REPLY
Thank you for the opportunity to respond to Dr Downey's letter to the editor, which was undated, but received by CIN: Computers, Informatics, Nursing on May 20, 2003, regarding our article "Predicting NCLEX success with the HESI Exit Exam: Fourth annual validity study" published in the November/December 2002 issue. We have provided responses to what we believe are the key issues raised by Dr Downey.
Purpose of the study.
When initiating this study, we posed the inverse of Dr Downey's stated research question, ie, How many students in each of the identified scoring intervals failed the NCLEX on their first attempt? This research question has been the focus of 4 validity studies,1-4 and it is a useful question in determining if a relationship exists between Health Education Systems, Inc (HESI) Exit Exam (E2) scores and NCLEX success.
Composition of the sample.
Those students who did not take the NCLEX should always be accounted for in the data collection procedures, but they should not be included in the data analysis. If a student did not take the NCLEX, the respondent noted this fact on the questionnaire, and the student was eliminated from the aggregate that comprised the final sample for the study. In the interest of saving space, however, we did not discuss this data collection methodology because we saw it as a perfunctory task. In fact, the 77 students who did not take the NCLEX comprised only about 1% of the total RN population.
Failure to collect data relative to NCLEX success.
In each of the 4 studies, data were collected from schools of nursing regarding the NCLEX success of their students. In the current study, 98.3% of the RN students predicted to pass the NCLEX (identified as scoring in Category A/B) were successful on their first attempt, while 1.7% of the predicted to pass group were unsuccessful on their first attempt. Additionally, licensing examination success was calculated for each of the scoring intervals, which were identified in the bar charts as follows: of those whose HESI scores were in Category C, 94.08% were successful on the NCLEX; in Category D, 89.18% were successful; in Category E/F, 76.28% were successful; and in Category G/H, 49.81% were successful. Similar licensure successes and failures were found for the PN programs. (See Fig 1 and 2 on pages 265 and 266 of the article.)
Verification of the HESI Predictability Model (HPM).
Dr Downey's statement that the research consumer is unable to verify the accuracy of the HPM calculation is absolutely correct. Neither can consumers verify the calculations that are used for the SAT, the ACT, or the NCLEX. As a matter of fact, consumers cannot verify the formula for Coca-Cola(TM), but we can judge its worth as a beverage!! Just as Coke(TM) provides refreshment, published findings from the 4 studies conducted over a period of 4 academic years (1996-2000) with 19,387 subjects indicate that the E2 predicts NCLEX success.
Nursing domain tested by the E2.
We found Dr Downey's query as to the domain of nursing tested by the E2 puzzling. All 4 published studies stated that the E2 follows the test blueprint published by the National Council of State Boards of Nursing (NCSBN), and the publications describing these blueprints are included in the reference lists of all 4 articles. Dr Downey's comment that none of the 4 validity studies addressed HESI Specialty Exams is correct. It was beyond the scope of these studies to address any of the HESI exams other than the E2.
Use of a representative study sample.
Dr Downey questioned the representation of the sample to the population. More than 65% of the total population that took the E2 during the study year were represented in the study population (59.74% of the RN population and 71.78% of the PN population), and approximately 91% (95.18% of RN programs and 86.11% of PN programs) of those surveyed, responded. We therefore concluded that the data were representative of the total population. Because of incompatible file formats, 35% of the data from the total population could not be exported into the questionnaire format and 65% of the data were exported into the questionnaire format, resulting in a random assignment of the sample.
Use of uneven scoring intervals.
In the interest of presenting meaningful findings as succinctly as possible, we did not describe our rationale for the division of scoring intervals, but we are pleased to have the opportunity to present that information. The decision to separate category C/D was based on the fact that many schools reported using 85 as a benchmark for progression,4-6 and we wanted to scrutinize C and D individually to determine the degree of risk for NCLEX failure associated with such benchmarks. Also, splitting the C/D category had no impact on the study findings or the calculation of predictive validity. Therefore, the rationale was not belabored.
Statistical analysis.
Statistical analysis included assessment of predictive accuracy, summaries of descriptive data using frequency distributions, and calculations of chi squares (which are appropriate when a nominal level variable is under study).7 Predictive accuracy calculations were explained in both the third3 and fourth4 validity studies.
Review versus remediation.
Our study does not address the amount of review or remediation that students should undertake prior to NCLEX. The purpose of the study was to identify what schools of nursing were doing to facilitate students' remediation process. This area had not been comprehensively addressed in the prior validity studies. It would be ideal if schools would take the opportunity early in the academic process to begin remediation programs for students who need additional assistance. We suggest that more precise information needs to be gathered regarding the remediation process, and we agree with Dr Downey that further study into the effectiveness of these interventions should be undertaken. In fact, this suggestion for future study can be found in the "Recommendation" section of our article.
Characteristics of the sample.
The focus of this study was to determine the predictive accuracy of the E2. We identified that a key limitation of our study was the inability to control student activities that occurred between administration of the E2 and administration of the NCLEX. The study of such activities constitutes a topic for another study, but it was beyond the scope of our study.
Clinical trial (experimental versus descriptive design).
Dr Downey makes reference to requirements for clinical drug trials and their relevance to nursing education research. Our study did not use an experimental design. Research must be conducted within the constraints of the setting. Our study demonstrates an association between HESI scores and NCLEX success. Examination at each scoring interval demonstrated that as HESI scores decreased, NLCEX success also decreased. This information was based on a large sample from diverse academic settings. Additionally, these findings are congruent with 3 prior studies that addressed this same issue. This information was presented in a peer-reviewed journal so that nurse educators could make their own, independent decisions about the generalizability of our research findings within their settings.
In conclusion, we thank the Editor-in-Chief of CIN, Dr Leslie Nicoll for allowing us to respond to Dr Downey's letter. We appreciate the opportunity to engage in a scholarly discourse regarding our research findings. High-stakes computer-based testing remains a timely topic in nursing education. The growing use of computer technology to both prepare nursing students for clinical practice, as well as to evaluate their competency prior to licensure, ensures that computerized testing will remain a significant subject for future nursing research efforts.
Ainslie T. Nibert, PhD, RN
Anne Young, EdD, RN
Carolyn Adamson, PhD, RN
College of Nursing Texas Woman's University Houston, Tex
REFERENCES