# You haven’t learned it unless you can do it

It’s now day five of the 20-day blogging challenge, which, if you are late to the party, FYI is an idea of Kelly Hines to blog 20 days in a month on topics related to teaching.

“Share any tips for designing/ grading/ giving assessments.”

I have two really good ideas for assessment, one of which I always use and the other I’m kicking myself because I have not done it lately but I’m really thinking of using it again for my next class.

**1. Analysis paper with real data.** I teach statistics, multivariate methods, data mining, stuff like that. As a business owner, I used to say that I would not hire anyone fresh out of graduate school because they could never DO anything. I don’t need someone to prove the Central Limit Theorem for me, calculate sums of squares with a calculator or look up degrees of freedom in a table in the back of some book. It’s 2014 and we have computers here at The Julia Group. It occurred to me at some point that since I teach graduate students each year, I am part of the problem. Now, I always require students to pose a research question, analyze data to answer it and write a paper discussing their method and results.

Details may vary from course to course but what the assignment always includes is REAL data, which means some of it is missing, some is impossible, meaning it was data entry errors or the person just wrote down the wrong information. No one is 992 years old. Data may not comply with distributional assumptions. Your measure may turn out to be unreliable. In all of that, you need to figure how to compute an Analysis of Variance or logistic regression and interpret the output without an answer in the back of the book.

I require the paper to be submitted in pieces, first a draft of the descriptive statistics, then a draft of inferential statistics, then a final draft. One reason for this, unfortunately, is that cheating is rampant at universities, and if you have to turn in lots of drafts, it is going to be difficult and expensive for you to get one of those storefront paper mills in West L.A. to write them all for you.

There is a more positive reason, though. You may (probably will) forget 90% of what was on a multiple-choice final exam that you crammed for. You’re a lot more likely to remember how you solved a problem that you posed yourself, because it’s likely to be of interest to you. For example, a young woman in a recent course wanted to do research on the relationship between obesity and health problems such as diabetes and cardiovascular disease because her family had several members who were obese and had health problems. When I asked whether she’d like to look at BMI or breakdown the sample in different ways, she was emphatic that she was interested in obesity. She used the data from the California Health Interview Survey. Because it was something meaningful to her, I believe she’ll retain a lot more of what she learned than if we had just had chapter tests.

**2. Class notebook.** This is an assignment I used to give and I would run into students years later who told me they still had theirs and used it. This wasn’t just notes but more of a lab notebook detailing in your own words just how you did each part of the project. They could also copy and paste in anything that would help them. The purpose was for them to have something they could use if they ran into this problem on the job. The notebook was THEIRS. How did you compute descriptive statistics with SAS? How did you compute reliability? How did you compute an ANOVA, what type of post hoc test did you use and why? How did you compute a MANOVA? What did each of those numbers on the printout mean? Because students wrote it for themselves, when they needed to do one of these procedures even a year or two later, they could pick it up and use it.

What both of these assessments have in common is that they allow the students to personalize their learning.

They also both take a really long time to grade because I read every page. I think this would be really hard for a large class unless you had a teaching assistant or grader. Even for graduate courses that tend to be less than 25 students, it’s a lot.