NURS 766 Week 5 Test Administration Paper Example

Administering, Analyzing and Reviewing a Unit Exam in Adult Health Nursing: From Blueprint to Item Decisions

The Exam

The unit exam on heart failure is the second of four exams in a junior-year adult health course. It has 30 points: eighteen single-answer multiple-choice items, two multiple-response items scored with partial credit and one unfolding case with a matrix, a cloze and a bowtie item. Each item was written from one of the unit's objectives and reviewed by a colleague before use. Forty-two students took the exam on the course's computer-based testing platform.

Before the Exam

McDonald (2018) treats the test blueprint as the first step of administration as well as construction, because it is the evidence that the exam samples the objectives in proportion to their importance. Our blueprint lists each objective, its cognitive level, the number of points allotted and the items that measure it. The two objectives on recognizing cues and prioritizing care carry 12 of the 30 points because they represent the unit's priorities and the thinking the licensure exam requires.

Accommodations are arranged before the exam through the university's disability services office. Three students had approved accommodations for extended time, and one for a reduced-distraction room. The exam is identical in content for these students; only the conditions change, which preserves both fairness and the meaning of the score.

Security has changed with generative AI. Salamonson et al. (2026) argue that rules against AI use are difficult to enforce and that assessments must instead be designed so that their structure protects validity. For a unit exam, the structural protections are a supervised room, a locked-down browser, randomized item and option order and a policy that items are not released after the exam. Students review their results in a supervised session rather than taking the items home.

During the Exam

The exam runs in a quiet computer lab with one proctor for every twenty students. The time limit is 60 minutes, set by allowing about one and a half minutes per single-answer item and more for the case items, which require reading. Instructions are read aloud at the start, including how partial credit works on the multiple-response items, so no student loses points by misunderstanding the format. Proctors log any disruptions, such as a computer failure, so they can be considered when scores are reviewed.

Make-up exams raise a fairness question that faculty often answer by suspicion. Faculty commonly worry that students who test late gain an advantage by hearing about the questions. Bultas et al. (2024) examined 156 late exams in a large nursing program and found no statistically significant difference between late scores and either the student's own on-time average or the class average. That evidence supports a make-up policy based on documentation and timing rather than a penalty: students with an excused absence take the same exam within three days, with a new random order of items and options.

After the Exam: Item Analysis

Makhija et al. (2024) note that many educators build tests without any means of evaluating them after administration, and they describe how to use test data to find flaws and improve exams, beginning with item difficulty, the index of discrimination and distractor analysis and adding reliability when the testing platform reports it. Our platform reports all four.

Item difficulty is the proportion of students answering correctly. For a classroom exam, the faculty treat values between about 0.30 and 0.90 as informative; an item answered correctly by nearly everyone measures little, though it may still be worth keeping if it tests essential safety content. Discrimination, reported as the point-biserial correlation, shows whether students who did well overall answered the item correctly more often than those who did poorly. Values of 0.20 or above are acceptable, and a negative value means something is wrong with the item. Distractor analysis shows how many students chose each wrong option; a distractor chosen by no one is doing no work, and one chosen more often by high scorers than by the key suggests a second defensible answer. Hartigan-Rogers et al. (2019) found that three-option items performed as well as four-option items on difficulty and discrimination, which makes it easier to drop a nonfunctioning distractor without writing a new one.

Reliability for the whole exam is estimated with the Kuder-Richardson formula 20. The exam's value was 0.71, acceptable for a 30-point unit exam with mixed item types. Short classroom exams rarely reach the values expected of standardized tests, and a single unit exam should be read as one piece of evidence among several.

Five Items and the Decisions Made

Item 13 is the one that required immediate action. Its negative discrimination showed that the strongest students preferred another option, and on review the faculty agreed that option C was also defensible because the stem did not specify whether the patient took the diuretic once or twice daily. Both B and C were credited, every student's score was recalculated and the stem was rewritten for the item bank. Item 9 is easy, but it tests a classification that students must know, so it stays with a better third option. Item 23 performed adequately but showed a pattern: students chose the right condition and actions but missed the monitoring parameters, which says more about teaching than about the item.

Partial credit scoring helps here. Norton and Yockey (2026) found that partial credit scoring improved how exams measured students' knowledge without evidence of grade inflation, and on this exam it let the analysis show which parts of the case students missed rather than reducing each case item to right or wrong.

Reading the Whole Exam, Not Only Items

Item statistics answer questions about single items; three other numbers answer questions about the exam. The mean score was 22.4 of 30 (75%), slightly below the course's passing standard of 77%, and the distribution was skewed toward the low end by six students who scored below 60%. Those six students did not cluster on any single objective, which suggests general preparation problems rather than a teaching gap, and each was invited to meet with faculty before the next exam. The scores by objective told a different story: the class averaged 88% on the medication and classification objectives but 63% on the prioritization objective, which matches the bowtie finding and shows where the next week's teaching should go.

The decision to credit two answers on item 13 raised the mean by less than half a point, but it changed the course standing of two students who had been just below the passing threshold. That is a reason to make such decisions on the basis of the item, before looking at who benefits, and to document them. The faculty recorded the rationale for the change in the course file and announced it to the whole class at once, so that no student learned of it through rumor.

Planning the Next Exam

Analysis is useful only if it reaches the next exam. Three changes follow from this one. The bank now holds the rewritten item 13 and the revised item 9, both flagged for re-analysis after their next use. The third exam will include a second case with monitoring parameters, so that the gap found on the bowtie item can be measured again after reteaching. And the time limit will stay at 60 minutes, since the platform's timing report showed that 95% of students finished with at least eight minutes to spare, which suggests the case items did not rush them.

Reviewing the Exam With Students

The exam review takes place in a supervised session one week after the exam. Students see their own responses and the rationales for items they missed, but items are not copied or released. Faculty begin with the class-level findings, including the credited item and the monitoring pattern from the bowtie, so that students see that analysis can change scores and that faculty learn from the exam too. Students then complete a brief error analysis, classifying each missed item as a knowledge gap, a misreading or a test-taking error, which gives them a plan for the next exam.

Conclusion

An exam is not finished when students hand it in; it is finished when its statistics have been read and its weak items have been fixed. The heart failure exam was planned from a blueprint, given under conditions that protected fairness and security, analyzed item by item and reviewed with students. One flawed item was credited and rewritten, one weak distractor was replaced and one teaching gap was found. The next exam will start from an item bank that is better for the work, and the students will start from a clearer picture of what they still need to learn before the final exam.

References

Bultas, M. W., Schmuke, A., Armstrong, K., Rubbelke, C., Alnawman, M., Moran, V., Fuller, K., Cole, B., & Lovan-Gold, K. (2024). "Make-up" examinations: Comparison of on-time and late examination student scores. Nurse Educator, 49(2), 80-84. https://doi.org/10.1097/NNE.0000000000001508

Hartigan-Rogers, J. A., Redmond, S., Cobbett, S., Cifuentes, K., Honan, D., Richard, E., Sheane, V., & VanTassell, T. (2019). A, B, or C? A quasi-experimental multi-site study investigating three option multiple choice questions. International Journal of Nursing Education Scholarship, 16(1). https://doi.org/10.1515/ijnes-2019-0061

Makhija, H., Schneid, S. D., Kalinowski, A., Mandel, J., & Davidson, J. E. (2024). Evaluating the quality of multiple-choice test questions in the postlicensure environment. Journal of Continuing Education in Nursing, 55(10), 487-492. https://doi.org/10.3928/00220124-20240718-03

McDonald, M. E. (2018). The nurse educator's guide to assessing learning outcomes (4th ed.). Jones & Bartlett Learning.

Norton, L. M., & Yockey, J. (2026). Partial credit scoring: A helping hand or a crutch? Journal of Nursing Education, 65(8), 502-504. https://doi.org/10.3928/01484834-20260224-05

Salamonson, Y., Irwin, P., Kornhaber, R., Kirk, D., McGough, S., Vakavosaki, K., Ramjan, L., Bloomfield, J., Matheson, A., & Greenwood, M. (2026). Refining assessment design and practice in nursing and midwifery education in the era of generative AI: A discussion paper. Journal of Clinical Nursing. Advance online publication. https://doi.org/10.1111/jocn.70484