Significant bacon
Q: The Herald says “Bacon makes everything better – research“. What research is this?
A: Wired magazine and the Food Network did a simple analysis of recipe ratings, comparing recipes for roughly the same thing with and without bacon (and a similar analysis for some other ingredients). For example, they compared sandwiches with bacon to sandwiches without bacon.
Q: And did bacon make everything better, like the headline says?
A: No.
Q: Ok, did it make nearly everything better?
A: No.
Q: Lots of things?
A: Yes. Though not necessarily by very much. Here’s the results. The biggest improvement is about 0.2 points on a 5-point scale.
Q: Are there actual numbers given somewhere, or results of other comparisons of ingredients?
A: No.
Q: These eight groups look a bit strange: kale vs all pasta dishes, for example. Did they look at other groups?
A: They might well have, but they aren’t telling.
Q: Do they explain why bacon is so good?
A: The Herald has an explanation (in fact, they have it twice, so it must be good). It’s the Maillard reaction
Q: Is that something that’s unique about bacon, which would explain why it’s different from other ingredients.
A: No. It’s in a huge range of foods.
Q: Is this sort of thing the future of data journalism?
A: Well, Wired says “Food is so personal and subjective that we’re always talking about it in vague and imprecise ways. But one of the many amazing things you can do with big-ish data is give precise questions to answers that always seemed so subjective.”
Q: Is that as bogus as it sounds?
A: Yes. For example, this is a sample where the average kale dish is rated more highly than the average dessert. Generalisable and objective food preference ratings are not what we’re looking at.
Q: Does this story, perhaps, come from the Daily Mail?
A: Why, yes. Complete with the misspelling of sriracha.
Thomas Lumley (@tslumley) is Professor of Biostatistics at the University of Auckland. His research interests include semiparametric models, survey sampling, statistical computing, foundations of statistics, and whatever methodological problems his medical collaborators come up with. He also blogs at Biased and Inefficient See all posts by Thomas Lumley »
So what you’re saying then is that the stats presented are inadequate for rejecting my null hypothesis, which is that bacon not only makes everything better but also confers minor superpowers?
11 years ago
Bacon Ripple Ice Cream. Yum! Right up there with Chicken Ripple Ice Cream. ;-)
11 years ago
It’s a problem that can be solved with data visualisation!
Specifically, they displayed very small categories (e.g. Kale, asparagus) equally alongside with very large categories (e.g. All Recipes).
A slight sense of proportion would help explain why “making kale better” is, in the grand scheme of things, not that significant.
http://www.drivehq.com/web/keithng/demo/EggSlicer.html
11 years ago