Social scientists, can you give me a brief explanation of coding?

January 15, 2020 4:46 PM

The character in my novel is a sociology grad student who's working on her dissertation. The year is 1991. She's conducted qualitative research, and I know she has to code her interviews, but I don't exactly know what that entails. Can you explain what this is?
posted by swheatie to Society & Culture (9 answers total)

Linguistic anthropologist here. I’ve done lots and lots of this. It involves combing through your meticulously transcribed interviews or field recordings of naturally occurring discourse or elicited data and marking each instance of the phenomena you are examining or quantifying. In my case, for example, for my dissertation I meticulously transcribed several hundred hours of both formal and natural recorded data (including phonetic and poetic/performative properties like tone of voice or pauses or phrase intonation etc). I then went through the thousands of pages of transcription specifically coding for different “reported speech” constructions — direct and indirect discourse, and as my work eventually showed, and ever more granular set of options in between direct and indirect discourse that were idiomatic for the dialect I was analyzing in ways that spoke to larger relationships between grammar and culture. To do this meant coding also for pronoun choices, verb tense, and spatial-temporal deixis, which along with tone of voice were all variables that systematically clustered into recognizable discrete quotative constructions — ways of saying what other people (might have) said.

This was done in my era (early 1990s) by editing a massive word processing document with systematic codes that referred to a more elaborates key (“6A for verb tense shifted”) and then put into a spreadsheet to look for consistencies, although I didn’t do anything more quant than that many people did and do. I was layering this into a more qualitative ethnography of speaking.
posted by spitbull at 5:06 PM on January 15, 2020 [13 favorites]

Themes Don’t Just Emerge — Coding the Qualitative Data (Erika Yi, Ph.D., Medium) includes "A Quick Guide To Qualitative Coding"

Gorden, Raymond (1992). Basic Interviewing Skills. Itasca, IL: F. E. Peacock. Coding Interview Responses

Elliott, V. (2018). Thinking about the Coding Process in Qualitative Data Analysis. The Qualitative Report, 23(11), 2850-2861. Retrieved from https://nsuworks.nova.edu/tqr/vol23/iss11/14

An Introduction to Codes and Coding, The Coding Manual for Qualitative Researchers, Johnny Saldaña.
posted by katra at 5:09 PM on January 15, 2020 [4 favorites]

The two-sentence example they used in school to give you the intuition: You know how I can say, "Okay," or "Yeah," or "Uh-huh," and you'll understand all of those to mean, "Yes"? That's coding.

If you're coming at this from a more quantitative background, it's basically feature extraction.
posted by meaty shoe puppet at 5:38 PM on January 15, 2020 [3 favorites]

What spitbull said. A couple other things to consider:
Your coding varies depending on the purpose of the work, so you might for example be coding linguistic, structural, or meaning-based data elements.
Many people nowadays use software to code/organize their codes. NVivo is the most popular choice in my discipline. Some people also hand code using printed paper & highlighters/pens, if they have a smaller amount of data.
posted by DTMFA at 5:42 PM on January 15, 2020

Here's a video. I did coding for credit and pocket money in college in the early 90s, in the psychology department, that looked a lot like the examples shown. A lot of ours were written questionnaires, occasionally there were tape recordings or videos.

We often used printed spreadsheet-ish grids to write and tally codes, but I have no idea if that was a homegrown system in that department or if that was some kind of standard at the time. I suspect the student coders were the first-pass team.
posted by Lyn Never at 5:50 PM on January 15, 2020

I did a lot of coding of qualitative data (read: transcripts of children talking) as a research assistant in psychology and inter-rater reliability was always a big deal for us. We were coding for expressions of various ideas and concepts so it was super open to interpretation (especially because kids often speak in word salad).
posted by soren_lorensen at 6:08 PM on January 15, 2020 [1 favorite]

All respect to people doing that, but it seems unlikely to me that a soc grad student is going to be doing the kind of coding that linguists do, because [janet] not a linguist. [/janet] Not strictly sociology, but I did some similar arguably political-sociology coding for Dan Lipinski's dissertation and later book on congressional communication because he'd gotten a small grant from nsf and could throw a couple grand at other grad students.

I had a couple of boxes full of (IIRC randomly selected) mass mailings sent by MCs to their constituents. Dan had put together a worksheet to fill in for each one. They centered on the thing he was interested in, MCs that were negative about Congress itself, but included a fair amount of other stuff in addition to the basic identifying info because why not while we're there.

So I would pick up a mailing, pick up a blank worksheet, and start. Some basic identifying information first -- this is district XX and mailing number YYYY. Is the mailing single-topic, few-topic, or many-topic like a general newsletter? Is the mailing policy-oriented or credit-claiming oriented? Does the mailing include any negative statements? If so, are the negative statements directed towards an individual, a party, a simple circumstance that's happening like a natural disaster, a committee, or the entire institution? Other stuff too that I forget, including a bunch to more precisely nail down how it was being negative. Enter all that on the worksheet, then onto the next mailing and the next blank worksheet.

This would have been between 1995 and 1997. The work I did was entirely with paper and on paper; Dan would have entered it into stata later. Your graduate student would be doing something similar for the interviews. Either walking in with a worksheet to fill in live, filling them out immediately afterwards, or working from recordings or notes taken immediately afterwards.

It would matter less but not zero for stuff being coded by a single person, but in this case there were two or three of us doing the coding, so there was some smallish percentage of the mailings that were in everyone's boxes. Dan could then see whether we all coded the same mailings in the same way. Your solo grad student would still *want* to hand a small randomly-selected percentage of material for someone else to code to see if they code it the same way, but this would be difficult to do without money and might be impossible given anonymity guarantees to respondents.
posted by GCU Sweet and Full of Grace at 6:41 PM on January 15, 2020 [2 favorites]

Many linguists consider ourselves to be social scientists.
posted by spitbull at 9:09 PM on January 15, 2020 [1 favorite]

That said yes many non-linguists will code for specific content, themes, tropes, etc.
posted by spitbull at 10:03 PM on January 15, 2020