succumbing to peer pressure

Saturday, June 14, 2008

Hot damn I had forgotten how nice it was to lay in bed all morning, then waste away half the afternoon sipping coffee and watching bad weekend tv! Actually, I wasn't even watching bad weekend tv, I was watching Swingtown thanks to On Demand (hooray cable!). And so, I offer you this little review:

Overall, I have to say I'm not yet won over, so I can't embrace fully recommending this show to others. However, I was pleasantly surprised to see Jack Davenport (Coupling, Pirates of the Caribbean) as a main character, even if his American accent wavers a bit from time to time. Canadian actress Molly Parker plays his wife, and she is sublime. Simply watching her excel at her craft makes all of her scenes enjoyable. Added bonuses include Grant Show sporting a horrific 70s 'stache and a Mark Valley cameo in episode two. The basic storyline is that Davenport and Parker have recently moved, just a few blocks, into a new neighborhood, where they befriend new neighbors who happen to be swingers. Their old neighbors are pretty conservative, and are jealous and freaked out by the new neighbors. Awkward sexual encounters ensue, often accompanied by drugs. I'm going to wait and see if it grows on me.

Data Entry
In other news, should you ever find yourself designing a survey instrument, please, for the love of god, don't replicate this particular problem. One of the datasets I'm working with includes information about BMI, which is missing for a few observations. No problem, since those observations do include information about height and weight, so a little arithmetic later and I can fill in the BMI values. Except that some genius decided to tell the data collectors that they could enter the height in either inches or centimeters and the weight in either pounds or kilograms! Sure, on a case-by-case basis I could probably figure out which was used, but there's no good way to explain that to a computer program, and it's just not sound data-cleaning to make assumptions like that. Bah!


Blogger Ben Ostrowsky said...

Well, the shortest adult is 72cm tall, and the tallest person was 107 inches tall. I'd say that within a range of 72-107 height-units, for an adult, there's some question as to which was meant.

Less than 72, it's almost certainly inches.

Greater than 107, it's almost certainly centimeters.

You can use similar reasoning for weight units, and of course a person whose height was recorded in centimeters is more likely to have their weight recorded in kilograms...

Not foolproof, no, but when the numbers are outside the range of world-record height or weight for one particular unit, I think you can flag that record as "high confidence" and try reporting statistics on just the high-confidence group to see how well the overall sample holds up...

6:36 PM  

Post a Comment

<< Home