Behavioral anchors on Bondy's scale, critical behaviors scored separately, and a calibration study with weighted kappa rising from .48 to .67 define the clinical evaluation tool built for NU524 Unit 8. Searches like "nu 524 unit 8 assignment example", "nu524 unit 8 sample" and "nu524 unit 8 example" land here.
Agreement From 48 to 79 Percent: Rebuilding a Clinical Evaluation Tool With Bondy's Criterion-Referenced Levels and Behavioral Anchors
[Student Name]
Purdue University Global
NU524: Curriculum Development Process, Assessment, and Evaluation Strategies
Unit 8 Assignment
[Instructor Name]
[Date]
The program, faculty and calibration results are composite; the scale and statistics cited are published.
The Problem: The First Calibration Round
The old tool rated six areas of clinical performance as satisfactory, needs improvement or unsatisfactory, each with a one-line descriptor such as "communicates effectively with the health care team." To test it, two experienced clinical faculty independently rated eight recorded simulation performances by third-semester students on all six criteria, 48 paired ratings in all. They agreed exactly on 23, or 48 percent. Weighted kappa, which credits near-agreement on an ordered scale and corrects for chance (Cohen, 1968), was .48. Communication was the loosest criterion, with exact agreement on three of eight students.
When the two faculty discussed their disagreements, the pattern was clear. They were watching the same behavior and judging it against different standards, and neither was wrong by the old tool's wording, which never said how much help was acceptable at each level. The problem was the tool, not the raters. One rated a student satisfactory in communication because she eventually gave a complete handoff; the other rated her needs improvement because she did so only after two prompts.
The Redesign
Three changes address that pattern.
Criterion-referenced levels instead of letter-like categories. The tool now uses Bondy's five levels, independent, supervised, assisted, marginal and dependent, which were designed to describe clinical performance by the quality of the behavior and the amount of guidance a student needs (Bondy, 1983). Levels describe what the student does, not how she compares with classmates.
Anchors that name the amount of cueing. Every cell describes an observable behavior and states how much prompting the student needed, because the amount of cueing was exactly what the two faculty had judged differently.
Critical behaviors pulled out. Four behaviors are rated separately as met or not met, independent of the scale: patient identification before any intervention, recognition of a change in condition requiring escalation, medication safety checks, and no false documentation. A student who fails a critical behavior is not rescued by high ratings elsewhere.
The Tool
Six criteria, each linked to a course outcome and an Essentials domain (American Association of Colleges of Nursing, 2021): assessment; prioritization; medication safety; communication; documentation; and professional conduct.
Example anchors, communication criterion:
1. Independent: gives a complete, organized handoff and SBAR report without prompts; adjusts language for the patient and family.
2. Supervised: complete and organized with one prompt, such as a reminder to include the most recent vital signs.
3. Assisted: complete only with two or more prompts; organization needs direction.
4. Marginal: incomplete or disorganized even with repeated prompts; relies on the instructor to supply key information.
5. Dependent: unable to give a usable report; the instructor must give it.
Example anchors, prioritization criterion:
1. Independent: identifies the most urgent patient need and acts first on it without prompting, explaining the reason.
2. Supervised: identifies the priority correctly and needs one question to act on it.
3. Assisted: identifies the priority only after guided questions.
4. Marginal: attends to lower priorities first despite cues.
5. Dependent: cannot identify the priority with guidance.
The midterm and final expectations are stated: by the final evaluation, students are expected to perform at supervised or better on every criterion.
Why Not a Percentage Score
Some faculty proposed converting ratings to points and averaging them into a clinical percentage, as some programs do. The committee rejected it for two reasons. First, averaging lets strong performance in documentation offset weak performance in medication safety, which is exactly the trade clinical evaluation should not allow. Second, a percentage invites comparison between students, when the question clinical evaluation must answer is whether each student meets a standard. Criterion-referenced levels keep the question on the standard. The course still reports pass or fail for clinical, determined by two rules: every criterion at supervised or better at the final evaluation, and every critical behavior met.
Evidence Behind Each Rating
Each rating must be supported by at least one written example of observed behavior, dated, from the student's clinical days or simulation. The requirement slows faculty down slightly, and that is intended. When the two faculty in the first round were asked for examples, they found that several of their ratings rested on a general impression rather than a specific observation, and those were the ratings on which they disagreed most. Written examples also give the student something concrete to work on, and they protect the program if a failing grade is appealed.
The Second Calibration Round
The same two faculty, after a one-hour orientation to the anchors, rated eight new recorded simulations with the rebuilt tool. They agreed exactly on 38 of 48 ratings, 79 percent, with weighted kappa of .74. Communication agreement rose to seven of eight. The remaining disagreements were mostly one level apart, between supervised and assisted, and most concerned whether a question from the instructor counted as a prompt. On the four critical behaviors, the two faculty agreed on every student in both rounds, which suggests that clear pass or fail items are easier to rate consistently than graded levels, and supports keeping them separate.
What Remains
Agreement rose, but two faculty rating sixteen recorded performances is a small test. The program will extend calibration to all eight clinical faculty each semester, using a shared set of recordings, and will add a note to the anchors defining a prompt as any question or statement from the instructor that directs the student's attention to missing information. Evaluation texts in nursing education caution that rating tools in clinical settings are vulnerable to rater bias and that training and calibration are needed to keep ratings consistent (Oermann & Gaberson, 2019).
Using the Tool Fairly
Students receive the anchors on the first day of the course, so students know what independent performance looks like. Ratings are discussed with the student at midterm with specific examples, and a student rated marginal on any criterion at midterm receives a written learning plan. The critical behaviors are explained on the first clinical day, with the reasons each one matters. In the pilot semester, students' written comments on the midterm evaluation shifted from questions about why they received a rating to questions about what they needed to do to reach the next level, which is the conversation the anchors were meant to start.
References
American Association of Colleges of Nursing. (2021). The essentials: Core competencies for professional nursing education.
Bondy, K. N. (1983). Criterion-referenced definitions for rating scales in clinical evaluation. Journal of Nursing Education, 22(9), 376-382. https://doi.org/10.3928/0148-4834-19831101-04
Cohen, J. (1968). Weighted kappa: Nominal scale agreement with provision for scaled disagreement or partial credit. Psychological Bulletin, 70(4), 213-220. https://doi.org/10.1037/h0026256
Oermann, M. H., & Gaberson, K. B. (2019). Evaluation and testing in nursing education (6th ed.). Springer Publishing.
How this NU524 Unit 8 example is structured
Problem, redesign, evidence: the rationale opens with the first calibration round, where two faculty rating the same recordings on the old tool's one-line descriptors agreed exactly on 48 percent of 48 paired ratings, with linear-weighted kappa of .48 and communication the loosest criterion at three exact matches in eight. Each design choice is then explained: criterion-referenced levels instead of letter grades, anchors naming the amount of cueing, and critical behaviors pulled out of the scale so they cannot be averaged away. The second round, with rewritten anchors and a thirty-minute calibration meeting, reached 65 percent exact agreement, every pair within one level, and kappa of .67. Figures were computed by script from composite ratings. A plan for recalibrating faculty each term closes the report, since agreement drifts.
Get an NU524 Unit 8 example written to your instructions
Describe the clinical setting, the course outcomes the tool must reflect, and any tool your program already uses, attaching the Unit 8 grading criteria as well. No fee applies to a first sample, delivered in 24-48h. Ratings of real students stay out of it entirely, and the calibration figures in the sample are composite. 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 8 questions, answered
What is Bondy's scale?
A five-level criterion-referenced rating scale Bondy published in 1983 for clinical performance: independent, supervised, assisted, marginal and dependent. Each level describes the quality of performance and the amount of cueing a student needs. The tool uses it because cueing is observable, which helps two evaluators reach the same rating.
How is agreement between evaluators measured?
Two evaluators rate the same performances independently, often recordings, and the ratings are compared. The report gives exact agreement, agreement within one level, and linear-weighted Cohen's kappa, which corrects for chance and gives partial credit for near misses. Percent agreement alone overstates consistency, which is why kappa is added.
Why keep critical behaviors separate from the rating scale?
Because an average can hide them. A student rated well on five criteria but who skipped patient identification might still pass if every item were averaged. The tool lists critical behaviors in their own box, and any one triggers review regardless of the other ratings. That keeps safety from being traded against performance elsewhere.