Related Posts
Any suggestions on securing an interview with Apple in Austin in a senior quality manager role? I interviewed with them late last year for a role but didn't make it to offer phase. I've been in quality for over 20 years spanning several industries. Just looking to cut thru the applicant pool if anyone can help. Apple
Salary of clinical research associate at PPD?
Additional Posts in Consulting
Only 2 more days till Consultant Friday!!
The best 'advice to my younger self' yet
Any layoffs in big4?
Best book on personal finance?
Book recommendations for new executives?
How is everyone planning for potential layoffs?
Anyone have a Branch standing desk? Thoughts?
New to Fishbowl?
unlock all discussions on Fishbowl.
People like to hear stories than read numbers!
Chief
If they're opposed to stats, you might be focusing too much on the math and not enough on the implication.
I regularly do multiple regression in revenue growth projects. I often don't use the output because high correlation often does not reflect causation. If I can get an insight from it, I'll throw up some high-level numbers that say changing your price by x% under y% circumstances will affect P&L by z%. Sometimes I get these numbers from multiple regression other times I'll get it by looking differences of averages. The method doesn't really matter, as long as there is confidence in the output.
Broadly agree, but some folks confuse correlation with causation and then it becomes a problem ...
There’s usually such poor data and so many confounding factors that regressions should be taken with a pinch of salt.
Like M1 said, you should have them in your toolbox but even when you use them the math is not the message.
In general, it is probably due to the focus on the math/method more than the result. Going to a high-level narrative is more useful to communicate the insights and generate recommendations than passing along the actual statistical results. Those could be saved in the appendix instead.
Agreed
Strategy is as much art as science (stats). Too much of either makes it useless.
Are you sure you understand stats enough to peddle that to clients? Spend and sales typically have a lagged relationship, which tends to indicate the use of time series methods. Unless you’re referring to the obscure terminology of “dynamic regression”, which is technically a time series method, regression is not the solution to the problem you’re trying to solve IMHO.
^ did you know that Kevin bacon movies correlate to pool deaths?! Numbers alone rarely tell the story
And just to have a little more fun with the argument: did you know that chicken ten mustard nail credit card?! Words alone rarely tell a story.