The article’s title clearly defines the boundaries of the research question and identifies the critical issue related to childhood obesity: this makes it clear what the piece is about. Interestingly, the research question or null and alternative hypothesis was never articulated in the article. However, Hessler (2015) introduced the reader to the overall topic of discussion and showed his attitude so that both questions and hypotheses could be defined if necessary.
Thus, the research question was to determine the effectiveness of an online self-study program for nurse practitioners on counseling families about childhood obesity. The alternative hypothesis was that the proposed program had efficacy for nurses, whereas the null hypothesis postulated no significant effect. Nor was there a separate section in which Hessler establishes the variables and discusses their classification. Instead, the author again formulates the text so that the reader can identify the entire set of variables on their own. For example, the dependent variable is
- the mean score showing the nurse practitioner’s level of knowledge before and after the intervention, and
- also the level of satisfaction with her knowledge as examined by the questionnaire.
The independent variable, in this case, is online learning: by manipulating this, the authors observed changes in the dependent variables. The control variables that did not change throughout testing were all other variables not directly used for analysis: sample composition, procedure and duration of the training, and questionnaire design.
One of the functional roles of nursing is to counsel patients on better quality of life, including helping families with childhood obesity. It has been shown that nurses without a sufficient evidence base may feel uncomfortable discussing the sensitive topic of their child’s overweight with parents, especially if the parents themselves do not have similar problems. In addition, knowledge on childhood obesity is constantly being updated, so nurses may have felt insecure about such a flood of academic information. Consequently, the use of online learning is a promising strategy for addressing the intended problems for nurses, and thus the topic of this study proved to be relevant to the nursing community.
The findings reveal several important trends that simplify nursing practice in childhood obesity. First, the short online training made sense for nurses to increase their knowledge and confidence. Second, the nurses felt more comfortable after taking the course. Third, it was shown that despite the intervention, time constraints remained critical. Using these findings, clinicians can improve work practice and make discussions about childhood obesity more structured and effective.
The study design fully justified Hessler’s goal. Since the primary question was to test the effectiveness of the online intervention, the before/after algorithm testing was valuable and appropriate. This approach was not entirely ideal for the project, as this algorithm is not without systematic errors and biases, but it is a relatively reliable statistical tool. The model used to identify evidence from the current design was the evidence pyramid and EBP Levels of Evidence practice (Evidence-based practice, 2021). Thus, the level of evidence was Level III (Observational Studies with Comparison Groups) and, more narrowly, the design could be called a retrospective cohort study, but the same sample of people exposed to the intervention was used as groups of individuals.
Regarding the sample, it is worth saying that no specifics of its cancellation were discussed, and a voluntary participation mechanism was used as a sampling tool, which means that the sample was non-probability and non-randomized: participants decided for themselves whether to take part in the questionnaire or not. The sample size (no. = 354) was not justified by the desire to achieve a specific confidence interval, nor by screening out unsuitable questionnaires, nor by the final proportion of participants who agreed to take the survey out of the total number of applications.
Additionally, there was no analysis of sample power, although a KMO metric was measured for it, showing the adequacy of the sample for factor analysis. A good feature of the experiment was the diversity of the participating nurses: the data in the article describe differences in age, gender, and previous experience, thus reducing bias. These demographic traits likely determined the criteria for inclusion of respondents, but this was not reported. Nor was the possibility of using inferential statistics and obtaining cross-sectional data for each of the groups reported, although this could have been interesting.
In terms of the procedural part of the methodology, the tools used were a paired two-tailed paired t-test, determination of the mean, and Cronbach’s alpha measurement. For this psychometric study, this is a reasonable and appropriate basis for a statistical study. For example, the mean scores were not simply compared but using the paired t-test, which assessed the significance of scores over time. At the same time, Cronbach’s alpha assessed the consistency of the questionnaire items. Data collection was consistent over time: test and questionnaire results were collected before and after the intervention. Significantly, the independent variables were collected the same way, and no modifications were described for all 354 respondents.
Regarding the results, it is fair to acknowledge that the t-test allowed us to immediately decide which changes were significant at the critical level of 5%. This allowed us to understand that of all items accurately, only (p=0.363>0.05) was statistically not significant and supported the null hypothesis, which answered the author’s question of the experiment. This was hardly the result of type 2 error, as this result makes sense–the lack of time is an apparent reason for the lack of work on childhood obesity.
The researcher identified the significance of these results and showed their impact on nursing practice. All results were presented in one table, and a few additional items were described in a paragraph: for an average level, that is enough. The paper is not overloaded with statistics but uses a minimum of them, focusing on textual conclusions. Therefore, no figures, charts, or histograms were used, although they could have been.
There is a discussion of some limitations of sampling bias and the possible influence of the first test on the second test in the material. This is a common validity risk for retrospective studies that can be addressed by blind randomization, but this was not discussed in the paper (York, 2016). The results show a clear benefit of the online nursing course from an in-person perspective, but the applied value of such findings is questionable. If an in-person study were available, it would be appropriate to control for more variables, randomize participants, ensure they could not communicate, and observe cohort results related to demographic groups (Handley et al., 2018).
It would have been appropriate to add a control group that did not complete the training and tried testing again after time: the training probably did not make any sense. All of this combined would have produced an expanded design that would have produced more sensitive results of high value for health care systems and the personalization of therapy.
Evidence-based practice: Levels of evidence and study designs. (2021). Ascension. Web.
Handley, M. A., Lyles, C. R., McCulloch, C., & Cattamanchi, A. (2018). Selecting and improving quasi-experimental designs in effectiveness and implementation research. Annual Review of Public Health, 39, 5-25.
Hessler, K. L. (2015). Self-efficacy and knowledge of nurse practitioners to prevent pediatric obesity. The Journal for Nurse Practitioners, 11(4), 402-408.
York, R. O. (2016). Statistics for human service evaluation. Sage Publications.