When an Exam Doesn’t Go as Planned: Questions to Ask Before You Curve or Toss

Author: Gabi Likavec – Teaching and Learning Consultant,  Office of Curriculum and Instructional Support

Every instructor has experienced it.

You grade an exam expecting the usual range of scores, but instead you see something unexpected: a cluster of low scores, one question that almost everyone missed, or a distribution that just feels… off. Suddenly, you’re faced with a series of decisions that no syllabus policy fully prepares you for. Do you curve the exam? Throw out a question? Give everyone credit? Or trust the results and move forward?

Moments like this can feel uncomfortable, but they’re a normal and valuable part of teaching. Exams are tools for evaluating student learning, but they’re also tools for evaluating our questions, our instruction, and how well the assessment measured what we intended.

Educational research refers to this process as item analysis: reviewing individual questions to determine whether they accurately measure student learning (Haladyna & Rodriguez, 2013). While the term sounds technical, the basic idea is simple: look closely at the data before deciding what to do.

Below is a series of questions I often ask myself when an exam result raises concerns -- before I make any adjustments.

What does the overall score distribution look like?

First, step back and examine the overall pattern of scores. Assessment researchers often emphasize that exams should differentiate levels of understanding, meaning students with stronger mastery should generally score higher than those still developing their understanding (Nitko & Brookhart, 2011). If all scores cluster tightly together, the exam may not be distinguishing levels of learning very well. If the average is dramatically lower than expected, it might signal a mismatch between instruction and assessment difficulty. Looking at the average score and distribution first helps ground decisions in evidence rather than reacting to a few surprising scores.

Ask yourself:

• Is the distribution roughly what I expected?
• Did the exam distinguish between stronger and weaker performances?
• Is the average dramatically lower or higher than usual?
• Are scores clustered tightly together or spread across a larger range?

Would a curve address the real issue?

Curving can feel like a quick solution when an exam average is lower than expected. But it’s worth pausing to consider what the curve will actually accomplish.

Reflect on:

• Is the issue with one or two questions, or the entire exam?
• Would adjusting specific items solve the problem more precisely?
• Would a curve maintain fairness across students?

Curving can be appropriate when the entire exam proved more difficult than intended, and the overall average is significantly lower than expected. But if the issue stems from a few problematic questions, adjusting those items individually may be the more transparent and targeted approach.

Is it the question?

Blackboard provides a question analysis report that shows how individual questions correlate with overall exam performance.  When many students miss the same question, it’s tempting to conclude they didn’t learn the concept. But sometimes the issue is question clarity rather than learner understanding. Research on assessment design consistently finds that question wording can significantly influence student responses, particularly in multiple-choice formats (Haladyna, Downing, & Rodriguez, 2002).

When deciding how to approach question outliers, consider:

• Could the wording have been ambiguous?
• Did multiple interpretations seem reasonable?
• Did students misread a key phrase?
• Could more than one answer be interpreted as correct?
• Would someone unfamiliar with the course interpret the question the same way?

One clue is the pattern of wrong answers. If responses are spread across many incorrect options, guessing may be at play. But if a large number of students selected the same incorrect option, the question may have unintentionally led them there.

A useful strategy is to ask a colleague to read the question without context and explain what they think it’s asking. If their interpretation differs from yours, students may have experienced the same confusion.

If the question wording was misleading, instructors might choose to discard the question, reducing the exam's total point value, or give everyone credit for the response. This really depends on the course structure and the number of points involved.

Was the question aligned with what was taught?

Assessment should measure what students have had a fair opportunity to learn. Sometimes questions feel reasonable because they align with our expertise in a discipline, but students may not yet have that level of fluency.

If wrong answers are spread across multiple choices, it is worth taking a moment to reflect on the alignment between the question and your course:

• Was the concept clearly addressed in class, readings, or activities?
• Did students have a chance to practice similar problems or concepts?
• Was the level of complexity comparable to examples or assignments?

A useful check is to look at your learning objectives for that unit. If the question requires skills or knowledge beyond those objectives, it may not be measuring what the course intended to assess.

How difficult was the question compared to the rest of the exam?

Every exam should contain a mix of question difficulties. Some should be straightforward checks of foundational knowledge, while others ask students to apply or analyze concepts.

However, problems arise when one (or more) question is dramatically harder than the rest.

Consider:

• Was this question significantly more complex than others?
• Did it require multiple steps or layers of reasoning?
• Was it testing a small detail rather than a key concept?

If a question stands out as disproportionately difficult, students’ performance on it may not reflect the assessment's learning goals.

In those cases, instructors sometimes choose to reduce the point value of the question (if it seems to be disproportionately affecting students' overall results) or to offer partial credit (if there are ways to identify which components have been successfully mastered). The key is to ensure that the exam still fairly reflects students' overall learning. 

Is there more than one reasonable answer?

Occasionally, a question reveals something unexpected: students make strong arguments for answers that were not originally considered correct.

This can happen when:

• Multiple interpretations of the question are valid
• Two choices to the answer could reasonably apply
• The concept itself is more nuanced than the question allowed

If you, or a student, can articulate a defensible rationale for an answer, that’s often a signal that the question may not have been constructed tightly enough.

Rather than viewing this as a mistake, it can be an opportunity to recognize thoughtful reasoning. In these cases, instructors often award credit for multiple answers or give full credit to everyone.

What decision best supports student learning?

Ultimately, the goal of assessment is not simply assigning grades, it’s supporting learning. When deciding how to handle a problematic question, consider:

• What message will this decision send to students?
• Will it reinforce fairness and transparency?
• Does it acknowledge legitimate issues without undermining the assessment?

Students often respond positively when instructors explain their reasoning. A brief message such as: “After reviewing the results for Question 12, I realized the wording may have allowed multiple interpretations. I’ve decided to award full credit for that question.” helps students see that the decision was thoughtful and intentional.

What can this exam teach me for next time?

Finally, exam analysis is not just about fixing the current assessment; it’s about improving future ones.

After making adjustments, consider:

• Were there concepts that students struggled with more than expected?
• Did certain question formats work better than others?
• Are there areas where additional practice or explanation might help?

Many instructors keep a simple note for themselves after grading exams, documenting questions that should be revised or replaced next time. Over time, this process leads to stronger assessments and fewer surprises.

A final thought

Every instructor occasionally encounters an exam that doesn’t unfold exactly as planned. When that happens, it’s easy to feel pressure to make a quick decision. But by taking a few minutes to analyze the results and ask yourself the right questions, you can turn that moment into a productive reflection on both assessment and teaching.

In the end, the goal isn’t perfection; it’s thoughtful, fair decisions that support learning while helping us refine our practice as educators. And sometimes, the most valuable outcome of an exam isn’t the score students receive—it’s what we learn from the results.

References:

Joint Committee on Testing Practices. (2004). Code of fair testing practices in education. https://ncme.org/wp-content/uploads/2025/10/CodeofFairTestingPractices-1.pdf

Haladyna, T. M., & Rodriguez, M. C. (2013). Developing and validating test items. Routledge.

Haladyna, T. M., Downing, S. M., & Rodriguez, M. C. (2002). A review of multiple-choice item-writing guidelines for classroom assessment. Applied Measurement in Education, 15(3), 309–334. https://doi.org/10.1207/S15324818AME1503_5

Joint Committee on Testing Practices. (2004). Code of fair testing practices in education. https://ncme.org/wp-content/uploads/2025/10/CodeofFairTestingPractices-1.pdf

Kane, M. T. (2013). Validating the interpretations and uses of test scores. Journal of Educational Measurement, 50(1), 1–73. https://doi.org/10.1111/jedm.12000

Nitko, A. J., & Brookhart, S. M. (2014). Educational assessment of students Anthony J. Nitko, Susan M. Brookhart. Pearson.

Schinske, J., & Tanner, K.. (2014). Teaching more by grading less (or differently). CBE—Life Sciences Education, 13(2), 159–166. https://doi.org/10.1187/cbe.cbe-14-03-0054