9

Preliminaries

I am giving my first lecture at university. The topic is about Big Data and takes 40 hours (1 hour = 60 minutes). I am wondering how to test the students. I am at this university for the first time and I have no idea how full the student's schedule is and how good they are. I am also not sure how many students I will have, but I think it should be between 20 to 30. I think they are going to be at the end of their Bachelor / beginning of Master.

Options

I see three options:

1. written exam

It's the least amount of work for the students (only learning) - which might be a good thing if their schedule is very full.

Disadvantages: They learn the least I think.

2. presentation

I'd let the students present a topic - probably a software technology that wasn't part of my lecture so far.

Advantages are that

  • they learn more,
  • they learn research technologies (I've been working in the industry for a while and researching how technologies work is an important part of the work in this area, so it's a real-live skill in itself).

Disadvantages are

  • that it might get difficult for me to judge their effort in terms of grades,
  • it might take quite some time for them to prepare (issue if their schedules are really full),
  • they might not be able to find good information and might not have the skills to put the information into a good presentation.

3. project

I'd have them do a small software project.

Advantages are that they would gain practical experience.

Disadvantages might be that

  • these take quite a lot of time,
  • it might get difficult for me to judge their effort in terms of grades,
  • the projects might fail if they are not proficient enough, which might be rather frustrating.

Question(s)

I tend to take either 2 or 3 since - back in the day - I learned most from those and little from tests. However I fear that those might fail, because the students don't have the time or skills to excel in those.

For 2 and 3 I would put the students into groups.

I might be able to decide after the first week of lecture, but not just before the course ends.

Going out to fellow professors, lecturers etc.:

  1. What are your experiences? What options would you suggest? What do I have to consider?
  2. If I'd take 2 or 3: How many students should a group have approximately (minimum, maximum)?
  3. Can I simply ask them how they would like to get tested?
Make42
  • 979
  • 8
  • 21
  • 2
    I'm not clear. Is this a single lecture or a lecture course? How many student study hours does it represent? I think we need to know more context before we can answer... – Brian Tompsett - 汤莱恩 Jun 17 '16 at 11:19
  • What exactly is it that you are teaching? Is it a more theoretical topic or a practical one that would involve more hands-on coding? – mbaytas Jun 17 '16 at 11:49
  • @BrianTompsett-汤莱恩: Updated question: 40 hours = 40 * 60 minutes. – Make42 Jun 17 '16 at 13:33
  • @mbaytas: Updated question: It's about Big Data. It's both, theoretical and practical. It's about architectures, data bases, processing engines, etc. – Make42 Jun 17 '16 at 13:35
  • I was a BSc student at applied physics-informatics. Our exams were written or orally, but we had for each subject to do a project(individually or in teams), and a more theoretical homework individually. The percentage for the grade were 10% course presence(quizz in some cases)+30%final exam+30%partial exam+20%project in general. – Nikey Mike Jun 17 '16 at 15:25
  • For 40 hours of class (contact hours), how much out-of-class time can you use? In my university 40 contact hours would imply 360 out of class study hours; thus a large implementation could be expected as an assignment..... – Brian Tompsett - 汤莱恩 Jun 17 '16 at 17:48
  • In my under-graduate data analysis course, we actually did 1 & 3. We started by being tasked to build a project centered around some data (the datasets were provided or had to be approved) and ended with a summary/methodology report - as that helped to reflect the industry and provide some practice for writing research papers later. Also the project was allowed to fail (or produce no results) but we had to be able to explain why in report. – LinkBerest Jun 17 '16 at 18:29

5 Answers5

4

  1. My experience: One thing I do see in your question is a sensitivity to your students work load. In my experience when I have taught courses involving less "homework" the course may be looked at as "easy." In higher education, students expect a challenge and teachers need to bring it. Worry less about their workload and more about what you want them to learn. If all else fails, grade on a curve.

  2. Groups are hard. It's an unpaid environment. Slackers will slack, pushovers will get pushed, and a group of real team players will rarely shine through. If you are going to do groups, let it be a part of the grade, but not nearly the full grade. Information comprehension/retention should be tested individually. Working in groups is a whole other set of skills that some thrive at and some falter at, and that's not the class you are teaching(I assume.)

  3. You should know what your students expectations are and they should know yours. Have that conversation, but come prepared.

Simon Andreas
  • 41
  • 1
  • 2
    +1 for #2. My university pushed working in groups, because the workplace involves that, but group work in school has almost nothing in common with group work in the workplace. – Kathy Jun 17 '16 at 14:08
  • Always disagree with grading on a curve, except possibly in the first trial run of a course. It's a total pox. – Daniel R. Collins Jun 18 '16 at 02:50
3

What options would you suggest?

As you only have 20 to 30 students, I'd suggest giving them individual oral exams.

Advantages are:

  • This format tends to show grade-level granularity very, very quickly.
  • It allows you to be a little free-ranging to assess the A+ vs A or D- vs F levels.
  • It takes about as long as marking individual written assignments.

Disadvantages are:

  • Students who haven't taken this form before tend to freak out.
  • You can't really do it in a group scenario.
  • If you haven't done this before (as a professor), it can be a little strange.
  • You need to take better notes to allow you to verify / remember at a later date.
Community
  • 1
Peter K.
  • 4,938
  • 1
  • 26
  • 37
  • 1
    This can also be presented to the students as preparation for interviews in later job market (or when applying to graduate school). – LinkBerest Jun 17 '16 at 18:35
  • @JGreenwell : Yes, it certainly can. You can run through the exam curriculum as a sort of list of required and nice-to-have list of job requirements. – Peter K. Jun 17 '16 at 19:29
  • Is it properly documentable? If someone challenges/grieves the grade, can you properly defend it? – Daniel R. Collins Jun 18 '16 at 02:52
  • @DanielR.Collins certainly. If I were to do it today, I'd video it. But you can get the student to sign off on your notes at the time of the examination. Just takes preparation and a consistent set of questions for all students. – Peter K. Jun 18 '16 at 03:09
2

I'm not sure whether what you are proposing is one kind of assessment out of many, or whether you intend to assess students in only one way and you are asking what that way should be. So first I just want to say, assessment in a university class should be varied. In particular you want to make sure you have both formative assessments (where you gauge student learning while it's still in the process of being "formed") and summative assessments (which are done at the end of a learning process). Having just one of these without the other is not going to provide you with sufficient information about what and how well students are learning.

All three of the things you mentioned are legitimate summative assessments. Have you thought about formative assessments too? Some possibilities include:

Regular quizzes can be formative as well, although formative assessments often work better when they are ungraded, or graded on a Pass/Fail basis rather than something more complicated. The purpose of formative assessment is to gather data about student learning so you can know, in the moment, how students are doing and make teaching adjustments before it's too late.

As far as summative assessment goes, I think any one of the three things you mentioned could be useful. Traditional tests do tend to be more auditive than educational -- they only tell students what they did wrong, rather than offer an opportunity to learn. So one wrinkle on the traditional test you might consider is having a revision policy. For example if you give three tests during the semester, allow students three revisions (so they can revise Test 1 three times if necessary, or each test gets revised once -- their choice) that must be accompanied by a detailed failure analysis that delves into how they prepared for the test, what was missing from their preparation, what they got wrong and why, how they fixed it, and what they will do differently next time. Your acceptance of the revision is contingent upon their turning in a failure analysis that meets your standards whatever those may be.

Or, you might set up a system in your class where students can choose from among the summative assessments you have in mind. How that might be done depends on the specifics of your grading system but the teaching and learning research shows that giving students choice increases motivation which then improves learning outcomes.

Finally if you are concerned about grading practices on non-standard items like presentations, I encourage you to consider specifications grading which is predicated on grading items Pass/Fail on the basis of whether they meet professional standards or not -- no points, no hair-splitting.

Robert Talbert
  • 1,693
  • 12
  • 14
0

Here's a partial answer (too long for a comment).

One consideration I don't see here: How secure is this method against cheating (having someone else write the paper)? I would recommend at least some in-class tests as a check that the individual student actually knows what they're talking about. On top of that, I'll note that your non-testing options already have the longest list of significant and real disadvantages.

Also: Definitely don't ask them up front how they like to get tested. This sets the tone that the students get to run the course all semester, and you definitely don't want that. As the instructor you need to set a tone of authority and that you're prepared and an expert at the start of the course. If at the end of the semester you want to survey students on what they thought helpful or would suggest changing, then that's fine (although hopefully you'll have a good yourself at that point on iterative improvements).

Daniel R. Collins
  • 36,531
  • 13
  • 99
  • 143
  • 1
    I myself had a prof once that said at the beginning of the course that it is going to be rather experimental and we will know as much as he does at the end of the course. He also said that the course will be as good as our suggestions and participation. In the end it was one of the best courses I ever took. So much for "authoritative" behavior ;-). What do you think? – Make42 Jun 17 '16 at 13:43
  • @Make42: I think that this probably varies by educational institution. Where I am in the U.S. community college system, I very quickly learned (and see every day for other instructors) that this is uniformly disastrous. I'm gladdened if you teach in an environment where that's workable. – Daniel R. Collins Jun 18 '16 at 02:54
-3

I think, If you want to check a student in cs, you should ask architectural question like if you have to make train booking system how you will design complete architecture.

aryanathakur
  • 1
  • 2