could see this end completely.

It is well known, to those who know me well, that I am not and never have been a fan of word clouds, but why? and also why do people want to use them?

## Word Clouds are "the Mullets of the Internet"

I do love this, I am biased I know, but I think it is a fab description from 2005 of the faddish adoption of word clouds (tag clouds) back in the day.

## What are word clouds?

Word clouds are simply a pictorial representation of words that are taken from unstructured (or structured) text grouped together, and typically larger or smaller depending on the number of times the word is used.

## Whats my issue with word clouds?

If we think about the purpose of data visualisation. Data visualisation takes complex data and turns this into visuals that allow the user to see the information they need quickly, to spot trends, to see data represented in different ways and to tell a story about the data.

### They don't really tell a story

Telling a story is an important part of data visualisation and I am afraid word clouds simply do not do this, they strip out the sentences, the narrative and try to replace this rich (unstructured) data with "tags" based on how often a single word is said. I will demonstrate with a bad example; Macbeth, a classic:

> Methought I heard a voice cry, ‘Sleep no more! Macbeth does murder sleep: the innocent sleep, Sleep that knits up the ravelled sleeve of care, The death of each day's life, sore labor's bath, Balm of hurt minds, great nature's second course, Chief nourisher in life's feast.
 
The book is complex, with lots going on and many different viewpoints and narratives, understanding this requires that you read the book or at least a detailed synopsis of the major themes.


**Or could you just read the word cloud...**

![An example of a word cloud for the play Macbeth](/uploads/f0f5b5e92a4a420c8a308d2af80a49f5.webp)

Are we surprised that the main word used in Macbeth, a story about Macbeth is "Macbeth"?

### They provide no context to the sentiment in the text

Words are just displayed with no background, no context of how the person felt at the time.

We will often use the same words when we are asked a question that are in the question, if you ask someone about Teamwork people may say "we are great at teamwork" or "teamwork could be improved" or "teamwork is ok"

So in the word cloud we will see "Teamwork" dominating but all the other words will be based on the individual - great, awesome, fab, good, ok, done well and amazing all suggest good things, you just won't get that in a word cloud.

### Their design means you see something that potentially is not there

Some words are bigger than other words because they contain more letters - sounds obvious right?

If a short word has a large count and a long word a medium-sized count you run the risk that you lose the scale between the words simply because the word is bigger!

Look at how "CadburysDairyMilk" is so much larger than "Milka" or "Trident" below - are you confident that this is because it is statistically larger or just because it has more letters.

![chocolate brands word cloud](/uploads/d68ebf0b3c534a9490e9ef304f227cad.webp)

## They are not data science - you cannot see the facts

Here is a word cloud representing towns in the UK by population size, what does this tell us?

![word cloud representing uk towns by population](/uploads/99d6bf480909493cb0ef033fae29e8a9.webp)

We can see that Liverpool and Sheffield are the largest here, but we instantly lose any real frames of reference.

- How much larger is the population of Liverpool than Sheffield?
- What about Hull? is it the same as Derby?

> Every time I see a word cloud presented as insight, I die a little inside
 
Quote from Jacob Harris from the NY Times

## So, why do people use word clouds?

I think it is because of the following:

1. Unstructured text takes a long time to read so people want a fast solution
2. Word clouds were everywhere once and therefore well known and associated with a fast solution
3. Most people do not think about data visualisation issues
4. We all have biases that things like word clouds play on

Let's look at those biases and how we respond to word clouds:

### I do not have enough meaning

We have already said that word clouds are sparse when it comes to actual meaning so what happens - "We tend to find stories and patterns even when looking at sparse data"

Some of the biases at play:

[Clustering Illusion](https://en.wikipedia.org/wiki/Clustering_illusion) - The clustering illusion is the tendency to erroneously consider the inevitable "streaks" or "clusters" arising in small samples from random distributions to be non-random

(Illusory correlation)[https://en.wikipedia.org/wiki/Illusory_correlation] -  In psychology, illusory correlation is the phenomenon of perceiving a relationship between variables even when no such relationship exists

###  have too much information

Free text by it's nature is a lot of information so what happens? - "Visually striking or bizarre things stick out more than non-bizarre, less striking things" Word clouds are designed to be visually striking

Some of the biases at play:

Von Restorff effect - The Von Restorff effect, also known as the "isolation effect", predicts that when multiple homogeneous stimuli are presented, the stimulus that differs from the rest is more likely to be remembered

### I need to act fast

We have already said that in business we default to needing to act quickly, reading lots of free text is time-consuming.

This is one of the reasons word clouds are chosen because - **"We favour simple-looking options over complex ones"**

## So, what is better than word clouds?

Hopefully, by now you are in agreement that we do not need a word cloud in our life if we are looking for insights into any large unstructured free text,

So, if not word clouds then what?

In our experience, while there are software tools that can help you make sense of unstructured qualitative data, we've yet to find anything that works just like that. As part of the attraction of world clouds is their convenience, it's also true that even the best tools provide more accurate insights when you put some effort into coding the text.


Check out our other blogs on {data visualisation}(https://pxhub.io/blog/category/data-visualisation)


## The People Experience Hub

A tech-led, human approach to measuring and improving [people experience](https://pxhub.io) at work, helping you understand your people better.

Head in the clouds.png

We're often asked is how to present data from unstructured free text. And sometimes then we get asked, "Do you do a word cloud?". And here's why we don't!

Is Your Head in the Clouds? Why We Don't Use Word Clouds

Is Your Head In The Clouds?

Director of People Science

THJBHHNRZ-U019UJNULH5-54f1615d8d15-192.jpg

When it comes to Employee Surveys, response rates matter. It doesn’t matter whether you’re talking about more traditional (annual) [employee surveys](/annual-engagement-surveys) or [pulse](/pulse-surveys) solutions, there’s a very simple equation at play. Higher response rates provide more reliable results.

This blog will provide you with some advice to help you ‘nudge’ your people to give their feedback that’s rooted in behavioural science, rather than “what’s worked for me”. This is because specific tactics that work in one organisation might not work so well in another.

## But, what is a good survey response rate?

Many different sources suggest that for an annual survey, above 70% is good (though it still feels kind of arbitrary), but for other forms of [employee feedback](/employee-feedback) such as pulse, you should expect less. The truth, however, is rather more nuanced and contextual. I’ve worked in very different large organisations where response rates were typically over 90% and the reasons for that are rooted in culture, past experiences with employee surveys, whether they were perceived as important and, crucially, line manager support.

### Why are response rates different for different types of survey?

There’s a lot of debate about whether annual engagement surveys are dead and should be replaced by Pulse or other forms of continuous feedback, but we’re cautious about taking a binary point of view on this. While the trend is towards Pulse, the reliability that comes with higher response rates is a good argument for annual surveys. That’s why - [like Josh Bersin] (https://joshbersin.com/2020/08/is-the-employee-survey-dead-nope-its-becoming-smarter-by-the-minute/) - we believe that annual surveys have their place in the 'employee feedback mix’ and shouldn’t be ruled out.

At [Facebook](https://hbr.org/2018/03/employee-surveys-are-still-one-of-the-best-ways-to-measure-engagement), where 95% of employees fill out their engagement survey and 61% provide written comments, they believe that the survey is seen as something that’s important and is taken seriously as an opportunity to be heard. That’s certainly food for thought.

## How to boost your survey response rates with Behavioural Science

Simply put, responding to an employee survey is – and must be – an act of volition. If not, we would fully expect employees to find their own way of exercising autonomy by spoiling the data. So let’s rule out coercion as a viable tactic up front! How can we use behavioural science to encourage people to participate in surveys, without feeling undue pressure?

## The COM-B System of Behaviour 

One of the most widely accepted developments in behavioural science is the [COM-B system of behaviour](https://implementationscience.biomedcentral.com/articles/10.1186/1748-5908-6-42) (developed by Michie and colleagues - image below). COM-B which represents the Capability, Opportunity and Motivation behind Behaviour, was originally developed in the health industry but is now having a much broader influence. 

![The COM-B System of Behaviour can be used to improve employee survey response rates with behavioural science](/uploads/38621d8a04064dbea25bd513aa81465b.jpg)
 
## So how can COM-B help improve survey response rates?

Let’s look at each of the components of COM-B in turn. A nice metaphor for COM-B comes from detective stories. We all know that the suspect must have had the means (capability), motive and opportunity to have committed the crime…

### Capability 
While it’s easy to dismiss capability as a factor in survey responses, because of course you use survey tools that are designed with the user experience in mind, and we are now in an age of near-universal digital literacy (at least to the level required to respond to a survey), let’s not overlook it entirely.

If you started with the following assumptions, you might not go amiss:
- Not everyone knows that the employee survey is important, so you need to make sure that they know why it is
- Your people are probably stretched; and so you are competing for their attention
- Even if they intend to respond, they might forget
- They might not feel that they have control over how they spend their time

So while your people may not need to be educated on how to respond to surveys, communication is critical. They need to know that it’s important, that they have permission to spend time doing it, the instructions need to be clear and they may well need a reminder. Use multiple channels, for example, email, posters, intranet, video – and if face-to-face communication comes from their line manager that ticks multiple boxes. And if you have a manual workforce, literacy could still prove to be a barrier.

Additionally, your workforce is neurodiverse and therefore some employees may find it difficult to respond to surveys that are over-reliant on colour, particular fonts or, for example visual ‘sliders’. These issues arguably take us into the territory of opportunity because your survey design could be a barrier to participation.

### Opportunity
 
Opportunity can be categorised as physical or social, and so too can potential barriers.

The barriers that a dyslexic employee (for example) might experience may be largely physical – for example, fonts that are difficult to process, but that’s not always the case because supportive colleagues, for example, create social opportunity.

Physical opportunity might include the ready availability of technology, which might not just mean being provided with access to a device, but with enough proximity or indeed numbers to make it practical to use them.

Social opportunity is about building norms, making it acceptable or desirable to respond through modelling. It’s one thing that managers complete the employee survey as individuals, but it’s another that they share with their teams that they’re doing it on company time.

If line managers are key to survey response rates, should they be given a completion target for their team? This is a tricky question, to which many HR people would automatically respond “no”, but I’m not sure. There are two things to consider:
1. The potentially negative impact of incentivisation on motivation in the longer-term (which we’ll come to) 
2. The potential for (possibly well-meaning) but inappropriate behaviour – hectoring or even worse.

I still wouldn’t rule out giving line managers a target, particularly in an environment with very low participation rates, but I also wouldn’t encourage it. There will always be a trade-off between the data gained, possibly the data quality (if people feel that they are being coerced), but more specifically the undermining of motivation to participate.

### Motivation

When it comes to motivation, the intrinsic motivation to act is broadly considered to be desirable. When a person chooses to engage in behaviour without the promise of a reward (or punishment), or in other words the activity is the reward in itself, then they are said to be intrinsically motivated. When they are given a ‘contingent’ reward for performing the behaviour they are extrinsically motivated.

Research has shown that when you directly reward things that people would happily do anyway, they can become less motivated to do them because their intrinsic motivation (which is more powerful) is reduced.

According to [Self-Determination Theory](https://positivepsychology.com/self-determination-theory/) (Deci & Ryan, 1985), intrinsic motivation is driven by three basic needs: 

1. Competence: The need to feel that I am effective
2. Autonomy: The need to feel a sense of control 
3. Relatedness: The need connect with / have relationships with others

![Self-determination theory is a useful framework for improving survey response rates](/uploads/32de0f9908f148579aef8cd2db865c52.png)
 
#### Enhancing intrinsic motivation to participate

These three needs give some powerful clues to improving employee survey rates:

First, the need for autonomy emphasises the importance of giving people the free choice to participate (or not). It also highlights the importance of acting on results and emphasising that the survey will be used to create change. It represents one way of influencing the organisation, its culture, and its management practices.

The opportunity to influence their working environment is fundamental to driving employee survey response rates, but the flip side is that when the results are not acted upon it builds a sense of helplessness and the broken ‘contract’ promotes cynicism.

Second, the need for competence suggests that you might emphasise that the survey will help to identify where people’s need to develop, grow or progress could be improved. It might be used to improve access to the tools for the job, processes that make it easier to succeed. You might also appeal to the desire to ‘win’ collectively, as an organisation.

Finally, the need for relatedness might mean that people will respond well to opportunities to improve collaboration and trust in the organisation. It might be effective to remind people that it is not only their opportunity to speak their mind but to help improve the experience of their colleague too. And of course, if you can connect the annual survey to the broader purpose of the organisation, that will help.

### History Matters

COM-B also distinguishes between ‘automatic’ and ‘reflective’ motivation. Automatic motivation is about the immediate emotions and mechanisms such as incentives and rewards that may be intrinsic or extrinsic. Reflective motivation is more nuanced and learned over time – and includes things like our beliefs, values and identities. 

It’s therefore critical when thinking about all of the factors that might boost employee survey response rates, to consider the historical and cultural context. While it’s easy to advise about what tactics to use to improve response rates, it’s harder to do it, because you have to consider what will work here and now. 

How might our cultural values impact upon survey responses, for example, our relative openness or desire for privacy? Do our people tend to conform? How individualistic is the culture, and so on…

You may have to overcome cynicism and mistrust because of what’s happened in the past, or you might inherit a situation where people are willing to give their views freely because surveys have been well-managed and have demonstrably led to change.

That’s why we work closely with our clients to help them develop their plan for maximising participation in their employee survey and, by implication, driving more insightful results and meaningful change for their people.


Response Rates.png

When it comes to your employee survey, response rates mean more reliable results. So how can behavioural Science help you improve participation?

Boost Your Employee Survey Response Rates with Behavioural Science

When it comes to Employee Surveys, response rates matter. Whether that's more traditional annual surveys or pulse solutions, there’s a very simple equation at play. Higher response rates provide more reliable results. So how can behavioural Science help you improve your survey reponse rates?