Smartest thing I've done while dissertating

Hiring a statistics grad student to run my numbers. You know, the ones I've been struggling to figure out since SEPTEMBER? The ones that could've been done then, if my husband I hadn't been so reluctant to hire them out, and therefore potentially have been done writing this thing by now?

I didn't do it earlier because this is part of getting your degree, you'll need to know them in the future, yadda yadda yadda. Guess what? When none of the professors know it, that's a sign that you won't actually have to know it. And when you aren't familiar with research methods and setting up an experiment so you know what stats to run ahead of time, you end up with much more complicated tests to run. Complicated things that even your psychology committee member, who teaches a stats class himself, can't do.

Then it's time to bring in the big guns. Well, not the really big ones - I checked with an online company that wanted to charge me $2400 to do it. $2400! What a joke. Then of course they said they could do less for half price (which is what I wanted them to do anyway - I don't really want someone else to write my results chapter), but that was still way over the top. Enter stats grad student - smaller guns but the bullets are just as effective.

The happiest day of my dissertation so far was when she emailed me the results.

Roller coaster

I'm back on campus and have taken these past couple days to meet up with all three of my committee members. Within every meeting there have been ups and downs, but here's a broad overview:

Monday
UP: My adviser finally got me comments on my first two chapters.
DOWN: The first chapter needs a ton of work.
UP: The second chapter is basically good to go!

Tuesday
UP: My within-dept committee member was quite helpful going through my statistics with me.
DOWN: That afternoon I had another statistics breakdown, and (again) was unable to follow her suggestions due to requirements of the tests.

Wednesday
UP: First thing this morning I found a way to make one test work! It showed exactly what I wanted to be significant!
DOWN: My outside-dept committee member informed me that none of the statistics I'd done were valid, because they were all violating an important assumption. The tests I've been fretting over for the last 4+ months? Basically toss 'em.


And that's not even including the personal side of my trip so far.

Who knows what tomorrow might bring? Hopefully a meeting with yet another guy who can finalize stats stuff. And oh yeah, giving a colloquium to my dept and presenting my research to all them for the first time. I should really put together that talk, huh?

Anything but that!

Wow, it's been a busy couple weeks and will continue to be. As I'm simultaneously writing my results draft and putting together my powerpoint for next week's talk*, I've come to the dismal tentative conclusion that I need to rework some of my statistics. These were the stats that I thought were finished and good to go. And yet, after being reminded of a different method this past weekend, I'm afraid it's more appropriate for my work. (In a nutshell, instead of testing F1 and F2 separately, as I did, testing the distance between them.) Right now I'm stalling and trying to think of reasons why I shouldn't have to do that. So far all I've got is that I won't have time to do it by next week, and if no one in the audience suggests it, maybe I can get away without it.

And yet... for all my wanting to hurry and get this done as quickly as possible, I just don't know if I can let myself get away with it. I don't want to sacrifice quality for speed. (Actually, I want to be okay with that, but I just can't.)

And while I'm complaining about statistics (as usual), do you mind if I whine that I'm still not sure about my other numbers? Even the professional statistician didn't actually answer my questions - he just suggested I present what I have and then try (yet another) test. I think this is actually a case where I will base my decisions on what the audience says. If they don't like what I did and suggest something else, I'll do it. If they can't think of a better way, I'll probably say forget it.

If I could go back to Undergraduate Me and give her one piece of advice, it would be to take as many statistics classes and psychology research methods classes as possible. (Actually, I'd probably talk to Undergraduate Me about a few other things as well, but those are unrelated to linguistics.)


* I've found giving presentations on your work as it's in progress to be a helpful way to get a handle on what you're doing. For example, in order to create my presentation, I need to write the conclusions for these sections (something I've put off doing). In figuring out what the most important things to say are, my presentation and my written draft inform each other.

Sweet sweet affirmation

You know how I've been moaning about my issues with statistics for months now? A professional statistician doesn't think I'm doing so bad! What?! Whoo-hoo! An excerpt from his email:

Hi Elizabeth,
What a neat study! You have a good handle on stat methods. ...
It really was a fascinating study. We'll get back to you soon.

Did you read that? He thinks I have a good handle on statistical methods! And that my study is cool! I can go for months on this little bit of praise!

The productivity continues!

I made my goal and got my first two chapters to my adviser by the end of last week - yay! Nothing says "Happy Friday" and "Happy end of the semester" like getting an email with a huge attachment and a desperate plea to read it and get back to me before Christmas!

Now the work continues on my results chapter(s). I'm in contact with friends of my father-in-law, who just happen to be statisticians in the pharmaceutical industry, and who are glad to help me out by looking over my results! Finally! Some much-needed reassurance. I should hear from them soon about my listener results, so I'm trying to get a lot done on my speaker results before that.

Isn't it funny how my productive streak coincides with the time of the year that I would most like to be slacking off? Oh well, gotta strike while the iron's hot!

Further proof of my nerdiness

I was so excited to create several clear charts and a big beautiful vowel space this afternoon! Things are starting to come together and instead of just cranking out numbers, I'm starting to think critically about what these things mean! Whoo-hoo!

Visual aids are the best

Especially when you do them right.

When I was compiling my results, I basically just copy/pasted from the SPSS output. I ended up with charts like the one below.

Yes, it was very confusing and I had to include an elaborate key below it - but I didn't know how to change the graphs within the program, and didn't really think much more about it. Don't those programs know best, anyway? Won't they always give me the perfect output?

No?

Thankfully my professor suggested I revise my plot, I took the time necessary to create it, and now I have this.

Much better! I can actually learn from this one!

One thankful grad student

I was thrilled to get an email last night from my professor. All my frustration was flung away as I skimmed her several pages worth of comments on my research. My jaw dropped when I saw her brilliant "possible interpretations" and suggestions for further analysis. It's all I can do now not to dance for joy! Thank you, Lord!

As if all that wasn't enough, my committee member thanked me for continuing to contact her, and apologized for any inconvenience in taking so long. I am brimming with happiness and relief!

"So what do I do now??"

I try to write a tiny bit each day about what I've done that day. Usually I just jot down my thoughts as I'm going through things, so that I can look back and remember why I did something a certain way. Other times I think this might be the start of something brilliant, so I'd better write it down so I don't forget it! Sometimes I just write anything down so that I can pretend I actually did something that day. And finally, many times I just write down my complaints.

Yesterday's note ended with a melodramatic "So what do I do now??"

This morning Soothing Encouraging Me responded to Freaked Out Me.

"What you do now is breathe and reevaluate. Go through and think again about what questions you want your data to answer. Then take a break and work on something else, like writing a different chapter. Then jump back to where you were having difficulties with renewed vigor! You are smarter than this dissertation! You will prevail!"

And yes, I actually wrote that down.

It's November

And I'm still not done with my statistics. Excuse me while I bang my head against my keyboard.

fjkjfijl.hjki9oy7;egk'/hijlio;oyghilrkuymblru

What kind of pesky student

... calls their professor at home on a Friday afternoon, when that professor has been out sick for two weeks, to beg them to please look at their data?

This kind. It's not something I'm proud of. But multiple unanswered emails leave me feeling more and more desperate, until I do something rash.

Sorry.

And thank you to my kind professor, who is reviewing my work as I type and calling me back shortly.

Celebrate!

I'm tired of complaining all the time. So I would like to celebrate the fact that this afternoon I ran some new statistics (on a different aspect of my data) and they worked! Hallelujah!

Yes, I actually belted out Handel's "Hallelujah Chorus" when it happened. I sounded like this:



Or something close to that. I was missing the horn section.

Slow going

My poor old laptop is painfully slow. When I bought it I thought it would last me through all five years of grad school - ha! Last year my husband the scientist convinced me to get a new one, and I'm so glad he did. Because trying to do my research and write my dissertation on that thing would be, as I said, painful. I'm actually writing this whole post as I wait for it to open up a document within a folder. Yeah, it's that slow.

But I shouldn't rag on it too much, because I was able to put a trial version of SPSS on it, and it worked the first time! Yesterday was a really productive day - I was able to run all of my tests on that, despite the slow going. All the chi squareds for independence were disappointing - very few significant findings. But my binomial tests were beautiful, with most p values registering at .000!

I was also able to talk to my adviser yesterday and discuss the results I did have. She suggested I make sure and call my committee members (not just email), because they need to be constantly reminded. She's told me more than once to just keep calling and bugging her, or I'll never get in touch with her. It's weird to have to pester the person who's supposed to be guiding you, but you get used to it. That doesn't mean I like to feel like that annoying student who keeps calling, but I'll do it out of necessity.

Oh, the file finally loaded! Off I go!

I thought I'd be done with this a month ago

Actually, my plan was to be finished with data gathering and statistics by the end of August. Data gathering was done. Statistics still are not. Who knew they'd be this much work?

I do have to say I'm learning a lot. Depending on what I do after graduation, that may or may not ever come in handy for the rest of my life. Maybe I'll tell myself that I'm not just learning about specific tests, but learning valuable lessons in perseverance.

I think I can see the light at the end of the tunnel, though. I found a trial version of SPSS that I can put on my old laptop. So today I created a bunch of data files and put them on my jump disk, so first thing tomorrow I can plug those in and start testing! I'm hoping I can finish that all in one day, so I can type it up on Friday and send it on out to my committee members. Time is getting short after all!

Sliding

I didn't post yesterday. Did you notice? It's because it was MY BIRTHDAY!

At first I told myself that the best gift to give myself would be a good morning of solid work - then I would feel great the rest of the day! Then I decided that I should really spend some time indulging in the blogosphere and facebook. And then I'm not sure what happened after that, but it was lunch time. And then I did work for an hour or two, and then got a call from my friend Jenny and we always talk for a while. And then I had to hurry just to get in a run and shower before it was time to go out for dinner! So yeah, not much got done in there.

Oh, and remember how I said I didn't mind family responsibilities interfering with my work? I still don't really feel guilty about that. But it turns out I only like to be helpful when it is on my schedule. I need to work on that.

I don't think I mentioned that I did hear back from one of my committee members! He suggested a logistic regression, so I ran that. I see some very nice p values, but I don't know which (if any) of them are applicable. Today I also ran some loglinear models. I think they said I have significant three way interactions. I'm not really sure what that means, either. I think I'm trying to complicate things too much.

I really need to create my NWAV poster. But that's based on my listener results. And those are the very statistics I'm struggling with. But I will prevail!

Dear SPSS,

We are no longer on good terms. I used to think you were the answer to all my statistical problems. Actually, I still believe you hold the key to beautiful p values. You can make my research significant!

But then you had to go and hold my Mac against me. I'm sorry. I honestly would not have purchased this computer if I'd realized you weren't fulfilled by it. I didn't know you could give me more on a PC. That "exact" button is all I'm looking for from you now. And you can't give it to me here.

I'm not sure if we can still be friends after this. It's going to take time for me to reach out and meet new people, to find someone who has your 16th (or later) version installed on their laptop who is willing to let me use it free of charge. Because we both know you don't come cheap.

This is painful for both of us. I'd like to say it's not you, but, well, it is you.

Your former friend and fan,
Elizabeth

How exact?

Well, things are looking up. I sat down all refreshed and ready to do the Chi Square tests. I made beautiful little tables and was able to put everything into SPSS. I'm feeling very happy about that.

My book walked me step-by-step through each click. I realized I needed to do an extra step at the end, and went back to do it. Then I realized that the button I was supposed to push was MISSING. It was not there like in the handy picture. It was nowhere to be found.

I decided not to panic and went to the help menu. That's where I saw this:

Can you read that? It says "Note: Exact Tests is only available on Windows operating systems."

Of course - that's what I get for having a husband who convinces me to get a trendy Mac. (You like how the blame is immediately shifted?) Still no problem, I thought. I'll just put this on my old PC.

So I went to my university's IT website and then it hit me. In order to put it on my PC, I have to go to the IT building and use the installation cd. The IT building that is currently 1,590 miles away. (Yes, I just mapped that online.) That's a problem.

Over 200 pages later...

And I still can't figure out what kind of statistical test to run on my listener responses. Ugh. I guess I'll just do a bunch more Chi-Squareds and call it good. Then if that's not correct, at least I can ask my committee what would be more appropriate. In the meantime, I made a pretty outline of a table to show them when they tell me what I did was wrong :) Then again, they're the ones who told me to just do that test in the first place, so maybe just maybe I'm worrying over nothing and it will be right!

On the plus side, I have finished my statistics for the speaker results! And I have several nice p < .001, which makes me quite happy. I spent a couple hours yesterday agonizing over which tests to use for these, then as I was browsing through powerpoints I found online, I discovered that while in theory it's better to use a certain type of test (the one I couldn't figure out), in practice most people just do this other test twice (which I had already figured out and successfully ran). Sweet! So I did, which is how I got my results. Now I just have to be able to explain exactly what they mean.

To all of you professors out there...

who put your powerpoint slides from class out on the web, free for anyone to learn from:

THANK YOU.

Rejoice!

So many reasons to rejoice on this beautiful afternoon!

1. My LSA abstract was accepted! The Linguistic Society of America's annual meeting in January is the official big conference for my field, and it looks really impressive to have presented there. It's also where all the academic jobs interview, so if you're planning to graduate, you basically have to go whether you present or not. I got accepted as a poster, which is great. I know in some other fields, everyone does a poster; but in linguistics, a poster is a step down in prestige from a presentation. But I'm actually quite happy doing a poster, because (a) only people who really actually are interested will check it out and talk to me, and (b) I wasn't expecting to be accepted at all :)

2. I got the book I've been waiting on!


I was hoping and praying that it would be able to give me the instruction I needed to do the appropriate statistics for my research...

3. And so far it has! Today I was able to run the mixed method ANOVA on my duration data. It was an A x (B x C), if that means anything to you. It does to me now! And two (of the possible six) factors/interactions were significant! Whoo-hoo!

Now I just have to figure out what exactly that means :)

4. It's Friday! And I know this isn't the case for many grad students, but I take the weekends off! In fact, I've got to run and shower now - I have a hot date with my husband tonight!