Attributions in student writing

Various actions tend to result in students' being accused of plagiarism:

• citing a source but not putting quotation marks around the exact words; 
• putting quotation marks and referring to the source somewhere but not in the right place; 
• using phrases from a source directly without indicating that they are doing so; 
• paraphrasing in a way that mostly uses the original source's language while changing some of the words, usually to close synonyms (a practice that Rebecca Moore Howard named "patch writing"); 
• submitting someone else's work as if it were their own

Of these, the last one is the only clear case of cheating, with the rest usually being errors students commit while attempting to write in new ways. The errors are even more understandable as errors given that citation practices, including how much of another source's wording one may use, vary from one field to another. This means a student might do the same thing in chemistry (re-use of unique phrasing often allowed) and be fine, then do it in art history (re-use of unique phrasing allowed) and get nailed for cheating. Many faculty don't realize these distinctions between fields exist, and many faculty also don't understand the distinctions between different errors and their causes.

None of this is good for student learning.

Changing how faculty react is a long process that involves lots of education (e.g. during TA training, or faculty workshops) and, to be honest, waiting for many faculty to retire.  Changing how students work is another matter, especially since many students learn about source use practices in their general education (GE) classes, and faculty teaching GE classes are, in my experience at least, more open to new teaching practices than faculty teaching in their specialty areas.

Even with GE classes, though, and even with the most adaptable faculty, changes related to writing instruction are hard because writing instruction itself is hard. More than almost any other kind of teaching, it involves looking at each student's work not only for what the student did, but also for what the student was trying to do, and then helping them do it.

One thing I want to figure out is how to help students get better at things that will, eventually, become routine for them -- things like referring to a source when they use an idea or excerpt from it. I also want to figure out strategies for helping them that can be used by faculty who don't have a lot of time to respond to each student's work, since that's the most normal situation in the university these days; most classes are just too big.

As part of trying to figure this out, I have been asking students in my own large-enrollment class (80 students, with TAs) to allude to their source in every sentence that relates to it, no matter how obvious they think it is. I'm not judging how smoothly or stylishly they do it -- just that they do it -- and doing/not doing it is a minor part of their grade for any given assignment. This gives them lots of chances to practice getting it right, with low stakes so it's not a source of high anxiety (which is never good for learning).

I think it's working. In Winter quarter I saw more instances of "according to X" and "X says" in their work over the course of the quarter. The question is, are there really more, or am I just seeing what I'm looking for (a phenomenon called confirmation bias)? And if there is more, how much more? And how many students are improving?

To find out, and to answer some other, more complicated questions  about what students are learning, I asked students if I could study their work. Of the 80 in the class, 63 said yes, and of those, 55 (over 68% of the class) had submitted all the assignments I wanted to study. These are pieces they wrote at the start and end of class, responding to the same questions about the same article, graded only on whether or not they answered the questions. In other words, they were not being graded on attributions in these pieces; this means that if they used more in the later piece, it was because they were getting into the habit, not because they were being careful for a grade.

The next steps were:

1. I coded (labeled) sentences as either clear (has an attribution) or unclear (does not). 
2. My former TA, Lina, coded 10% of the total files. (Lina will be a co-author on anything publishable from this work.)
3. We checked to see if we were agreeing enough. In general, this means at least 80% of the time, but we wanted a higher level. We came in at 82% and decided this meant our definitions of "clear" and "unclear" were still, well, unclear. 
4. We refined our definitions, then each re-coded that 10%. Our agreement went up to over 90.
5. Next, I recoded the rest using the new, better definitions, Lina did another 10% of the total, and we came in well over 90%.
6. Once we had agreement, we ran the software to count to the totals so we could compare the before and after texts.

That's were things stand now. The results look very promising; overall, there was an increase of almost 25%. At the same time, the average for all students at the end was still only around 75%, which suggests I might want to do a bit more instruction around this the next time I teach 011. I say "might" instead of "will" because more attention to that means less to something else, and I have to think about what the trade-off would be and if it's worth it. In the meantime, Lina and I will do some detailed analysis, and Lina will do magical stuff with numbers to make sure our results mean what they look like they mean. Stay tuned for more in the months ahead.

Also, Orhan and I get the dudes and leave for dinner and then the train station soon. Here's Catapult (aka The Fickle Kitty, aka Fick) looking supremely unconcerned:

Cataturk (aka The Ass Cat, aka Ass), is more worried: