Forty-eight students sat a composite fifty-item cardiopulmonary exam, and this NU524 Unit 7 item analysis report turns difficulty, corrected point-biserial and upper-lower discrimination into decisions: one rekey, two rewrites, one replaced distractor and one very easy safety item kept on purpose. Searches like "nu 524 unit 7 assignment example", "nu524 unit 7 sample" and "nu524 unit 7 example" land here.
From Statistics to Decisions: An Item Analysis Report on a Fifty-Item Cardiopulmonary Exam Taken by 48 Students
[Student Name]
Purdue University Global
NU524: Curriculum Development Process, Assessment, and Evaluation Strategies
Unit 7 Assignment
[Instructor Name]
[Date]
The exam, students and all statistics are composite; the methods cited are published.
Method
Item statistics were computed by script from the scored response file. Three statistics were used for each item.
1. Difficulty: the proportion of students answering correctly, p. Higher values mean easier items.
2. Corrected point-biserial: the correlation between each item and the total score with that item removed. Removing the item matters on a fifty-item exam, because an item's own contribution inflates its correlation with a total that includes it.
3. Upper-lower discrimination: the difference in proportion correct between the top and bottom 27 percent of scorers, thirteen students in each group.
Thresholds were chosen in advance: difficulty above .90 or below .30, and a corrected point-biserial below .20, flag an item for review. Flagged items are not automatically removed; each is read against its content and purpose (Oermann & Gaberson, 2019).
Reliability was estimated with KR-20, the coefficient for dichotomously scored items (Kuder & Richardson, 1937).
The Exam as a Whole
Mean score 36.2 of 50, or 72.4 percent; standard deviation 7.64; KR-20 .87; standard error of measurement 2.75 points.
A reliability of .87 is acceptable for a classroom exam used, with other evidence, for course grades. Reliability coefficients should be interpreted with care: a high value can reflect a long test or a varied group as well as a coherent one (Tavakol & Dennick, 2011). The standard error of measurement matters for students near the passing score: a student scoring 37 has a true score likely within about three points either side, which the course's policy on borderline scores should reflect.
Flagged Items
Eleven items had difficulty above .90; one had difficulty below .30; eleven had corrected point-biserials below .20. Several items appear on more than one list. Each flagged item received one of four decisions: keep, revise, rekey or remove.
The Decisions
Item 23, rekey. Difficulty .28, the only item below .30, with a negative point-biserial of minus .14. Eleven of thirteen top-scoring students chose option C, the keyed answer was B. Review showed that the key had been entered incorrectly; C was correct. When the strongest students overwhelmingly choose an unkeyed answer and the item correlates negatively with the total, the first thing to check is the key, not the students. The item was rekeyed, and for this administration credit was given for both B and C, so students who chose the old key were not penalized for the faculty's error; [nineteen] students gained a point.
Items 8 and 31, rewrite. Both had difficulty in the acceptable range but point-biserials of .06 and .11, meaning strong and weak students were almost equally likely to answer correctly. Item 8's stem could be answered from a clue in the options; item 31 had two defensible answers depending on an assumption the stem did not state. Both will be rewritten before the next administration, and both were kept in this exam's scoring because the flaw affected all students equally.
Item 17, replace a distractor. Difficulty .72 and point-biserial .31, a sound item, but option D was chosen by no one. A distractor that no one chooses adds nothing to measurement. It will be replaced with a plausible alternative based on a common error seen in clinical practice.
Item 4, keep deliberately. Difficulty .98 and point-biserial .09, weak discrimination because almost everyone answered correctly. The item asks what to do first when a patient on a heparin infusion shows signs of major bleeding. It stays on the exam because it measures a safety expectation that every student must meet, and a very easy safety item is doing its job when nearly everyone passes it. Its statistics say it does not separate students; its purpose was never to separate them.
The remaining easy items were reviewed and most were kept, because they sample essential content the blueprint assigns to recall; [three] will be revised to raise their cognitive level.
Distractors Across the Exam
Beyond the flagged items, every distractor was checked for how often it was chosen and by whom. A working distractor is chosen by at least a few students, and more often by weaker students than by stronger ones. Across the exam's 150 distractors, [19] were chosen by fewer than 5 percent of students, and [four] were chosen more often by the top group than the bottom group, which usually signals an ambiguity. Those four were read closely: two were defensible alternatives that the stems did not rule out and will be addressed in the rewrites; two reflected a misconception common among stronger students about oxygen targets in COPD, which the course will address in teaching rather than in the item. Nonfunctioning distractors are the easiest improvement available, because replacing them with plausible errors makes an item harder to guess without changing what it measures.
What Faculty Will Do Next Time
Three process changes follow from the analysis. The key will be checked by a second faculty member before scoring, since the item 23 error would have been caught by a simple review. Items will be banked with their statistics, so that future exams can draw on items with known performance. And the course team will meet within a week of each exam to review flagged items together, so that decisions are made by the people who wrote and taught the content, not by the analysis alone.
What the Analysis Adds to the Blueprint
The analysis also checks the blueprint in reverse. The analyze-level items had a mean difficulty of .64 and a mean corrected point-biserial of .34, higher than the recall items, which suggests that the items asking for judgment were doing most of the work of distinguishing students. The heart failure row, with nine items, had the highest mean discrimination of any content area.
Communicating the Results
Students were told that one item had been rekeyed and that scores had been recalculated, which raised the mean by about four-tenths of a point, and that no student lost points. The course team will review the rewritten items before the next exam.
Limits
Forty-eight students is a modest sample, and item statistics can shift noticeably from one cohort to the next. Upper-lower discrimination on thirteen students per group is coarse. And item analysis measures how items performed with this group; it does not show whether an item measures what it claims to, which requires review of content as well as numbers.
References
Kuder, G. F., & Richardson, M. W. (1937). The theory of the estimation of test reliability. Psychometrika, 2(3), 151-160. https://doi.org/10.1007/BF02288391
Oermann, M. H., & Gaberson, K. B. (2019). Evaluation and testing in nursing education (6th ed.). Springer Publishing.
Tavakol, M., & Dennick, R. (2011). Making sense of Cronbach's alpha. International Journal of Medical Education, 2, 53-55. https://doi.org/10.5116/ijme.4dfb.8dfd
How this NU524 Unit 7 example is structured
Method first: how the analysis was run, which statistics were computed, and the thresholds chosen, with the reasons for each. The corrected point-biserial removes each item from the total before correlating, which matters on a fifty-item exam, and the upper and lower groups are the top and bottom 27 percent, thirteen students each. Results follow at two levels, the exam as a whole and then flagged items. Each flagged item receives a short paragraph ending in one of four decisions: keep, revise, replace a distractor, or rekey. Item 23 is rekeyed and the exam rescored, which moves that item's difficulty index to .56 and KR-20 to .88, and the report notes that rescoring changes grades and was approved by the course lead. A closing section warns that forty-eight students make every statistic unstable, so decisions stay provisional until the next administration.
Get an NU524 Unit 7 example written to your instructions
Item analysis needs data, so send the statistics your instructor provided or describe the exam so a composite data set can be generated. Include the Unit 7 assignment and rubric. First samples cost nothing, arrive within 24-48h, and show the calculations, so every figure can be checked against the formulas your course uses. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.
NU524 Unit 7 questions, answered
What does a negative point-biserial mean?
That students who scored lower on the whole exam were more likely to answer the item correctly than those who scored higher. It usually signals a miskey, an ambiguous stem or a distractor that attracts strong students for a good reason. Item 23 here had a corrected point-biserial of -.24, and the distractor table showed the recorded key was wrong.
How is KR-20 different from the item statistics?
KR-20 estimates the internal consistency of the exam as a whole for items scored right or wrong, while difficulty and point-biserial describe individual items. The report gives .87, a comfortable figure for a classroom exam. It is a reliability estimate, not evidence that the exam measures the right things, a distinction many reports blur.
Should very easy items always be removed?
No. An item answered correctly by nearly everyone discriminates poorly, but it may test something every student must know. The report keeps a patient identification item answered by 98 percent for exactly that reason and says so. Easy items that test trivia, like the potassium range, are rewritten instead.