Sunday, August 29, 2010

Research

I think I'm quite good at research.  I'm not always great at executing it perfectly, but I think I have a pretty good grasp of the basics of good study design and data analysis.  I try not to forget about simple things like specifying the population, intervention, comparator and outcome.

Yesterday the London correspondent (yes, really and would you believe she's from New Zealand) was running around the news room like a newly decapitated chicken looking for a 'researcher'.  Can I help?  Jean Adams, media fellow, here for a month. 

Yes, the deputy editor wants me to do a story on travel disruption around sporting events.  I need to know if there was any disruption on the tube lines serving the London football grounds during home matches.  Just do since January.  Just on the weekends.  Premiership matches only, don't bother about the FA cup or anything.  Just do Arsenal and Chelsea.

She seemed frantic, so I just did what I was told - found all the home, weekend, premiership fixtures for Arsenal and Chelsea since January.  Worked out what tube lines the stadiums were on.  Searched for disruption on the Transport for London website.  I knew it was stupid, (the population is neither complete nor a good sample - not all the clubs, not all the matches; there's no comparator; the specified 'outcome' doesn't seem necessarily relevant - station closure at the opposite end of the line from the stadium might not be important) but she seemed pretty clear that's what she wanted.  I rather forced at least a fuller sample on her by including Fulham and Spurs as well.

As soon as she saw the data, Ms London Correspondent worked out it was stupid too.  We shouldn't look at the whole of the lines serving the stadiums, just around the stadiums.

As soon as she saw that data, she worked out that was stupid too.  Jean, what do you mean by "around the stadium"?

As soon as she saw that data, she worked out that was stupid too, but I think by then time had run out and we were going with whatever we had.  At least she was kind enough to give me a credit on her article and there was another funky graphic.


You will notice that there is no mention of Arsenal at the Emirates here.  That's because there was no tube disruption around the Emirates during the study period.

Critical appraisal anyone?

1 comment:

  1. I have actually been saying for a very long time that everyone (yes EVERYONE) should learn epidemiology and causal inference at school...

    ReplyDelete