Comments on A Computer Scientist in a Business School: Probabilities and MTurk Executives: A Troubled Sto...

Consider this: if you have 3 workers of 68% accura...

2011-11-14T21:43:32.772-05:00

Consider this: if you have 3 workers of 68% accuracy, the combination of the three (e.g., using majority vote) will result in an average accuracy of only 75%. In other words only 3 out of 4 times the majority will generate the correct answer. To reach 90% accuracy, we need 11 workers with 68% accuracy each. And to reach 99% accuracy, we need 39 workers of 68% accuracy! (I will present the math in a later blog post.)

The math formula would be very much appreciated!

Thanks

I noticed that they removed the old blog post, but...

2011-09-19T21:50:42.628-04:00

I noticed that they removed the old blog post, but did not see the updated one. Thanks for the pointer!

Regarding the market being dysfunctional: Yes, I will agree with you. But there is also an information inefficiency from the employer side: I do not think that anyone understands how much more valuable accurate workers are. Based on my own informal survey, people think that a worker that is 80% accurate, should get 80% of the salary of a worker who is 100% accurate. Intuitive, yet very very wrong!

I guess it is time for a blog post to illustrate how wrong this impression is.

Interesting post, but people get probabilities wro...

2011-09-05T17:38:37.936-04:00

Interesting post, but people get probabilities wrong all the time. whatever.congrats on finding it. sometime this year there will probably be a rehash of the monty hall problem somewhere on the web. however, the bigger issue is the removal of statisticians from business schools-someone has to be able to spot these errors, and ideally teach the correct foundations. Chicago, very few stat people left, NYU a few very good people, Columbia one or two, Michigan maybe one person, Wharton of course has a dept inside the bschool, Georgetown 1 or 2, UVA 1 or 2, Stanford none, Berkeley none, some at UT-Austin, some at USC, Harvard none, Kellog none, Darden none, and so on.

bschools used to embrace classical statistics, such as DOE and quality control. Then time series became useful for finance, and now Bayes is hot for marketing. But now where are the mba's and phd students getting their quant training? Its one thing to train mba's to use solver or fit a line in Excel. What about critical thinking? Modeling? Basic probability? I fear that the mba curriculum (not the faculty) is getting soft, and not pushing students to study good fundamental ideas.

Exactly my thoughts too. The problem of the origin...

2011-09-04T13:58:16.231-04:00

Exactly my thoughts too. The problem of the original post is that the stated propabilities (i.e. 0.99*0.99) do not calculate the accuracy of the answer . Nice post!

You are, of course, correct. Thanks for catching t...

2011-09-04T09:43:34.292-04:00

You are, of course, correct. Thanks for catching this, it is now fixed.

Hi, Great post, You might have a typo somewhere in...

2011-09-04T09:06:51.319-04:00

Hi,
Great post,
You might have a typo somewhere in the middle,
With 99% accuracy, the probability of a worker being correct is p=0.01.
Should be probability of a worker being incorrect...