The PhD will only be useful in a small (but growing) subset of data science jobs. Their are data scientists who develop new algorithms and techniques, and those who apply the. For application, the PhD is probably overkill and extensive experience with a bachelor's, or a master's is better. For theory the PhD can't be beat, of course. I say the opportunity costs are high, and should be carefully weighed, because I have my bachelor's and am making more than a data scientist friend of mine who has his PhD. The vast majority of what he learned in the PhD program he hasn't found useful in the real world (yet, anyways). So it's all about what problems you want to work on. Personally I thrive in the application arena of data science, while others thrive in the R&D data science arena. If you like one you probably won't like the other.

It may not necessarily be the best option for the OP's daughter, but it is possible to do a PhD that uses data science as a tool rather than the focus. There's a lot of interesting work going on in academia using data science in a variety of fields (like medicine) related to predictive analytics (predicting hospital readmission, predicting therapy outcomes), time series analysis (analyzing EKGs, and other vital sign data) etc. A masters would likely allow you to work on similar projects, but if the goal is a PhD, there are certainly application rather than theory focused research groups across the country

I didn't say necessarily a PhD in machine learning. I think, generally speaking, that a technical PhD vs an MBA could make for a better data scientist.

This said, I don't have an MBA, and most of the MBA people I know are not technical at all or try to stay away from the "technicalities". There's for sure a huge need of technically and business savvy people.

Lol, no. Just learn programming and start working for a company. No better way to learn data science than to work as a data scientist.