Research Assistanship before PhD

[Where To Go]

I would not list it, because one should already be writing typesetting their resume in Latex.

Valid point. I would probably still list it though if it fits in some kind of list of programming ability. Making extra space for it is probably superfluous, but I don't think adding it to a list / line of other languages would hurt you unless you seem to be bragging about it.

[rawls234]

From personal experience, I really don't think that programming skills gives any positive weight to an application. Even if you can build large distributed systems in C++ or Go, it won't matter. A knowledge of the standard statistical programming languages is probably helpful, but beyond that it's probably not worth much.

[ColonelForbin]

I think you all are over-thinking this. There's no issue with putting LaTeX on your CV even if you made it in Word *GASP*! If you know a language put it on your CV, if you don't then don't.

[jakethedog]

what is considered advance skills as far as excel? For instance, at my current job its important that I meet a threshold of data points per hour(I always exceed it). I never do any of my work manually, I format BLS, Census, and all different types of data sets by writing different formulas(vlookup, sumproduct, etc.) and using different tricks with text to columns delimited, advance options of find and replace etc. I used excel to efficiently clean up the issues I had with merging strings on STATA. For instance, my dependent variable was median 10 year earnings of graduates from 1000+ universities, which was a pain to merge with some of my independent variables such as SAT percentiles. For example, some data series would spell the University of Virginia as "U of Virginia" while others spelled it as "Univ. of Virginia".

As for my experience with STATA, I used it for running regressions and statistical analysis on both of my writing samples and I had the chance to work with panel data as well(my roomate wrote about cross sectional differences in trade of OECD countries from the 1970's until now). I also use STATA do files when dealing with larger data sets. Aside of that, I'm familiar with the syntax in general, and whatever idk I usually just google it and pick it up lol. Should that be enough for an RAship?

I'm currently teaching myself Python right now, so I have a familiar idea of the syntax involved. I'm specifically looking into learning about the Pandas package which seems like something pretty useful to know. What else should I teach myself in regards to Python? I can I state this on my resume/cover letter?

As for Matlab, I still have no idea what they use Matlab for? I do not understand why it would be useful if one already has a general knowledge of STATA and python?

[Where To Go]

Personally, I would hope that any more extensive work in Excel is done using something like VBA and not semi-manual search and replace, to keep things reproducible. (But I know that's not what typically happens.)

 

For PhD applications, I only listed programming languages I'm familiar with on my CV, didn't really mention it in my SoP. Space is just too valuable there, and programming is not the main part for a PhD (and at the same time can be picked up later on when needed). For a RAship, I guess it depends on what you are expected to be doing and what you are asked to say about yourself. The ones I looked at often mentioned something in the descriptions, and then it would make sense to say something about it.

 

Can you do something like Monte Carlo simulations in Stata without too much trouble? Easily program a new estimator? (I only used Stata for simple empirical work so far)

Python can theoretically do what you can do in Matlab. But if your main operations will be done on matrices, Matlab is likely to be more convenient. If you look at e.g. R, you probably find more econometrics methods in packages available for that than for Python. There is more for Python than just pandas though, check e.g. this post

r - Python as a statistics workbench - Cross Validated

[Econhead]

As for my experience with STATA, I used it for running regressions and statistical analysis on both of my writing samples and I had the chance to work with panel data as well(my roomate wrote about cross sectional differences in trade of OECD countries from the 1970's until now). I also use STATA do files when dealing with larger data sets. Aside of that, I'm familiar with the syntax in general, and whatever idk I usually just google it and pick it up lol. Should that be enough for an RAship?

I'm currently teaching myself Python right now, so I have a familiar idea of the syntax involved. I'm specifically looking into learning about the Pandas package which seems like something pretty useful to know. What else should I teach myself in regards to Python? I can I state this on my resume/cover letter?

As for Matlab, I still have no idea what they use Matlab for? I do not understand why it would be useful if one already has a general knowledge of STATA and python?

 

My can't say what is expected of you for an RA position, but in terms of practical experience with Stata, it sounds like you have done less than a master's student would learn in a "applied" econometrics course (sometimes called econometric analysis). It just doesn't sound like you'be been exposed to a lot of different things-of course, part of this is likely a lack of exposure to the theory that acompanies it. It is difficult to sometimes know what to do practically of you don't fundamentally understand the problem. An easy example would be thinking that you can do regression because you've done it before but having no idea that you need to log certain variables.

[jakethedog]

I don't see a problem putting it on a resume if you actually know how to use these. Your MATLAB experience seems rather limited though. Personally, I think it would be nice to have programmed a more complex, non-standard estimator / test yourself or run some kind of monte carlo simulations.

But then, at some point knowing a programming language (Python, C,...) and a matrix language (MATLAB, Gauss, R) means that using another matrix language should not pose too many problems. As an economist, you don't need to be able to write a complicated program without looking at the documentation after all.

I don't know whether STATA is seen as indicative of your programming ability in general, but it is likely considered quite useful.

Excel - I assume you mean VBA? Otherwise, I don't think people will see much value in it - it would be tedious manual work, and experience in that kind of work only speeds up things so much. (You can probably do most things in R,... just as fast as in VBA, in particular if you have a better grasp of the concepts used in matrix languages than in Excel. Last time I used VBA was to apply conditional formatting to hundreds of cells - it does give a nice overview.) So my personal opinion is: writing Excel but not VBA does not tell them much because few students who "know" Excel (but not VBA) will really have a significant advantage over the normal Excel user (which probably includes everyone), and there is little you can do to show that you are such an advanced user.

LaTeX I would list. Probably does not help a lot (unless that's one of your main tasks), but all else equal it could tip decisions in your favor.

 

I tried reading/going over this document

http://www.tcd.ie/Economics/TEP/2010/TEP0110.pdf

Everything is self explanatory up until page 30. I downloaded the econometric toolbox and I opened the editor but I have no idea how to run that script... Is this what I'm supposed to be teaching myself? I tried to look up monte carlo simulations but I really have no idea what I'm doing or what I should I be teaching myself. Do you have any links of information I should study so that I have a better idea of what type of simulations/estimations I should be understanding?

[Where To Go]

I'm at the airport right now (visit days :) ), and somehow can't open your document. Monte Carlo simulations are just an example of something that people do in Matlab, R, Gaus,..

I did not mean to imply that you should learn to do that or that it's something particularly useful for learning a language.

But as a simple exercise:

create a sample

y = 3 + 5 * x + e

where x ~ N(10, 5), e ~ t(2)

Take a sample of, say, 20 observations to start with.

You can try to study the small sample properties of the OLS estimator.

Regress y on a constant and x (OLS)

Create a 95% confidence interval for beta (the coefficient of x)

Check whether the true value (5) is in the confidence interval you created.

Do the above (simulating, regressing, confidence interval) 1,000 times and count how many times the true value was in your confidence interval.

Show the result as output.

[jakethedog]

so far I have applied to several top ranked schools and have received zero offers let alone interviews... I don't know what I'm doing wrong or what I am missing. Based on my academic profile, am I wasting my time applying to top ranked schools? Its kind of discouraging consider the amount of time I'm investing with each application...Or sometimes I'll finish writing the cover letter for the corresponding school but by the time I try to apply, the posting will be gone. There was an research assistant position at Standford on indeed.com. The posting was gone within a week. The same thing happened for Chicago. I don't know whether that means those positions were on a first come first serve basis or they employed someone that already attends the University

[thagzone]

so far I have applied to several top ranked schools and have received zero offers let alone interviews... I don't know what I'm doing wrong or what I am missing. Based on my academic profile, am I wasting my time applying to top ranked schools? Its kind of discouraging consider the amount of time I'm investing with each application...Or sometimes I'll finish writing the cover letter for the corresponding school but by the time I try to apply, the posting will be gone. There was an research assistant position at Standford on indeed.com. The posting was gone within a week. The same thing happened for Chicago. I don't know whether that means those positions were on a first come first serve basis or they employed someone that already attends the University

 

Applying to several top-ranked schools and not receiving any offers or interviews is not terribly unusual. These positions are very competitive--especially at elite institutions. You will simply need to keep applying as well as to expand the range of positions you're applying to. Rejections are discouraging, no doubt, but you'd better get used to letting them roll right off if you want to apply to PhD programs eventually. There are a lot of people that want these positions and not that many of them available--sometimes a rejection doesn't mean anything besides that.

 

Keep in mind that what any position requires varies and I am by no means anything close to an expert on this. However, I am currently a full-time RA at a top-5 econ school and so I will at least attempt to share what I know (or think I know). Having read the thread a bit and having looked at your profile, I can point out a few things that might be lacking in your profile.

 

Firstly, I don't see any mention of an econometrics background. It's my impression that many full-time RA positions, even if you aren't required to do any model-building yourself, like you to have some experience with econometrics under your belt. That would be something to go out and get--the more mathematically rigorous the course the better. By this I mean that having run some regressions in STATA without really knowing what you're actually doing or what the underlying mathematical machinery is will probably not be considered enough.

 

Secondly, it seems that your programming abilities may be lacking for a top-tier RA position. Many undergraduates have taken some pre-cleaned, tidy datasets that are plopped into their lap, run a couple STATA commands to churn out regression output, and come out of the experience believing they "know" STATA (this could be true of R, MATLAB, etc. as well). I'm not saying this is the case with you, but it sounds like it might be. To be an effective RA, you should be very competent working with large messy datasets programatically. As an example, you mention that you "use STATA do files when dealing with larger data sets." This suggests to me that you might not understand the scale of programming necessary to be a competent RA. You should be scripting everything you're doing, should probably have a grasp of basic algorithms and data structures, may need to have an understanding of parallel computing techniques, and should be comfortable writing hundreds or thousands of lines of code for any one project which encompass complex and creative transformations of data, creation of professional data visualizations, and feature an architecture which renders the code flexible, readable, and easily maintainable for anyone who might work with it in the future. An (at least basic) knowledge of Linux systems is probably helpful too, as a lot of more sizable computing tasks will require you to work remotely on systems that almost certainly will be running Linux. In addition, while someone rightly pointed out that it is definitely not necessary, a knowledge of some lower-level language like Java or C++ will most certainly be viewed favorably and generally will help you write better code in technical languages like R.

 

There are probably plenty of RA jobs that have far more lax technical requirements, but I know for certain that the skills I've outlined here are essential for getting an RA position where I am currently. If you're shooting for an RA job at a top university, then I imagine what I've said here applies at most other highly-ranked schools as well. The good news is that it is possible to teach yourself some of these skills (I did so without any formal background in programming or computer science and I know plenty of other people who have as well). For instance, you mentioned you're going to learn the pandas package in Python--this is a good idea. You should also be learning numpy, and should probably do so before learning pandas. Before that, however, I would suggest trying to get a very firm hold on R. In particular, after learning your way around the basics, I would suggest learning how to use the data.table package.

 

Aside from that, some of your math and economics grades might be red flags at the places to which you're applying. I can't say for sure, but it's possible you're getting screened out for those reasons. It's not worth dwelling on because you can't do anything about that now and, certainly, it's hard to say if it's been a hindrance for you; but, if you're looking for explanations as to your failure to land a top-tier RA job yet, that might be a factor. In addition, many RA positions are filled by students who attended the university and already know the professors who are hiring. Getting an RA job at your alma mater or seeing if any professors you've built a relationship with there can leverage their connections elsewhere to help you would be extremely beneficial. It is my impression that trying to get one of these jobs cold (i.e. without the help of any connections) can be difficult--especially if you don't have an overwhelming technical profile.

 

In summary, I would suggest adding a formal econometrics course, improving your understanding of what working with real data on a large scale looks like, networking with professors at your home university, and patience.

 

Best of luck!