Bad Graphs, or, I Have a BA in English

Bad Graphs, or, I Have a BA in English

Not to brag, but it’s almost impressive how many times I got something so simple so profoundly wrong, in so many odd ways.

I have a degree in Cultural Studies, which means my space of academic comfort is mostly theoretical articles, preferably those with sentences that frequently surpass the length of four lines. So, when I walked into Prof. Piper’s introductory course (LLCU 255: Introduction to Literary Text Mining), and he said some phrases like “run your code” and “statistical significance” I dropped the course while sitting in the class. But the following year I tried again, figuring that I like thinking about language quantitatively, and I like discourse analysis, and, really, I could just take the course pass/fail if it came down to it.

A year later I’m still text mining, and now it’s my full-time job to work at txtLAB. I’m as surprised as anyone that I ended up here, and I’m getting better at coding, but I’m still not very good, at all. I think it’s quite common for folks straddling fields to hit a lot of errors. I think this is especially blatant if you’re coding, because those mistakes actually show up on your screen; they’re also always in red.  It’s hard not to get discouraged when you spend a few hours trying to fix something that seems like it would take others only a few minutes. Recently, I got really frustrated trying to make a specific type of stacked bar graph for a project on gender patterns in contemporary fiction. It was supposed to look like this: 

But I couldn’t get it to work. And the only good thing about messing up plots specifically is that instead of other types of coding, or instead of most errors in any field, is that you are actually forced to see all the completely wrong absolutely useless outputs of your mistakes. They pop up on the sidebar. So, I figure I might as well make the best of the unspeakable number of hours I spent on this and display all these garbage graphs to the whole of the internet. Not to brag, but it’s almost impressive how many times I got something so simple so profoundly wrong, in so many odd ways.

First, I just made these two very unhelpful blocks that show absolutely nothing: 

Then I made a rainbow thing, which is fun, I guess? But also is completely useless. I truly cannot glean anything from this and also what a wild ride these default colours are: 

Then I made this strange small graph that seems to just show the same thing seven times?  But just incredibly small. Oh and yes my y- and x-axis the same, not sure how that happened:

Then, riddle me this, I ended up with another small graph, and the values of the colours of the bars changed? To what values I do not know. And also the Y axis became a collection of seemingly random numbers that don’t align with the size of the boxes?:

Okay, then it started to at least resemble a stacked bar graph. But was based  on raw counts, not percentages.  So that had to change. And the colours were wrong:

Then I GOT THE NUMBERS RIGHT, but for some reason 5 of my columns  were labelled as NA? So, I fixed that and ended up with the first graph I posted:

I’d like to write a takeaway about trying and trying and trying until you succeed, and granted, I did feel some strange code euphoria when I finally figured out this godforsaken graph, but I think more important than the eventual success is being okay with the messiness of the process. I’m trying to learn as much as I can to be better at what I do, but I think a necessary part of that is being comfortable with other people knowing that I rarely know exactly what I’m doing; it’s that admission that allows the space for others to give me the good, challenging, and critical feedback I need to be a better researcher. 

It can’t just be about the final graph. We, of course, ended up cutting it from the project, after all.