Ask the grumpies: Should I get a phd? And if so, should I do it in CS or linguistics?

To PhD or not to PhD?

I’m currently between programming jobs, having been laid off from my last job with a good severance package, healthy savings, and a spouse with a high paying job.  I haven’t yet started job hunting and instead have been focusing on getting a masters degree part-time in a program related to artificial intelligence and I’ve been working as a TA for the program.  I have enjoyed the TAing.

I have been thinking off and on throughout my career about switching from industry to a PhD program.  What things should I consider to help me make this decision?

My area of interest is Computational Linguistics. If I go for a PhD, should I apply to CS or Linguistics departments?  Should I apply broadly across the US or stick with the local program here (given I own a house and have a dual-body problem and the local program, though not a top 5 in my field, is decent)?

Should I take an industry job in this subfield prior to applying for a PhD or should I apply for a PhD first?  I love tech jobs, but I’m tired of being a generalist.

Our answer for the first, “should I get a phd” question is similar to our answer for the accountant who wrote in with the same question, with one important exception.  Salaries for accounting phds are higher than for those of accountants.  Salaries of CS PhDs are not that different than salaries of highly skilled programmers without a degree.  The main difference in our experience is in levels of specialization.  (Note:  this is based more on personal experience than on hard data, but there are also a lot of not so great and poorly paid programmers who would be unlikely to get a PhD.  I don’t think anything has been done looking at wages controlling for underlying ability.)  So yes, there’s still demand for CS professors, but there’s also demand for CS PhDs in industry and for programmers without degrees as well.  (My DH with the PhD makes a little less than #2’s DH who dropped out of college to become a programmer.  They both make very nice salaries.)

As with the accountant, we recommend that you work as a research assistant if you can.  This will probably be an enormous paycut from your last industry job, but it will give you a feel for the kind of work that PhDs do.

This paragraph is also still true, but replace “accounting” with “computer science”:

Even with an accounting degree, you get very little choice about where you move to after you’re done. We’re living in places we wouldn’t choose if it weren’t for the job. There’s a limited number of professor jobs in any discipline each year and you have to have a certain amount of flexibility. If you absolutely have to live in a specific city, it’s unlikely you’ll get a TT job there. It’s possible, but not likely. If you are location dependent, see what kind of jobs you can get with a PhD in accounting in industry and/or government (depending on the location).

which means lots of heart-to-heart talks with the other half of your two-body problem and you’ll have to consider selling/renting your house.

Also you will want to check how long it takes to get a CS PhD, especially given your masters work.  How many years are classes, how long do people generally take to finish the research portion?  And so on.

My general impression is that if you’re doing computational linguistics, you’re better off getting the CS degree (or one of the funky specific degrees you can get from places like MIT) than the linguistics degree for the same reason that people who do economic history are better off doing economics than history. The baseline of what you can do with the econ/CS is just so much higher than the baseline for history/linguistics that it’s better to go with the former even if you end up doing the same work with either label.  I could be totally wrong about this, so definitely talk with professors at these programs and look into job placement for people in the programs you’re considering.  But, I do have a friend who dropped out of a linguistics PhD program because she realized someone would have to die in order for her to find a job opening.

If you want to go the academic route, then you would most likely want to apply broadly and, if possible, to go to MIT or Stanford or another top school in your area because that will give you more options later. People at MIT are more likely to actually finish the PhD and go into academia (people at Stanford are more likely to drop out and do a startup and get really rich).  Stanford has better weather along with a higher cost of living.  The major problem with academia is that unless you are extremely good or extremely lucky it is very difficult to choose where you want to live with an academic job.  For industry, the local program is probably fine.

My husband prefers the work he does that requires a PhD to what the people in the same company/field without the advanced degree do.  Even in the tech industry the PhD does seem to be a gateway into less generalist work.  It sounds like you find that rewarding, which suggests that even if you do go into industry the PhD would not be wasted time.

Grumpy Nation, what advice do you have for To PhD or Not?

Null effects are fine, but you need to discuss power!

I like the way that a lot of social sciences are starting to push for publishing more studies that tried something plausible and then found nothing.  Null effects papers tend to be difficult to publish, which leads to publication bias, meaning you’re more likely to find something spurious than to not find something.

BUT.  One almost sure-fire way to find no effect is to have a sample size that is too small to pick up an effect.

If you find no effect, you need to discuss sample size.

If you find no effect from an experiment, then you really need to talk about the power analysis that you did *before* you ran the experiment that shows the sample size you would need to find an effect size.

And if you have a large magnitude that just isn’t significant, that isn’t as convincing as having an insignificant small magnitude or, even better, a small magnitude that flips sign depending on specification.

As the great Dierdre McCloskey says, statistical significance is not the same as oomph.  Or as I tell my students, meaningful significance is not the same as statistical significance.

A true null effect is something that has a small effect size, whether or not it is significant.  And if you find an insignificant null effect, then you have to discuss whether this is a true non-finding or if you just didn’t get enough observations.

Got that?  Null effect = fine, but it has to be a real null effect and not just a bad study.

Ask the grumpies: Favorite class outside your major?

Leah asks:

What was your favorite class outside of your major and why?

#1  German, choral conducting, maybe that one English class where we read mystery novels.

#2  History, probably the British monarchs class because the prof for that one was especially awesome. (I would have been a history major, except the prof that I had for the required freshman seminar was a total a@#$@# and he taught a bunch of required classes, so screw that, and thus I ended up on a more potentially lucrative path.)

How about the rest of Grumpy Nation?

Ask the grumpies: Writing an external tenure letter

Tenured economist asks:

I was asked to write an external letter for a tenure case. Do you have any advice to share? We don’t use these in our tenure cases so I have never even seen an example! How long/detailed are they usually?

The following is based on external letters we’ve gotten in the tenure cases I’ve sat on so far.  We’d love to hear from the Grumpy Nation for people with more extensive experience and with experience in different fields.

There’s a lot of variation in these letters even from economists.

Usually they’re 1-3 pages long (single spaced with extra spaces between paragraphs, 12 point font, TNR, etc. give or take). Here’s what I’ve seen generally:

You don’t have to give a recommendation yes/no if you don’t want to. If you do, it can either be based on, “They would get tenure at [my university]” or “They should get tenure at their university”

You start by saying if you’re aware of the person’s work if you are aware of it, and if so whether or not you know the person personally and in what context. If you’re not aware of the person’s work you can choose to say that or to not say that.

Then you talk about the different strands of literature and put them in context for the committee. Talk about their quality and how they fit into the broader literature.

If there’s other items they ask you to address like teaching or service, then address those as well. We specifically ask for it to be focused on research and fit within the broader community (so potentially service to the profession if they have any) because we’re an R1.  SLACs, policy schools, and business schools might have different things they care about so if there’s something that the specific type of institution cares about you might address that.  Ex. teaching, media visibility, etc.  If there are potential things you might think would be concerning, like lack of single authored papers, you can talk about that as well and why that may or may not be a concern in this specific case.

That’s really about all there is to it.  The hard part is reading through the articles and figuring out their worthiness, especially you don’t have a helpful overview letter written by the applicant that puts it into perspective for you.

Ask the grumpies: Can I really recommend accept with minor revisions in the first round?

Lucy asks:

I am trying to write a referee report on a paper and other than things I know are minutiae I have no comments!  Other than not showing anything causal (which the authors readily admit and isn’t really necessary for their question), I have no major criticism of anything they have done. The outlet is probably appropriate. So…do I seriously recommend publish as-is? Have you ever done that? It seems like such a cop out.

I think I have recommended some things be published with only minor revisions.  I know I’ve typed out under the Major section:  “I have no major concerns.”  And as an editor I’ve definitely gotten people making that recommendation, “Accept with only minor revisions”, even in the first round.  I just did one, in fact, that came back with “accept with minor revisions” from two reviewers in the first round.  And then I read it and was like, yeah, they should cut out that one section and see a copy-editor, but this is definitely an accept with minor revisions.

What you need to do so that the editor believes you (IMHO on the receiving end of these reports) is to explain in the cover letter why you think it doesn’t need revising.  So you say what you told me.  It exhaustively documents info, it doesn’t show anything causal but the authors are upfront about that and you don’t think it is necessary for them to show causation given the topic, the outlet is appropriate, etc.  The authors should be commended, etc.

It’s not enough to say, “accept with minor revisions” because then I’m all… should I trust you, or are you just lazy?  But if you can say why the paper is interesting and important/appropriate and anticipate problems that you don’t think are problems, then your letter is really helpful when I have to compare it to someone who, say, believes the paper should be rejected because it isn’t causal.  I had a situation like that once with two extremely enthusiastic reviewers and two who wanted to reject the paper outright and one of the rejects and one of the minor revisions were useless because they didn’t tell me anything useful.  If the second accept with minor revisions had told me why to accept, then the decision would have been a lot easier for me.  (Or if the other reject had said something other than, “this paper doesn’t cite [my papers]” even though it cited a literature review that contained said papers.)

Please use more topic sentences

In your technical writing.  Please!

What is a topic sentence, you ask?  Since they no longer seem to cover that in third grade…

A topic sentence is the first sentence in a paragraph that provides the main idea of the paragraph.  Essentially it introduces a paragraph and summarizes what the paragraph is going to say.  It isn’t, “Now we turn to Table 2”.   It isn’t, “[Author (DATE)] studies X.”  What does Table 2 say?  Why is it there?  Why are you talking about Author (DATE)?  What is the relationship to your paper?  Convey this information in the first sentence of each paragraph.

The topic sentence should tell you why that paragraph is there.  If you don’t know why that paragraph is there, then maybe it shouldn’t be.

This PSA brought to you by a grumpy rumbler who has had to do waaay too many referee reports recently.

Stocks and bonds, Writing and outreach

I had an idea.  Follow me, here:

For academic careers, writing is like investing in stocks.  Outreach and translational research are like investing in bonds.

Stocks and writing:  Get lots while you’re young.  You need to write prolifically enough to get tenure, and gain the national or international reputation you need for those outside letters.  Spread your name, become known in your field.  Start early.  Because the return is uncertain, put a lot of writing out there in the world (and buy stocks).  Stocks are a good investment when you have a long timeline until retirement; you have time to weather the ups-and-downs of the market and can have a higher tolerance for risk, in exchange for possibly higher returns.

Bonds and outreach/translation:  These are more effective when you’re older.  When you’re more experienced in your field, you have more experience and a reputation that you can leverage for influence.  Research-wise, you’ve got a better idea of what works and what’s worth developing further, as well as potential pitfalls and objections.  You also know people who can help spread your ideas.  You may have more time to devote to making the world a better place.  When you’re closer to retirement, you also want the safety and security of bonds: potentially lower return, but steady.

In financial investing, as in an academic career, you’ll need a balance and variety throughout your life.  You might want to be doing both of these things (and more!) at all times, but in varying ratios.  Diversify and rebalance your portfolio and life.

This idea: off the wall, or right on target?  Tell me, Grumpeteers.