MS in statistics with hardly any math background?

[Juliska]

Hi, I'm wondering how difficult it is to be accepted to a MS Statistics program without an undergrad background in math. I double majored in philosophy and law & society, and the only relevant course I took in college was calculus as a gen-ed requirement. Still, I feel like I am good at math (I was one of the few philosophy undergrads who really enjoyed logic class). I want to go into social science research, which is why I want the MS in statistics.

Having never taken a statistics course, what are the chances of getting into a graduate program? Are there any universities in particular that I should look into? I just graduated with a 3.5 gpa and strong extracurriculars in social justice activism, which are the issues I'm interested in researching.

Advice would be much appreciated!

 

[Dason]

It might not be the easiest to get into a program but I guess it depends on the department. Have you taken the GRE? How did you do on the math section?

 

[thedreamshaper]

I am a math/econ masters student myself, mostly my courses are in statistics. I would say it would be very difficult to take statistical courses without extra math..

Do you know linear algebra(matrices, vectors) ? mathematical analysis(infinite series etc.) ?

Im sure you can find some powerpoint slides from various programs online, download them and see if the math causes you trouble would be my advice or go a softer (but still statistical) way, like managerial economics with a focus on statistics..

 

[trinker]

I have a friend who is a PhD in Stats who started his PhD (at the same time he got a masters too with the PhD which I think is standard for a PhD in Stats but not other fields) without an undergrad math degree. It's doable but you'd have to obviously work a bit outside your course work to become proficient.

I want to go into social science research, which is why I want the MS in statistics.

Have you thought about a PhD instead in a field that you are interested? You'd get applied statistics training there and be more on the research end of things. I'm not stating this is what you should do, just throwing out options/ideas (though I don't know is you have to have an MS to do a PhD) based on what you were saying.

 

[Dason]

I second trinker's suggestion. You can get active in statistics while studying for a PhD in a different field... but since you've never taken a stats course you wouldn't necessarily want to start a program in statistics and realize that it's nothing like what you imagined and that you hate it (which I hope you wouldn't but it is possible).

 

[spunky]

i think i have to second what pretty much everybody else has already said. for an MSc in Statistics i think any potential advisor would try and see whether you have at least mastered the basics already (linear algebra, multivariate calculus, regression theory, at least some introductory probability theory, etc...) and, at least from what you mention in your post, all of these are areas where may still need to work a little bit more on. first year calculus just wont cut it in any (respectable) math department...

now... with that being said... i switched form an MSc in Statistics to an MA in Quantitative Psychology/Psychometrics... have you considered something like that? if your aim is to focus on social research, perhaps an MA in Research Methods, Quantitative Psychology, Educational Measurement, etc will both give you the exposure you need to advanced statistics while, at the same time, be able to speak the common language of the social sciences...

 

[jpkelley]

Seems like the OP has an excellent resume for getting into a great Ph.D. program in the social sciences. Since you feel like you're comfortable with some math, then you could try out more hardcore statistics/math courses during your first year in that kind of program. If it's like other fields outside statistics (e.g. ecology, psychology, conservation biology, or whatever), people with great stats skills --even without 100% knowledge of all the underlying mathematics, etc.-- are still hard to come by.

 

[noetsi]

I point out in passing, that it makes a great deal of difference what type of "statistics" degree you are getting. In a statistics department then a good deal of math is likely critical (particularly if you want to pass...). I am in a master's in statistic in the college of education and although we use some basic calculus and matrix algebra, the primary focus is on running various statistical methods without the involved grounding in math of a statistics program.

I took "doctoral" courses in social science statistics without any recent math at all.

 

[Janus]

With that background it is unlikely that you would get into a stats program. Even an excellent GRE would not help much. I believe I saw a statistic for the school were I received my MS in stats (University of Illinois) and the majority of applicants had an 800 on the math section.

Other people's emphasis on other programs is reasonable only if you are actually interested in a different field. I originally entered a Ph.D. program in poli sci. The stats department offered an option to graduate social science students to obtain an applied master's in stats along the way. However, the applied master's was pointless, though it seemed attractive at first. It required only 4 classes taken in the stats department and 5 taken in your social science department. Coincidentally, the poli sci program required 5 in-house stats classes. However, they could not be double counted. So, I decided to get the normal MS. No application required. Just a few approval forms from my advisors. I ended up leaving poli sci early with an MA and the MS in statistics. Btw, I took one math class as an undergrad (I was a journalism major) even though I loved math. There were a few prereqs for the statistics program -- I took Probability Theory, Statistics and Probability, and Intro to Linear. The rest I picked up along the way.

If you are serious about this, take some stats/probability courses to prepare and to have proof that you have the ability to successfully complete the MS.

 

[noetsi]

You should think what you plan to use the stats for. If you want to do research, or higher end jobs like say much of biostatistics, then you likely need a good deal of math. If you are going to use it as a data analyst in most corporations/state agencies you don't.

Dason likely won't agree that is my experience....

 

[Link]

You should think what you plan to use the stats for. If you want to do research, or higher end jobs like say much of biostatistics, then you likely need a good deal of math. If you are going to use it as a data analyst in most corporations/state agencies you don't.

I can't tell you how much I agree with this. As a data analyst, you just have to set up the models and follow instructions. If you really want research or higher up though...be ready to get lost and confused, spending countless hours sitting there trying to dissect those abstract theories.

PS. I now know that statistical parameters are mappings from a probability distribution to a parameter space. hehehe...took me a while to completely understand this and just thought I'd share that.

 

[noetsi]

As a data analyst, you just have to set up the models and follow instructions.

I would say that you have to know which box to click in SAS or SPSS (and to know if the results make sense given your data), but it would probably drive Dason crazy

Data analyst who spend too much time considering alternate methods (let alone the math behind their models) in the private sector will quickly end up being ex-data analyst. You have to get the work done on time.

 

[spunky]

PS. I now know that statistical parameters are mappings from a probability distribution to a parameter space. hehehe...took me a while to completely understand this and just thought I'd share that.

the only mappings i ever understood were conformal mappings... i never thought i'd ever see them again so thanks for sharing!

 

[Link]

hehehe. You're welcome.

Also, if anyone wants to discuss fluctuating probability distributions and pathwise differentiability, I'm all ears! Here's a sample (hope I'm getting it correctly):

For a given target parameter, \(\Psi (P_0) \), we can define the pathwise derivative as: \(\frac{\partial}{\partial \epsilon} \Psi (P_0(\epsilon)) \mid_{\epsilon=0} \), which will tell us the rate of change of our parameter \(\Psi (P_0)\) at \( \epsilon=0 \)