In the essay question, one main question includes several sub-questions that are ordered based on the thinking skills, from low (knowledge level) to high (analysis level).

Main question (total: 10 points) Sub-question 1 (knowledge) Sub-question 2 (comprehension) Sub-question 3 (analysis)

Situation 1: Since each sub-question is built upon the previous one, when answering subq1 wrongly, it is likely that the student will also answer subq2 wrongly. Situation 2:It is also likely that the student answers subq1 and 2 correctly, but answers the subq3 wrongly.

What is the best way to make a marking scheme for both situations?

  • $\begingroup$ I'm not sure that the situations are really all that different. The sub-questions are ordered in difficulty, so if you get Q1 wrong you're unlikely to get Q2 correct, and if you get Q1 correct then you're (a little) more likely to get Q2 correct. I would just say give a person a point for each question they get correct, but the fact you're asking makes me think you have a reason for believing this would potentially be inappropriate. Am I right? $\endgroup$
    – Ian_Fin
    Aug 30, 2016 at 12:47
  • $\begingroup$ My focus in more on the second situation. If the learning goal is that students should reach the analysis level, then should we assign points to the underlying lower levels? $\endgroup$
    – ahs
    Aug 30, 2016 at 13:05
  • $\begingroup$ Imagine I have three students: student A is below average, B is average, C is above average. Student A gets Q1 correct, but Q2 and Q3 wrong, so I give them 1 out of 3. Student B gets Q1 correct and Q2, but Q3 wrong, so I give them 2 out of 3. Student C gets Q1, Q2, and Q3 correct, so I give them 3 out of 3. The below average student gets a lower score than the average student, who gets a lower score than the above average student. Such a scale makes reasonable sense, no? $\endgroup$
    – Ian_Fin
    Aug 30, 2016 at 13:11
  • $\begingroup$ I'd recommend you read any introductory text on IRT, and most introductory psychometric texts written in at least the past 45 years. This idea of tests consisting of questions with varying difficulty is extremely common, and this approach of summing the number of correct answers is typical. $\endgroup$
    – Ian_Fin
    Aug 30, 2016 at 13:13
  • $\begingroup$ I still have another question: don't you think it is reasonable to assign more points to the more difficult subq3 than subq1 and 2? $\endgroup$
    – ahs
    Aug 30, 2016 at 13:50


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.