Wednesday, July 20, 2005

Worst job interviews

I'm curious to hear about job interviews gone wrong because I just butchered mine. It was for an economics research position and I was grilled with econometric questions that should be basic, but I turned into a bumbling idiot. What's the problem with running an OLS regression on survey data with yes or no responses, you ask? Well, my answer was you should run a probit, logit, or tobit. Fine. What's the statistical problem with that? I managed to say any zeros will not affect the beta. wat tha? Not only is that wrong, that wasn't even the question. To make matters worse, it was with two leftist Brits at the CMPO at university of bristol. I did manage to get questions about panel vs. pooled data correct. That's what I'm resting on in order to get the job. ugh.

10 Comments:

Blogger Steve said...

I've actually never had a real job interview. I have two coming up, but I doubt either will be as hard (read: important) as yours. Good luck. =)

10:32 AM  
Blogger See Jay run said...

Oh Priscillia, you are so funny! What the hell is Econometrics anyway? Like Steve, I have never had a real job interview either and so I really feel for you. But you do have something that I bet the other candidates lack and that is a real cool, confident, and humorous demeanor. You're a stud muffin, basically. See you later, perhaps in a couple hundred years :)

11:33 AM  
Blogger Priscillia said...

By the way I conducted that interview, I dont really know what econometrics is either!:) Oh, it's just fancy math&statistics.

I've since spoken to other people who are fairly shocked by the questions I got, which makes me feel a little better.

Can u imagine what the world will look like in the year 2205?

steve- good luck on ur interviews.

clyde- watchu glyding?:)

12:18 PM  
Blogger Steve said...

Funny anecdote: In a debate last year between a team from our squad and another school, the other team had this "ban the death penalty" plan, wherein they cited this "econometrics study." They were far from economists though, so that part of the speech went something like this:

"To support our point, there was this econometric study done about blah blah blah. Econometrics is good because... well, it's this really complicated mathematical formula, and... it takes statistics and... well, it's just really good!"

Priscillia: Thanks. One of the part-time jobs is for my school's advanced placement program, and the other is for a tutoring position (15/hour for an undergrad student is really appealing). =P

12:57 PM  
Blogger Jessica Spaulding said...

In a job interview with the VP of NPR I talked about porn. I didn't get the job. But I still fill in for her secretary so I guess it didn't go that badly.

The situation was this: She asked me to tell her about a time when I had mediated between two people/groups. What came to mind was the year I was treasurer of the student body and had to go in front of the Board of Trustees to defend the fact that the student council had just funded a porn club. I didn't know it at the time, but I basically used the Harm principle to convince them. They weren't hurting anyone, they were responsible adults and they wouldn't allow anyone under 18 in to the screenings.

I need to do a radio story on the porn club. The whole thing was really amusing. Perhaps not the best thing to talk about on a job interview though. Especially a phone interview when it's harder to tell that it's time to shut up.

1:11 PM  
Blogger Priscillia said...

Most PhD students can't tell you the difference between panel or pooled data, which is something any basic researcher should know..but they can throw around the names of these fancy models that no one actually uses except for academics. My friend just met with a guy at the dept.of justice who evaluates mergers & acquisitions...he uses Bertrand & Cournot models. If you open up a game theory book, it will be on the 5th page. So all this complicated math & statistics is total crap.
btw, if ur interested game theory games, check out gametheory.net

Jessica- that's good stuff. Heluva first impression...nothing like promoting the publication of sex when trying to get a job!

1:56 PM  
Blogger Tony said...

On a phone interview I was asked who my major philisophical influences were. It was my first interview so through some nervous double-speak I managed start talking about quasi-anti-religious issues and calling Lucretius (a religion basher) my "guiding light." Interestingly, my interviewer was a devout catholic (whoops)... (I only learned this later). Luckily, I got the job anyway (and that's actually where I'm writing from now).

As far as using Bertrand and Cournot at the DOJ, I would be suprised if they used the version you find in an introductory Industrial Org./Game Theory text. There are certainly more complicated/subtle versions, especially of Bertrand - I mean using basic Bertrand (unrestrained price competition leading straight to pricing at cost in a duopoly) is something even the DOJ can't get away with. Then again, it is the DOJ.... [Priscillia - we should start our own econ nerds blog :) ]

[BTW: I'm addicted to Cournot, it's more or less my all-time favorite econ model, hence defending the DOJ]

5:01 PM  
Blogger Priscillia said...

I can appreciate Cournot models, but what if players dont act at the same time? Can you still use it? To be honest, I dont remember it all that well but I think you have to use stackleberg & my friend (getting a PhD specializing in IO) says they don't really look at that. Maybe I'm missing something obvious, like they don't have to deal with sequential movers...?

ps. i will permanently be on the computer coz i got a shocker when i returned & my dept said i had to finish my masters paper within the month or pay fall tuition.

9:42 AM  
Blogger Joseph Schultz said...

Econometrics... yeah. I'll just say that I thoroughly disliked my econometrics professor and leave it at that.

I really like the repeated OLS stuff - though I'll be damned if I ever find a good reason to use it.

Most of the time I get stuck using hyperbolic stuff (basically OLS but with a different function type). Problem is, I can never figure out if it's actually better than a standard OLS (can't compare models based on different dimensions right?). Oh well, enough BSing for one night.

6:37 PM  
Blogger Steve said...

Back to the less nerdy stuff, I got the tutoring job. =) Interview for my other job tomorrow!

6:58 PM  

Post a Comment

<< Home