Are they right?
By Daniel Davis
You’ve probably read the stories. Articles in publications like the Atlantic, McKinsey, and Bloomberg saying that employees don’t want to return to the office full time. Specifically, that most would rather work from home two or three days a week.
Most of these articles are based on surveys that ask a variation of the question, “In the future, how many days per week would you prefer to work from home?”
The question looks simple and effective. But you’ve got to be careful – there’s a real science to phrasing survey questions. And as I’ll explain, this particular question has some hidden flaws that can lead to unreliable results.
Over the past six months, we’ve tested a variety of survey questions as we try to understand the post-pandemic workplace.
We’ve surveyed over 3000 people in four different countries using a variety of methods. In doing this research, we’ve found that results get skewed when you ask people how many days per week they’d prefer to work from home. In this article, I’ll detail the problems with this question and suggest three alternatives.
Problem 1: Intentions aren’t actions
One of the problems with asking people, In the future, how many days per week would you prefer to work from home? is that you’re asking about the future – you’re asking about their intentions.
People’s intentions are about as reliable as New Year’s resolutions.
People can say that they want to exercise daily and give up sugar, but generally speaking, people are pretty bad at predicting what they’ll do. Consider how many people join gyms and never go – 67% by some estimates. A person can tick a box in a survey saying that they’d prefer to spend 2-3 days working at home, but will they?
How many days a week would you prefer to work from home?
ALTERNATIVE: Ask about the present, not the future
The best survey questions focus on past actions rather than future intentions. For example, if you want to know whether someone exercises regularly, ask them whether they exercised last week rather than ask if they’re going to exercise this week.
If you want to know where people will work after the pandemic, you can look at what people are doing in places that have controlled the pandemic (places like Israel and Australia). In these regions, you don’t need to ask people to imagine where they’ll work after the pandemic – you can actually see it starting to happen. And as people in these countries continue to adjust to life without the daily threat of the pandemic, they indicate what is likely to happen elsewhere.
A couple of basic questions you could ask people in these regions:
These questions focus on past behavior, so they’ll give you a reasonably accurate picture of how people are splitting their time between the home and the office. Using this data, you’ll be able to work out where people are predominantly working.
These questions generate relatively accurate results, but they’re not perfect. Obviously it’s a leap to say that what’s happening in one country will likely occur in another. It’s a good approximation, but there are cases where you will want to hear directly from the people for whom you’re designing rather than surveying people in a different country. If that’s your situation, read on!
Problem 2: ‘Days of the week’ is a weird metric
Besides looking at past behavior, there are other ways we can improve the original survey question: In the future, how many days per week would you prefer to work from home?
The first thing to consider is the unit of measurement. For example, one of the problems with asking someone about the number of days they’d prefer to work from home is that we often don’t spend entire days at home or the office.
Before the pandemic, I worked with a colleague who would leave the office around 3 pm to pick up her children, spend the early evening being a mom, and then do a couple of hours work from home once her children were in bed. Exhausting! But how many days did she work from home? Was it five days a week because she worked from home every evening? Was it two days a week because she spent about a quarter of her overall workday at home? Was it zero days per week because she was in the office every day and never spent an entire day at home?
The truth is, all of these options work. And that’s the problem. When people answer surveys, they generally formulate an answer in their mind and then translate that answer into the options given. If the options don’t match, they make an approximation, which creates the opportunity for error to creep in.
ALTERNATIVE: A better question, a better scale
Rather than asking someone how many days they’d like to work at home, you can ask them:
Once Covid-19 is no longer a concern, where would you prefer to work?
This question eliminates the ‘days per week’ measurement and replaces it with a three-item scale that lets people indicate where they’d like to spend the bulk of their work-week: at home, at the office, or some hybrid. The three options don’t presuppose how they’ll split their time. Instead, it just asks for a general sense of where they’d like to work.
Admittedly, you don’t get the same amount of detail using a three-point scale as you would using the original six-point ‘days per week’ scale. That said, frequent public health changes mean people can’t be certain about the precise number of days they’ll spend working at home in the future, so a three-item scale is a better reflection of that underlying uncertainty.
A more detailed scale may feel more precise, but if the concept you’re measuring is fuzzy, you’re probably just collecting noisy data that gives a false sense of accuracy.
Problem 3: It is biased
The other problem with the question In the future, how many days per week would you prefer to work from home? is that the question focuses all of its attention on the home. Not only that, five of the six answers focus either on hybrid working or working from home. If you want to go back to the office full-time, there is only one option – and even then, the word ‘office’ is nowhere to be found. To go back to the office you have to say that you want to spend zero days working from home.
People responding to surveys can be pretty savvy. They pick up on subtle clues about what you’re asking and why you’re asking it. Some will even change their answer, often subconsciously, to give you what they think you want. This shift in response is sometimes called conformity bias or social desirability bias.
If someone reads a survey question that is entirely focused on working from home, they’ll get the hint; they’ll know what you’re expecting.
Even if they want to work in an office, they might feel pressure to say that they want to spend a day or more working from home because, in the context of the question, it seems more acceptable.
Fortunately, we’ve already fixed this bias problem. The earlier question – Once Covid-19 is no longer a concern, where would you prefer to work? (the second alternative) – is much more impartial. It asks about where you want to work (which is a neutral framing) rather than how much time you want to spend working at home (which is a leading framing).
In addition, the answers are more balanced because they give equal weight to the three main scenarios – working at home, working at the office, and working in a hybrid setup – and don’t unduly hint at a particular outcome.
Problem 4: The hidden cost
One final way we can improve the question In the future, how many days per week would you prefer to work from home? is by clarifying the implications of each option. In particular, people may not realize that their office will function quite differently if they spend more time working from home.
For example, if people are only coming into the office a couple of days per week, their desks will sit empty and their employer may look to implement desk sharing.
Someone ticking a box saying that they want to work from home three days a week may not realize that they’re probably also signing up for a hot desk when they do work from the office.
If they knew this, would they choose a different option?
ALTERNATIVE: Preference test
One way to make people aware of the tradeoffs between working from home and the office is to use a preference test.
In a preference test, participants are shown two options and asked to pick which one they prefer. For example, one option might be working in an office five days a week with their own desk; another option might be working wherever they like – at home or the office – on the condition that they give up their assigned desk.
Once COVID-19 is no longer a problem, which of these two workplaces would you prefer to work in?
To test more than two options, you can run the preference test a couple of times in a row, mixing the options each time. Then, using some math and probability theory, it’s relatively easy to calculate which option is the most popular.
Running the preference test is much more complicated than just asking people how many days they would like to work at home.
But the results are fascinating. When we run the preference test alongside other survey methods, we often get different results. In our surveys, many people changed their minds about their ideal workplace once they saw the details. This largely makes sense. We make tradeoffs all the time in our daily decisions, and this technique surfaces those tradeoffs when it comes to the workplace.
What should you do?
That was a lot of information about a seemingly simple question. If you’re trying to pick between the three alternatives presented, my general order of preference is:
- If you think another region or country can act as a proxy for the people that you’re studying, study what is actually happening in places coming out of the pandemic (alternative 1).
- If you still want to ask people about their future workplace preferences and have the time and ability to run a reasonably complicated survey, use a preference test (alternative 3).
- And, as a last resort, if you’re going to directly ask people where they’d prefer to work, at least make sure the question is balanced and uses a sensible scale (alternative 2).
For all of these questions, it’s essential to ask the follow-up question: Why?
You can ask people directly why they prefer a particular workplace, or you can infer it from other data you collect.
If someone tells you that they prefer to work from home, it’s not nearly as interesting as knowing someone prefers working from home because they find it easier to focus.
This extra bit of information will likely give you important clues about what people want from a workplace.
In general, it’s a positive sign that so many organizations are surveying their employees and asking them about their workplace preferences. We want decisions that are backed by data rather than intuition. And by being careful about how we ask these questions, we can be sure that our data, and therefore our actions, accurately reflect people’s preferences.