Is It Time To Change Engagement Surveys?
By Mark Murphy
Engagement surveys are like bread and butter to HRD departments – they’re the go-to staple for finding out the health of people and the organization.
But while conducting one is often the first port of call, the reality – according to a recent report from Leadership IQ – is that just 22% of organizations claim to see ‘good results’ from their them.
By ‘good results’ we mean progress. For instance, either survey scores were low, and they’ve improved significantly, or the scores were high, and they’ve remained that way. If survey scores aren’t improving or they’re declining, we can safely say that the survey isn’t generating good results.
Now there are plenty of reasons why an employee survey can fail to generate results, including being filled with poorly worded questions, or having processes in place that prevent the results from being shared with employees in the first place. But a lesser-known impediment can be just as damaging, and it’s the use of the five-point scale.
The curse of the five-point scale
The five-point scale is without doubt the most common scale employed in engagement surveys. Think of the classic Likert scale, ranging from Strongly Agree to Strongly Disagree and everything in-between.
Granted, it’s a common and effective scale in a wide variety of situations. But this scale has a major problem when it’s used for employee surveys.
The problem of skewed results
The five-point scale works best when there’s a reasonable chance the data derived from it will be normally distributed (ie as many people will rate 1’s as they do 5’s). For example, if you survey 1,000 random people and ask them to rate the statement, “America should have universal healthcare,” you will get roughly as many 1’s as 5’s.
Normal distribution is important because it allows the people looking at the data to easily track changes in opinion. For instance, if running an ad campaign pitching universal healthcare, researchers would want to know if they’d successfully moved any “strongly disagree” respondents into “disagree” or “neutral.”
But when it comes to employee surveys, five-point survey results tend to be far more skewed. Moreover, they evidence much less variability than the general public. In other words, it’s highly likely HRDs won’t see very many 1’s and 2’s, but the will have an abundance of 4’s and 5’s.
Employee surveys aren’t random surveys
Just think about the implications of this. If HRDs survey their employees with the question, “I recommend this company as a great organization to work for,” they’re just not going to have many employees strongly disagree or disagree.
Why? Well, even if people are frustrated or disappointed with their employer, they still come to work everyday and collect regular paychecks.
In other words, employee surveys are, by definition, not random samples because every single respondent is collecting money from, and choosing to remain at, the company.
Of course, there will be a few low scores, but nothing resembling a normal distribution of data. (Note: You can test everything I just wrote by running histograms on some of your big outcome questions).
Five point scales become three point scales
The upshot of all this, is that in an employee survey, the five-point scale de facto becomes a three-point scale. There will be so few low scores that the survey choices effectively become “Everything is awful,” or “Everything is wonderful,” or “Things are just okay.”
The problem with this, is that HRDs won’t be able to see subtle movements in their scores.
What this means on a practical level
Imagine an HRDs implements an employee recognition program in order to improve overall employee engagement. Even if it’s wonderfully successful, it’s probably not going to move someone from “Things are just okay” to “Everything is wonderful.”
It will positively impact employees’ engagement, but when data analysts are working with a de facto three-point scale, these ratings are far too blunt to pick up subtle improvements.
You need to pick up subtleties better
Shifts in workplace sentiment are typically subtle. And with a five-point scale (which is really a three-point scale), most surveys won’t be nearly precise enough to discern any actual improvements.
The solution is a seven-point scale, especially one that ranges from Never to Always. There will still be skewing of the data (ie there still won’t be many 1’s and 2’s), but at least HRDs will now have more data points with which to see subtle improvements and track progress.
Sadly, we know from the quiz, “How Good Is Your Employee Engagement Survey?,” more than half of companies are using five-point scales on their surveys.
The disappointing aspect is that while companies are often taking positive actions to improve their employee engagement, many of those improvements aren’t showing up in employee survey scores because the scale is too blunt.