Quant?

[thallters]

I'm somewhat torn as far as what I should be pursuing for my grad work. I'm in the process of strengthening my overall application profile for the 2016 season. Right now, I'm struggling between Quant and Clinical.

While I find Quant fascinating, I am slightly intimidated by the field. I did take a graduate level regression course as an undergrad (with Leona Aiken) and did well. The issue is that I've never considered myself a "math guy". I've always done well in math and stat courses, but I always have identifies myself as more of a verbal-oriented person.

For clinical, my interests are in the domains of addictions and health. I'm about to start work with an addictions researcher in hopes of contributing to a conference presentation.

All in all, I'm just looking for some insight from current quant phd students. Any advice would be great!

[lewin]

(with Leona Aiken) 

 

 

eeeeeee! I'd want her to sign my copy of Aiken and West (1991). I have no insight into your question, however. Sorry about that. 

[spunky]

I'm somewhat torn as far as what I should be pursuing for my grad work. I'm in the process of strengthening my overall application profile for the 2016 season. Right now, I'm struggling between Quant and Clinical.

While I find Quant fascinating, I am slightly intimidated by the field. I did take a graduate level regression course as an undergrad (with Leona Aiken) and did well. The issue is that I've never considered myself a "math guy". I've always done well in math and stat courses, but I always have identifies myself as more of a verbal-oriented person.

For clinical, my interests are in the domains of addictions and health. I'm about to start work with an addictions researcher in hopes of contributing to a conference presentation.

All in all, I'm just looking for some insight from current quant phd students. Any advice would be great!

 

 

well, as a quant i have to say i'm biased towards the awesomeness of my field ( 'cuz we're sorta awesome ) and like to point people towards this (horribly outdated but still) chart from the APA website:

 

http://www.apa.org/research/tools/quantitative/?item=6

 

when you say you find the field "intimidating"... what do you find intimidating about it? it's important to keep in mind that this field goes a little bit beyond just data analysis. you'll be pushed to sort of think like a statistician and deal with some non-trivial math stuff or handle complex analyses. you'll also have to be good at computer programming in statistical software (more specifically R. ok, STATA and SAS can come along as well). but SPSS will no longer be enough for the type of research us quants do. 

 

what makes you consider Quant Psych as a potential field of study? are you interested in measurement issues (i.e. psychometrics)? or more data analysis?

[Applemiu]

The most important thing is that you like it and enjoy working in the field. 

[thallters]

Spunky,

Thanks for the reply. Part of what is intimidating is the programming. I have some experience with SAS from that graduate Regression course, but knowledge of it is intermediate at best. I am interested in psychometrics, but I guess I lack confidence in my ability to contribute new knowledge and methods to the field. Also, I only reached calculus in my undergrad. Do you think w/ my limited programming and math experience I would have the chance to make it at that level?

[spunky]

Spunky,

Thanks for the reply. Part of what is intimidating is the programming. I have some experience with SAS from that graduate Regression course, but knowledge of it is intermediate at best. I am interested in psychometrics, but I guess I lack confidence in my ability to contribute new knowledge and methods to the field. Also, I only reached calculus in my undergrad. Do you think w/ my limited programming and math experience I would have the chance to make it at that level?

 

well, the fact of the matter is that very few people come into Quant Psych/Psychometrics graduate programs with a lot of mathematical or programming skills. most people in these programs are undergrads in psychology that also have a knack or an interest for data analysis and over the course of their MAs/PhDs and they become good at it. does it help to have a background in math/stats and some programming ability? oh, absolutely! i did my BSc in Mathematics (with a heavy concentration in Statistics) and that allowed me to jump right into research from the very beginning. i didn't need to take the research methods courses/statistics to be aware of the issues because i had already studied them. HOWEVER, once the other people in my program (actually i should say 'person' because only one other guy applied to my program when i started) were done with their courses, they (he) was sufficiently proficient to conduct quality research and ask interesting questions.

 

everything you need to know to become a good researcher will either be taught to you or you'll learn it as you go by talking to your more advanced/geeky peers, self-studying or by googling (Google will become your new best friend, trust me on that). but you also need to keep in mind that Quantitative Psychology (like other areas such as Mathematics or Statistics or Physics, etc.) will demand hours of alone time between you and a computer programming things over and over again until you can get them right. the main tool of the modern quant is the Monte Carlo simulation method (or 'simulations' for short). maybe you'd want to look into what they are because, in this field, you will be programming A LOT of those. this (and other stuff) is the one thing i do warn people about because this sub-area has a lot more to do with STEM fields than Psychology. but you could potentially go for a more 'applied' path in which you take on projects from other areas (clinical, social/personality, cognitive, etc.) that require complex modeling and maybe becoming an expert on this kind of analysis. for instance, a lot of Quants are being scooped up by people who do brain imaging research in biopsych/neuropsych/psychiatry/neurosciences/etc. because fMRI data is notoriously difficult to analyze, and methodology experts who can adapt methods to this type of data are badly needed as well. 

 

whether you feel you can 'contribute' to the area of not really depends on your research interests. i'm very theoretical and i really like to do math and programming for its own sake (which also impresses people when it comes to looking good on your CV). but you could also focus on learning how to apply complex methods to other areas of psychology and that can be your contribution. it's really up to which adivsor and which program you choose to enroll in because there area some people who are more 'theoretical' (like the UCLA or Urbana-Champaign crowd) and some are more applied (like the Educational Measurement people at Standford).

 

i think if you work hard you'll be able to become a good researcher in Quantitative Psychology but the question i'd like to pose to you is more along the lines of what Applemiu said of how in love are you with stats? because, trust me on this one, you better be ready to get married to this stuff or a few years down the line you'll start hating it. Stats is like music or a foreign language: you only get good at it if you practice, practice, practice! 

Edited October 23, 2014 by spunky

[Generis]

People who know statistics and can communicate their complex ideas clearly are worth their weight in gold.

[thallters]

Really great info, spunky, I appreciate it. Seeing as how stat is what really propelled my interest in research, I think I just may love it enough. My lack of programming experience stresses me out though. I suppose I should start learning R. Is it feasible to become fairly proficient with this language in about a years time? Any recommendations for learning resources?

[spunky]

well.. it depends on what you mean by "fairly proficient". i've been using R for almost a decade and i still hesitate to call myself "proficient" in it because of all the new stuff and undiscovered thingies i keep on finding about it. i mean i was not kidding when i said you *had* to learn R. SAS or STATA are pretty powerful and useful as well. R i just... well, it's quickly becoming *the* standard statistical programming environment (or as the NYTimes called it the "lingua franca" of data analysts http://www.nytimes.com/2009/01/07/technology/business-computing/07program.html?pagewanted=all&_r=0 ) but i guess if you feel more comfortable with SAS nobody's gonna throw a fit about it. you just reduce the potential pool of users who might find your stuff useful but they may see it in SAS and say "oh darn, i'll just wait for someone to translate it into R". or something like that. as you can imagine, the more programming languages you're 'fluent' in, the more influential as a scholar you can be because you can reach more people. 

 

if you have time to kill until gradschool i would not only encourage you to learn R but maybe to have a deeper look at areas where Quant Psych people work. Structural Equation Modeling/Item Response Theory are two VERY big ones and the software of choice for those (and most latent variable analyses) is Mplus, which is a lot simpler (but also a lot LESS flexible) than R. R still cannot do everything Mplus does but it's getting there through the use of the packages 'lavaan' and 'OpenMX'. although you'd probably be expected to be proficient in Mplus as well (i personally don't like it but i can use it without any major hassle). 

 

in terms of resources i'm usually a little low on those because i've found that there really is no better way to learn R than forcing yourself to use it and struggle with it. like the other guy in my program started by taking all his basic statistics homeworks and assignments (all done in SPSS) and re-doing them all in R just googling one thing at a time (e.g. 'how to do a t-test in R?' 'how do to an ANOVA in R?') he told me the first assignment took him about 3 hours (one of which was just figuring out how to read-in the data). but then the next assignment took him less... and less... and less until he became sufficiently comfortable with it to be able to keep on doing all his future assignments in R. i know people really like Andy Field's "Discover Statistics with R" (or something like that, i don't know the exact name) because of how chatty and down-to-earth it is. so maybe you can have a look at it and see whether you like it? but, in all honesty, Google is probably your best resource and you just ask google everything you want to do. since R has exploded in popularity, there are plenty of tutorials and blogs and youtube videos that teach you a lot of the basics.

[spunky]

even teh jobz come and find you in gradcafe if you're a quant!  

 

[thallters]

Thanks again for so much good info. My plan at the moment is as follows:

Reread Cohen, West and Aiken MRC for the behavioral sciences, familiarize myself with R and brush up on SAS.

I'll be doing all of this while also prepping for GRE (will obviously need to crush the quant section) and participating in research with a professor in hopes of presenting a poster in summer '15.

One last question, assuming I apply to several quant programs, would you recommend applying to some other areas of interest (ie clinical, social, dev) as backup?

[spunky]

Reread Cohen, West and Aiken MRC for the behavioral sciences, familiarize myself with R and brush up on SAS.

I'll be doing all of this while also prepping for GRE (will obviously need to crush the quant section) and participating in research with a professor in hopes of presenting a poster in summer '15.

 

 

i like this plan! i especially like the fact that you're willing to review your regression knowledge. understanding regression in and out is very important because there are many "regression ideas" in the more advanced methods that you will learn (we were actually talking about something similar here ). i feel like sometimes people learn their methods by "going through the motions", so to speak, without stopping to actually  think what exactly they are doing. and the biggest strength you'll have as a quant is that you'll be able to take any standard method that you know and adapt it to any type of data you're faced with. the analysis of real-life data is usually a lot more complicated (and interesting) than what you learn in your courses, but if you have a solid understanding of the theory behind the methods, then it becomes a lot easier.

 

 

One last question, assuming I apply to several quant programs, would you recommend applying to some other areas of interest (ie clinical, social, dev) as backup?

 

i... will have to defer this question to people from other areas. i know people in Social/Personality can be pretty savvy methodologists as well. many cannot run experiments due to the nature of their subject of study so they rely on pretty sophisticated statistics to establish "causal" (notice the " "  please) claims about their data (cue in Judea Pearl and the Neyman-Rubin causal model).

 

as a quant, however, it *is* sort of important that you're able to land your ideas about complex statistical methodologies in more applied settings so being proficient in a more 'substantive' (as opposed to purely methodological) area is ideal. mine is education/econ (particularly how they relate to the labour market) but yours can be anything you like.