Improving Prelicensure Healthcare Students' Confidence in Managing Aggression Through Interprofessional Simulation

Nursing, physician assistant, and respiratory therapy students rated their confidence in managing patient aggression before and after a shared high-fidelity simulation. Confidence rose on all ten items measured.

The study

This was a quasi-experimental pre/post design with a single group. Prelicensure students from three programs took part together in a high-fidelity simulation built around a patient becoming aggressive, followed by a structured interprofessional debrief.

The same ten items were asked immediately before and immediately after. Each item is rated from 1 to 11, and each asks about a different facet of managing aggression: comfort, perceived training, physical and psychological intervention, self-assurance, safety, technique, and the ability to meet the patient's needs. Because the same items are asked twice, every student who can be matched across the two surveys contributes a pair, and the analysis is a set of paired t-tests on those pairs.

Participants are identified only by a code they created themselves from their year of birth and their mother's first name. The pre survey collected 67 responses and the post survey collected 72. Of those, 57 matched on the participant code and form the analysis set. The mismatch is not simple dropout, since the post survey drew more responses than the pre survey. Some students appear to have entered a code that did not reproduce exactly across the two sittings.

What was found

Every one of the ten items improved, and every improvement was statistically significant. The mean difference is post minus pre, so a positive number means confidence went up.

The largest gains were in perceived training for handling physical aggression (1.74 points, d = 0.99) and in the perceived effectiveness of known techniques (1.67 points, d = 0.94). The smallest were in the ability to protect yourself physically (0.81 points, d = 0.40) and general comfort working with an aggressive patient (0.70 points, d = 0.46). Every confidence interval excludes zero.

How to read these numbers

Ten tests on ten items

Running ten tests raises the chance that at least one result is significant by accident. The p-values in the table are unadjusted. They hold up anyway: a Bonferroni correction for ten tests sets the threshold at .005, and all ten items fall below it, the weakest being .0045. No conclusion here depends on which correction is applied.

n is 54, not 57

The merged dataset holds 57 pairs, but each test uses only the pairs with both values present. That gives n = 54 on nine items and n = 53 on self-assurance.

What the design cannot tell you

There is no control group and no follow-up. The second measurement is taken immediately after the simulation, so it captures confidence at that moment. It cannot show whether the gain lasts, and it cannot separate the simulation from the simple effect of answering the same questions twice. Self-reported confidence is also not competence.

Where the numbers come from

Every value above is read from executed SAS output, not from a draft. The source is 202506_Survey_Paired_TTEST_Results.pdf, produced by SurveyPairedTTest202506.sas on June 27, 2025. The paired tests were then independently reproduced in Python with pandas and SciPy against the same survey exports, and agree on the matched n, the mean difference, and t for all ten items.

The materials

Data protection