17 June 2009

Lies, Damn Lies and UrbanSpoon Statistics

The UrbanSpoon site is a good resource to find restaurant reviews when looking for a place to eat. More so when visiting another city but it's plenty useful here as well. The site links to all restaurant reviews, including local blogs. Last week I found the "Cincinnati Blog Leaderboard" page which lists local blogs that do mainly restaurant reviews. It lists them by views (which I presume means clicks from UrbanSpoon to the blog) and posts (only restaurant review posts, since that's what the site is for).

I immediately recognized the opportunity to geek out over these numbers. What I wondered is how the blogs rank under the combined statistic of views-per-post.

Why? Mostly because I like playing with numbers.

This statistic reveals whether two things are correlated. In this case, views and posts. The Correlation Coefficient is 0.96 which means # of posts and # of views are closely correlated (0 means no correlation; 1 means strong correlation). This is not surprising; the more reviews that are available, the more opportunity there is for clicking to them.

After calculating the standard deviation I realized something interesting: although all the blogs have different views-per-post numbers, the difference is not statistically significant. Statistically, they are all the same except for two. One is lower than the others by a statistically significant margin (barely) and another-- and this was the interesting part-- is so far higher than the others that I had to remove it from the calculation because it skewed the numbers so much.

There is no special widget for it but if there were it would belong to Chickpeas, Please which is an astonishing FIVE standard deviations beyond everyone else. Congratulations on being #1 by this mostly meaningless statistic!


liz said...

nerdy yet interesting post, but only because it directly applies to my life. kudos to allison for this impressive achievement. she deserves it since she writes a great blog, if only she would get off her lazy ass and post more often (ahem). of course maybe she doesn't need to according to this statistic.

Allison Johnson said...

well thanks for the recognition, i'll gladly accept even meaningless awards for my blog. maybe i'll commission someone to make a widget.

liz i swear i am going to start posting more!!

Kate The Great said...

This post turned out far, uh, *nicer* than I expected.

Still, would you like a spoon to stir that $#!+? ;)

WestEnder said...

I'm sure I have no idea what you're talking about. Mathematics is always nice!

Food. By Mark. said...

This is actually beyond me right now. This was the no.2 search result in google when I looked up "how to remove kudos on urbanspoon". But interesting nevertheless!

Anonymous said...

You've proven that stats do in fact lie, because you're just completely off base.

They are not clicks; it has to do with the vote counts for the places that are reviewed. I haven't quite figured it out yet. But if there is a correlation between the blogs its because they have reviewed a lot of the same places.