Navigation
Search
New Comments
Handy Links
Syndication
« The Nihilist's Job Resume | Main | Hiatus »
Economists Examine the Costs of Job Satisfaction
On April 11, 2006 by Jamie Madigan
I'm always fascinated by the varied approaches that different branches of science tackle the same problems. There's an interesting report on research done by economists (found via Strategic HR Lawyer) that lays out some of the drivers of job satisfaction and their value relative to increases in pay. This isn't directly related to selection and assessment, but job satisfaction and tenure are often criteria that we're interested in when deciding if a selection system has value.
Here's the meat of the piece:
Trust in management is by far the biggest component to consider. Say you get a new boss and your trust in management goes up a bit at your job (say, up one point on a 10-point scale). That's like getting a 36 percent pay raise, Helliwell and Huang calculate.In other words, that increased level of trust will boost your level of overall satisfaction in life by about the same amount as a 36 percent raise would.
Conversely, if you lose some trust in management, the decline in your job satisfaction is like taking a 36 percent pay cut.
Having a job that offers a lot of variety in projects, Helliwell and Huang found, is the equivalent of a 21 percent hike in pay.
Having a position that requires a high level of skill is the equivalent of a 19 percent raise.
Interesting stuff. I'd love to see the original research to see how it compares to how psychologists typically go about looking at job satisfaction --self-report data on ordinal scales, measuring theoretical constructs.
Existing comments:
Posted by Bryan at April 11, 2006 3:09 PM:
Working paper here:
http://www.econ.ubc.ca/helliwell//papers/NBERw11759.pdf
Posted by Stephen at April 13, 2006 10:35 AM:
The CNNMoney.com article is based on a very dense research paper written by two economists who relentlessly use (some would say abuse) multiple regression of survey data to arrive at causal explanations. The cure for this methodological affliction was given in Pedhazur's book, Multiple Regression in Behavioral Research: Explanation and Prediction. This is why structural equation modeling was developed -- to handle many of the shortcomings of using multiple regression with observational data to support causal conclusions. Not to say that SEM isn't tricky.
The basic idea behind the research was to throw dozens of independent variables into an equation to see if any of them could explain Life Satisfaction after paritalling out the effects of confounding variables (health status, age, marital status, income, etc., ad nauseum). These statistical control techniques are questionable when it comes to drawing causal conclusions. Leaving out even a single critical independent variable can cause the model to crash and burn (referred to as L.O.V.E. or left-out-variable-error). Were the models misspecified? Look at it this way, after cramming over three dozen variables into the model, the authors were only able to explain 50% of the variance in Life Satisfaction. This type of analysis is what I call C.R.A.P. -- Can't Really Answer the Problem.
Posted by Jamie at April 13, 2006 10:44 AM:
Heh. CRAP. :)
Yeah, I get multiple regression and from what little I've read of economics it's a favorite among those kinds of researchers. General unweildiness aside, why haven't they adopted structural equation modeling or path analysis? It's advanced stuff requiring mountains of data, but it seems that studies like this aren't suffering from small Ns.
It generally seems, though, that economists aren't into theory building and hypothesis/model testing the way other disciplines, like psychology, are. Or maybe that's just my ignorance talking.
Post your own comment:
