These 8 metrics predicted 70% of satisfaction at work
We analyzed what actually makes people happy at work (hint: compensation is not #1)
Our team used regression analysis to study the impact of 8 different attributes on overall job satisfaction.
At TransparentCareer we love career data. In fact, over the past few months, we’ve crowdsourced tens of thousands of highly detailed compensation, satisfaction, and culture data points representing over 6,000 companies and 3,000 job titles.
One of the areas we are most excited about is helping people choose careers where they find fulfillment. So…we took our data and analyzed which factors of a job are the largest predictors of workplace satisfaction.
This is different than most other studies of this kind. We didn’t just survey people on what they THINK makes them happy, we actually use regression analysis to see how the various attributes employees rate about their jobs actually predict their overall satisfaction in the job.
Brace yourself, this piece gets a little stats-heavy, but we promise we’ll make sure it all makes sense.
So before we get to the results, let me explain a bit about the data we collect and how we analyzed it.
On our platform we collect an employee’s assessment of their overall satisfaction with a job as we’ll as 8 different metrics of how they rate different attributes of their work environment including:
- The Brand/Prestige of the company they work for
- Their assessment of the Compensation, Benefits, and Perks of the job
- How they perceive the Quality of their Coworkers
- The Balance & Flexibility of the job
- Their Opportunity for Advancement within the company
- The opportunities for Training & Development
- The Hours Worked during a typical week
- The Percent of Time Spent Traveling
Okay, so now that you understand the data we collect, let me describe a bit about the analysis. Using about 1,700 employee job ratings we wanted to see which of the factors we collect best predicted employee satisfaction. To do this, we ran a multiple linear regression of the 8 work environment attributes and looked to see how each of them alone as well as combined was able to predict an employees overall satisfaction on the job.
If you’re not familiar with a multiple linear regression, let me just give you a high level sense for how it works (again bear with me). A multiple linear regression attempts to model the relationship between two or more explanatory variables (i.e. our work attributes) and a response variable (i.e. overall satisfaction at work) by fitting a linear equation to the observed data. More simply, it essentially asks the question, how well do a set of inputs predict an output. In our case, it generates an equation of the following form:
Satisfaction = B0+B1(Attribute 1)+B2(Attribute 2)+B3(Attribute 3)…etc
**(here the Bs are coefficients that describe how much an attribute contributes to the final prediction of the dependent variable, satisfaction in this case. A B value significantly different from zero means the attribute is predictive of satisfaction. The B0 is the intercept of the linear equation. In the results below, the left most column are the attributes and the first column of numbers are the coefficients. )**
Now are you ready for the results? Drum roll please…
Okay, okay I know this looks like some gobbledegook, but here’s the order of the factors that were most predictive of job satisfaction from the most to the least (this is based on how far the numbers in the first column are from zero):
- Quality of coworkers
- Balance & flexibility
- Compensation & benefits
- Training & skill development
- Brand prestige of the firm
- Opportunities for advancement within the company
- Hours worked per week
- Percent of travel per year (no real correlation)
If you’re interested in understanding how to read the raw output, let me point you to the two most important things here.
If you look at the first column of numbers, that shows how a one point increase in a certain metric contributes to an increase in overall job satisfaction. For instance, quality of coworkers has a coefficient of 0.35. This means that a 1-point increase in coworker quality on our scale translates to a 0.35 point increase in overall satisfaction ratings. So the higher the coefficient, the more that metric contributes to predicting overall satisfaction.
The second thing to look at here is the number of stars on the right — these show how unlikely it is that these results can be caused by random chance. Three stars says that for the attribute in question there is a 0.1% probability that a metric is not actually correlated with overall satisfaction. As you can see nearly all the attributes qualify for this level of significance. Finally, overall, the combination of these 8 attributes has an r-squared value of about 0.7. This means that these 8 attributes can explain 70% of what determines overall satisfaction at work. Pretty cool!
This means that these 8 metrics can explain 70% of what determines overall satisfaction at work.
Okay, so now that some of the hardcore stats is out of the way, let me tell you about what I find most interesting in this data. The top metrics actually share something in common, which I think says a lot about how people are actually fulfilled at work.
My interpretation is that the attributes that are the best predictors of satisfaction are actually the ones that people have to deal with on a daily basis. Take quality of co-workers, for instance. You spend every day with these people. If you have a boss you hate or the people around you don’t inspire you to do good work, then no matter what, its going to be hard for you to enjoy going to work. The opposite can be true too. Having incredible people around you can make some of the most boring work fun and worth waking up in the morning for.
On the other hand, attributes that are more abstract or dealt with less frequently are lower predictors. Take brand prestige, training, or opportunities for advancement within the firm. My hunch is that some of these are big reasons to join a firm, but since they don’t impact how you feel on a day-to-day basis, they actually are less important for overall satisfaction.
One last thing — I was surprised that hours worked or the amount of travel weren’t really predictors, while the balance and flexibility of the job was a very high predictor. These seem to be at odds. My explanation would be that its not really about how much you’re working or traveling, but really how much control you have that determines satisfaction. You can work a lot, but be happy, but if you have no control over your ability to balance work with other aspects of your life, then dissatisfaction sets in.
Obviously, this is just one analysis: there are nuances to each of these attributes and their collection, as well as other attributes that we don’t even capture. That being said, I do think these learnings have significant implications for both job seekers and employers.
For job-seekers, many prioritize surface level factors when making a decision — like the “brand name” or the compensation package. While these matter, many fail to truly assess how well they will enjoy working with their future co-workers. For employers, many think that pay raises may be the best way to retain employees. Instead, focusing on policies that give people more flexibility and control over their lives or that create cohesive and high-performing cultures may be much more impactful.
We appreciate you reading this and hope at the very least it has made you think about what is most important to you.
As always, on TransparentCareer you can get company-specific data on compensation and culture, personalized to your unique background. On our end, we will continue improving so that we can give you the best platform for evaluating your career options, and (hopefully) help you find a job at which you can feel truly satisfied.
Thank you for reading,
The TransparentCareer Team