Rec letter from non-math professor?

[mathmollusk]

There are only two professors in math who I know could give me strong recommendation letters for stats graduate school (probably masters). 

Would admissions committees frown upon my application if my third rec letter was from a professor who I know really well in a non-math (and non-STEM) discipline? Or, would I be better off getting my third letter from a math prof whom I only had for one class?

Thanks.

[theduckster]

As others will tell you, you should always go for the professor/faculty member that has directly worked with you and can say something strong about you. If a letter is just reiterating your transcript or CV, then it will probably be next to useless (unless the letter writer is a really good embellisher).

Just make sure that at least one of your recommenders is a tenure-track professor (or academically-known individual) in a quantitative discipline. It seems like you've already got that covered, so no worries here.

Edited January 24, 2019 by theduckster

[Geococcyx]

As usual, I'm not an expert.  That said, I've gotten 2 Ph.D. acceptances and 1 Ph.D. interview thus far, and my recommenders were a research mentor in a sort-of quantitative social science, a statistics professor I took 1 applied class with, and a professor I took 1 journal club-style class with in a mostly non-quantitative field.  You'll probably be applying with a better math background than me too.  I don't know much about your GPA, GRE scores, or school's perceived prestige, but I'm of the opinion that if you already have 2 professors you've had multiple math classes with AND you know they will write strong recommendations of you, then you're doing pretty darn well, and don't stress the third letter writer so much.  Use whoever you think will write the strongest letter, the first two letters should probably convince the school you can do math.

Short Answer:  As long as you and your non-STEM recommender get an idea of how their recommendation fits into your case for admission into these programs, I don't see a reason you can't use them.

[mathmollusk]

Great - thanks for the replies!