This is another example where, if you lack analytic skills, you will jump to the wrong conclusions. This news article was published in MyNorthWest. It's about the new law that went into effect a year ago in WA, allowing grocery stores to sell hard liquor. Here we provide 16 reasons that could explain this paradox. You might even come up with additional reasons.
On a different note, how would you measure the impact of the following wine fraud: you go to a restaurant, order a $20 glass of wine, get offered a $10 wine glass and charged $20. Some claim that most Americans would not notice the difference. It happened to me once with a glass of champagne (substituting good champagne with something totally undrinkable), but it could have been an accident.
Related articles
Comment
Here’s the basis for my skepticism and maybe a way out of the problem.
With the suite of techniques currently available, we would use either a supervised or unsupervised algorithm. Forget supervised. There is one data point, WA state, and we can’t assign a measure. We have a problem even trying to classify this one outcome (the whole point). At most, we would ever have 50 data points, to train on. But, what is the test and what is the test data. Supervised is out. Maybe an unsupervised learning model? Dimensionality reduction. We lack a measure and there’s limited data.
Let’s classify some outcomes. My next idea would be to propose a literature review and secondary data analysis using results from other states over maybe the last 40 years of liberalizing access to alcohol. Classify each. We now have the basis of a training set.
The first thing I thought about when reading the article is causality vs. correlation.
I don’t think the problem is answerable with any confidence. Vince, as usual, you list a whole host of good questions that could be important.
But, we have to make progress somehow. In this case, I’d define my costs differently, I’d define it as public safety. I’d scope it in a way that I would have some confidence in my methodology and data. Underage drinking would be good. The null hypothesis: underage drinking remained unchanged. I think answering this question is doable. And, importantly, it matters.
I’m skeptical that models will be invented that take into account the reasons for the paradox while establishing causality.
Vincent, re wine/champagne, it's only fraud if it's mislabeled.
Taste is subjective. Run an experiment with your friends. Buy some $5, $50 and $100+ wines and pour them into identical glasses. No hints of any sort. You'll find that much of the time even those who purport to be oenophiles can't tell the difference between popular cheap wines and the more expensive varieties.
Re: the rest of the article, I'm not sure anyone could come up with any meaningful connection between that law and its effect on DUI. Some relationships remain beyond the reach of even the best models...at least for the foreseeable future.
How would you get the right data and the right methodology, to assess the impact of the new law? For the state of WA, the question is: did we reduce costs related to alcohol consumption (by increasing tax revenue from alcohol sales, laying off state-store employees, modest or no increase in alcohol-related crime, etc.)
© 2019 Data Science Central ®
Powered by
Badges | Report an Issue | Privacy Policy | Terms of Service
Most Popular Content on DSC
To not miss this type of content in the future, subscribe to our newsletter.
Other popular resources
Archives: 2008-2014 | 2015-2016 | 2017-2019 | Book 1 | Book 2 | More
Most popular articles
You need to be a member of Data Science Central to add comments!
Join Data Science Central