# The Science Learnification Weekly (April 18, 2011) – Homework Special Edition

**Posted:**April 18, 2011

**Filed under:**Homework, Learnification Weekly 4 Comments

*This is a collection of things that tickled my science education fancy in the last week or so. My internet was down for most of the week so this week it will all be on one topic, homework.*

**Cramster articles in The Physics Teacher**

*Over the past two months Michael Grams has published two articles on Cramster in the American Association of Physics Teachers publication The Physics Teacher. Cramster is a website where students can go to get poorly worked out solutions to textbook problems (more on this later) or get live help from “experts”. It has a tiered membership where people have some access for free but have to pay to have full access.*

**Cramster: Friend or Foe (free article)?**Grams discusses the website, and what some of his previous students and fellow educators think about it. He also sets up his experiment which is discussed in the second article.**The Cramster Conclusion (not a free article):**His research question is “Could giving students the answers to their assigned homework problems be an effective way of teaching them physics?” His students were provided with both a full membership to Cramster, and two days after homework assignments were given, the full textbook solutions to those problems. The homework was never collected nor graded, but an in-class quiz consisting of a numerical problem similar to one of the homework problems was given to follow-up on each assignment. He used common diagnostic conceptual surveys (Force Concept Inventory in his Mechanics sections and Conceptual Survey of Electricity and Magnetism in his Electricity and Magnetism sections), administered some exit surveys on their use of the solutions and had some follow-up student interviews. The most interesting findings (to my mind) were (1) that students greatly preferred the textbook solutions over the Cramster ones, and (2) the students with the best performance on the conceptual diagnostic surveys typically focused a lot of their attention on the “reasoning” part of the textbook solution when they used these solutions, where the students that did poorly on the conceptual diagnostic surveys completely ignored the “reasoning” parts of the solutions and just focused on the mechanical steps of the solution. The students found that the Cramster solutions often had errors, huge gaps in logic, and lacked any sort of reasoning steps, which is what made the textbook solutions their preference. This article really would have benefited from using graphs as its main method of communicating data instead of gigantic tables and endless numbers within the body of the article.

**More challenges to standard assign and grade homework practices**

- The Grading Dilemma: What’s Effort Worth? Shawn Cornally (Think Thank Thunk blog): “You can stop grading things. You can alleviate your students’ performance anxiety and points addiction by allowing them to engage with things purely for the fun of it. I’m so angry right now for how hippy-dippy this must be coming across.” Go read it.
- What does education research really tell us? Alfie Kohn (author of The Homework Myth and many other books) talks about three traditional teaching practices: behavior change through rewards, assigning homework, and teaching by telling, and discusses research which challenges common practices and/or findings from smaller previous studies. When looking at influence of homework on learning when factoring in other things such as instruction and motivation, researchers found “Homework no longer had any meaningful effect on achievement at all, even in high school.”

**Randomized numbers in pencil and paper assignments (or quizzes/exams)**

- DIY personalized, randomized assignments: Ed Hitchcock (TeachSciene.net blog) discusses how he uses some standard M$ office programs to make student assignments that have randomized numbers so none of the students have the exact same question, meaning that their discussions of homework can focus more on the process than only the solutions. This randomization is common for online homework software such as Mastering, WebAssign, LON-CAPA, etc, but it’s nice to see a quick way to take this into the pencil-and-paper question realm.

A colleague of mine tells all the students about cramster on the first day. He tells students they are free to copy solutions down, as long as they cite them properly. Not sure it’s the best solution, but he is happier then before.

I know of a few people that ask students to cite collaborations and other sources. I have been using Mastering Physics for the past while and set the assignments up so that it is pretty obvious if people are using internet solutions. I don’t ever act on this knowledge, but I do try to look at how these behaviors affect their overall performance (sort of like the Cramster papers linked above) with the plan of being able show future students what sort of behaviors are most consistent with different achievement levels. For Mastering Physics I make sure at least 1/4 of the questions are mine or from some other textbook. Those that use internet solutions do the other 3/4 lightning fast and close to 100%, but that last 1/4 of the questions is a struggle if they even get them done at all. Copying from friends is challenging since most questions use randomized numbers.

Anyway, I think I will ask students to cite who they were working with and any internet sources such as cramster or solution manual that they use. I’m interested to see what sort of data I get and how forthcoming they will be willing to be.

I agree with your statement about tables making the paper clearer. Sorry about that. I had some graphs made, but there were too many and they looked even more cumbersome than the tables. I was hoping for some advice from a reviewer, but got none. I assumed before they published the article it would have to be cleaned up but they just went ahead with it.

Thanks for dropping by Mike. I have come across a lot of data like yours where the tables are cumbersome, but it’s hard to figure out how to reasonably represent it another way.