Thursday 15 May 2014

How to Choose Between Job Offers Using Science

Suppose you get a phone call one evening about a great job offer from the CEO of your dream company. Just as you finish the conversation with a promise to respond with an answer, another call comes in.
It's your former mentor. He'd like you to join the executive team of his partnership: there you'd name your position, have a great salary and choose your perks. Again, promising to get back to him promptly, you hang up.
Strolling into work the next morning, you're greeted by your company's Chairman. She announces that the board wants to make you the next CEO once the existing CEO retires in a few weeks. It's the promotion you never expected but always hoped for. You're so overwhelmed you ask for a minute to mentally check whether you're, in fact, still alive.
Back in your office, your head's racing but then suddenly it settles on one, horrifying question: "How will I decide?"
Perhaps your gut reaction is to quickly jot down a list of all the pro's and con's of each job offer. You might even make a top 10 list of the most important features of each, then rank them and choose. Unfortunately, science shows that such approaches are often riddled with mental traps.
Focusing too much on one major feature could lead you to overlook a number of awful surprises in the making. That's a cognitive bias called, "focalism" or "anchoring".
In the other extreme, considering too many features in your analysis could lead you to overweight unimportant factors (such as the size of your office window) and underweight important factors (such as career trajectory). That's a mental bias called the "paradox of choice".
In the worst case scenario, the paradox of choice can even lead you to a state of absolute confusion and perplexity: where you delay your decision altogether and miss out on one or all of the opportunities.
What then can you do? Fortunately, decision science has an interesting tool that can help. It's called, "stochastic dominance".
Sexy sounding enough to be used in a pickup line, yet powerful enough to cut through the cognitive biases just mentioned, stochastic dominance can help you select the opportunity that's "most likely" the better option, considering what you know at the time you have to make the decision.
Suppose, for example, that the three most important features to you in a job are Salary, Career Trajectory and Location. On the surface, comparing jobs based upon these criteria seems easy. Take the one with the highest salary, the best position and the most ideal location, right?
Not! In fact, all three of these features are typically subject to numerous uncertainties. Salary may fluctuate based upon the company's performance, not yours. Career Trajectory may be held up by internal politics and Location may be too far from civilisation (i.e., no Whole Foods).
This is where stochastic dominance helps. Using it, you would evaluate each job's features taking uncertainty into account.
Here's how it works in three steps:
Step 1. List the key features/dimensions that a given job opportunity 'must-have' for you to take it seriously.
As I said above, these could be Salary, Career Trajectory, Location, etc.
Step 2. For each job opportunity, score (on a scale of 1 to 5, say) the features in relation to that particular job. Then compare each job, side-by-side, using the scores.
For example, Job A might be with an established company that pays a regular (i.e., almost certain) bonus. If the Job A's base salary is already high (and you expect the bonus to be good), you'll give Job A a score of "5" in the Salary dimension.
Job B, however, is with a startup. The Salary base is good, but the bonus could either be very high, or very low. Net-net, you expect the Salary + bonus to be a bit less with Job B, so you give Job B a score of "4" in the Salary dimension. Comparing the two, Job A stochastically dominates Job B in Salary terms.
Proceed like this in scoring and comparing all the other must-have features (i.e., Career Trajectory, Location, etc.).
Step 3. Choose the job that clearly dominates in at least one dimension and is at least as good as the others in terms of the other dimensions.
Job A (which, again, dominates Job B in Salary terms) is stochastically dominant overall IF it scores at least as high as Job B in all the other dimensions (i.e., Career Trajectory and Location). If that's the case, you choose Job A!
Another way of saying this is: choose the job that dominates all others in at least one important feature and where you're no worse off in any other dimension.
Sexy, right? There's just a few catches. Ties occur if Job A is dominant in one feature and Job B is dominant in another. Break the tie by flipping a coin, staying put at your current job or ranking the features, comparing the scores again and taking the job that dominates in the highest ranked feature.
A second catch relates to how you score. Use the same method of scoring each feature across jobs. If you use Glassdoor to assess salary for one job, don't use your buddy's drunken opinion for the other.
A third catch relates to how much you personally like to take risks. Thus far, I've only discussed "first-order stochastic dominance". However, if you're one of those people who really hates the unknown, then "second order stochastic dominance" (SOD) is more appropriate.
SOD would suggest you take the job with less uncertainty and an acceptable, overall, anticipated experience. Hence, take Job A if the features it promises are more certain or more clear to you and if you anticipate situations where you might be worse off at Job B. (Of course, you may kick yourself for being so risk averse later…)
A fourth and final catch relates to the number of dimensions you consider in scoring. You can use as many as you like, but you should focus on those absolute, MUST-HAVES. Going too much beyond this brings you back to the paradox of choice.
Admittedly, stochastic dominance is still a geeky way to choose between jobs. However, it can save you errors in thinking incorrectly about such an important decision.

No comments:

Post a Comment