Interview Success Simulations
Now that I’m in the thick of the job search, it’s become clear to me that the job search generally comes in two stages: 1) An initial video interview (oftentimes Skype) where you meet the search committee if they’re interested but not completely sure about you, and 2) An in-person interview where they are really trying to choose between a small set of bona-fide candidates.
I became rather curious about the ballpark optimal number of video and in-person interviews one should shoot for if trying to secure at least a single job offer with reasonably high probability, so I decided to do a back-of-the-envelope calculation to try to figure this out.
Firstly, estimates with video interviews.
Based on some context clues (eg. Number of offered potential meeting slots), I assumed that different departments / programs generally selected a dozen candidates to interview by video conference, but that the actual distribution of candidate cohort sizes fit a normal distribution, spanning a range of essentially 6 through 18, giving a distribution like so:So for every video interview offer, I’ll essentially be in a candidate cohort size randomly sampled from the above probability distribution. Great. But what does that mean in terms actually getting a job offer? Well, it’s impossible to accurately model the probabilities without having information about the specific departments, their specific candidate desires, and the candidates in the cohort, so I’m going to make an incredibly crude assumption that once you’re selected for an interview offer, that each candidate in the cohort has an equal probability of getting a job offer, and thus inversely proportional to the total number of candidates in your cohort. I can simulate this process many times to get the mean probability of getting an offer (the thick red line), as well as the 95% confidence intervals of the range of probabilities that are likely (or unlikely) based on originally sampling from the above distribution. Currently, my main concern is going through this application/interview cycle and not getting a single offer. Thus, based on these probabilities and assumptions, I can simulate my chances of getting AT LEAST one job offer as the number of video interview offers increase:
So that’s great; if your only information is on your number of video interviews alone, once you have 8 video interviews it’s a coin flip whether you’ll end up with a job, and once you’re in the 16+ range you should feel relatively safe that you’ll get at least one offer in the end.
So some departments seem to go straight to offering in-person interviews to their best candidates, and even programs that do video interviews will still subsequently offer in-person interviews to their strongest candidates before offering them a job offer, so the number of in-person interview opportunities will be more closely correlated with the number of job offers you will presumably receive. Since vastly more resources (eg. time, money) go into in-person interviews, the chosen set of cohorts here are likely a fair bit smaller. Based on the number of seminars I’ve seen by faculty candidates for each faculty search I’ve observed (one at Fred Hutch, and a couple at UW), I’ve estimated the cohort sizes to center on 5, with some 4’s and 6’s:
And the corresponding simulation when multiple in-person interview opportunities are given:
So while the job search process is long and exhausting, based on my very crude assumptions, you probably want to be somewhere in the range of 6+ in-person interviews if you want to be reasonably sure you’ll end up with at least one offer.
A huge caveat: This is a SUPER simplistic model that does not incorporate ANY specific information about each potential job opening, and assumes COMPLETE INDEPENDENCE (between opportunities, for example). Thus, real life will definitely be different from what is shown above. Still, as is the case with most back-of-the-envelope calculations, I think it’s useful to know what general range any given candidate may want to be in if 1) they want to ensure they get a job offer this job cycle, and 2) have little information about offer chances (eg. were not specifically recruited by any departments).