Fields in which Analytics Do Not Apply

Crowdsourcing.

What are some fields, or areas of study, that are resistant to analytical measures—that defy statistical attempts to understand them?

For instance, I wrote a story once about a company called Relativity Media. They were going to use Moneyball-like analysis to determine which movies to make. They thought they could crack some kind of code and place smarter bets than other studios. They ended up losing billions.

Are there other industries that still rely on expertise, on eye and ear and gut? Not out of some old-fashioned desire to cling to the way things were, but because they genuinely defy more dispassionate analysis?

 

Literary novels. Ya know, not thrillers or best-sellers, which have proven to be largely formulaic., but the kind of books that win Pulitzers or the National Book Critics Circle award. Basically -- like what you mentioned with Relativity Media and movies -- analytics seem less effective when applied to areas that have to do with tastes as opposed to wins/losses. Then again, food makers have proven that things that are high in sugar, fat and salt sell well. But that's a different kind of taste.

Do analytics work for music? How about art? As the saying goes, ya can't account for taste.

 

Analysis has a place in any endeavor.

 

Agreed, but statistical analysis?

 

Yes.
The issue, I think, in your example would be that statistical analysis does not apply to making profitable movies.
However, that seems like quite an assumption. Why wouldn't the issue be that the wrong statistical analysis was used or the statistical analysis was done incorrectly?

Making movies without statistical analysis also often results in money loss.

When you say 'fields,' what is your working definition there?
If we are talking business endeavors, then statistical analysis is always relevant.
If we are talking about endeavors in which objectives are not clearly defined, then it is impossible to say if they are helpful or not because the objective is not defined.

Take an example already offered - music.
Is the objective to make aesthetically 'important' or 'excellent' music? To provide creative fulfillment for the composer?
Hard to provide statistical analysis because the objectives are not really defined and measurable.
If the objective is to sell music - CDs, downloads - then yes, statistical analysis is relevant.

But statistical analysis cannot be applied to anything that does not contain a defined, measurable objective.

 

Yeah, "fields" is vague. I don't know how else to describe it. Like, I'm sure there are ways to analyze violins and what makes a beautiful violin. But Stradivarius did that by ear, and did it better than anybody. I guess the music version you reference wouldn't be how to write a hit song—because there are both formulaic and non-formulaic ways to do that—but an A&R person who stands in a crowd and knows he's seeing the next big thing.

There's a defined objective—find the future biggest band in the world—that doesn't necessarily lend itself to statistical analysis. That's the sort of thing I'm thinking of.

 

Statistical analysis is very good at predicting the past.

 

But are you looking for examples that illustrate statistical analysis does not apply to all endeavors as a means to prove a predetermined point?
Or are you honestly addressing the question of whether statistical analysis applies in all endeavors?

See w hat I mean?

If I'm trying to move units, I am analyzing currents and developing trends in music sales, and then I'm looking to sign acts that fit.

 

Bagmen for companies pushing athletic shoes.

 

This is well put.
Correlation does not equal causality. Most people fail to understand that, leading to the overestimation of predictive accuracy.
Further, statistical analysis is one piece of the decision-making process.

And in many instances, people are not honestly considering the data. They are looking for patterns in the data that reinforce, correctly or incorrectly, already held objectives.
I can analyze market trends and figure out what type of restaurant is more likely to be successful at a given location. That is appropriate.
Often, though, I have seen a space I like, or I've already determined that I want to have an Italian restaurant, or I've already determined both - now I'm really just seeking data that can reinforce what I want to do.
Lastly, most restaurants fail. Period. The failure of my restaurant is conditional on a lot of factors. Saying that statistical analysis was not useful in my business endeavor would be false, although saying that might make me feel better.

In the end, analysis, statistical or otherwise, requires a defined and measurable objective in order to have any utility.
If my objective is to move units, statistical analysis can help me sign the right act.
If my objective is to be able to tell people I signed the next Elvis and I bucked all the trends to do it, then my objective is tainted from a business perspective.