Is the final grade on the nationally registered paramedic exam based upon any variables?
That is the question I posed last month, and Bryan Prosek, EMT-B, and I sought to answer this question through the use of a regression analysis. I should say that I expected this analysis to demonstrate what I always suspected - that understanding mathematics would be the best predictor of success on the NREMT-Paramedic written exam.
In our study, we had clearly declared dependent and independent variables. The dependent variable was past students' test scores on the exam. The independent variables consisted of the following categories from these same students: entrance exam scores, self-reported social interaction style and self-reported stress levels.
Utilizing data from graduates of the paramedic program at Stony Brook University in New York, a regression analysis was able to be calculated. Our data included the results of test scores and self-reported levels from 64 individuals. Figure 1 displays the data for the first five and last five test takers to illustrate how the information in this study was collected.
When the data was originally gathered, social interaction was documented as either a "P" (Passive) or "A" (Aggressive). Since regression analysis is an equation, these values needed to be converted into numbers that could be recognized in the equation, and the numbers representing need to mean something. For example, 10 needs to be 10 times as great as 1; 100 has to be 20 times as great as 5, etc. Therefore, in order to correctly represent social interaction, dummy variables were established. If the original data displayed aggressive social interaction, it was changed one "1" in our data. For passive values, "0" was utilized. Since we only had two choices (P or A), only one dummy variable was needed. If three choices were presented, two dummy variables would have been created.
After this information was inputted into an Excel document, a regression analysis was performed. The regression was also performed in a Statistix program and provided the same results. Figure 2 displays the regression output. The R square is 0.41, which displays that this is a significant result and accounts for about 40% of the variables that determines the paramedic exam score. The equation to predict the paramedic exam score is as follows:
Exam Score = 62.77 + 0.05(HOBET Math) + 0.21(HOBET Verbal) 0.94(Social Interaction) 0.03(Family Stress) + 0.02(Social Stress) + 0.02(Money Stress) 0.04(Academic Stress) + 0.10(Work Stress).
Therefore, by inputting a student's responses to stress levels and social interaction as well as their entrance exam scores into this equation, a predicted paramedic exam score can be established.
Further review of the regression output displays that the Health Occupations Basic Entrance Test (HOBET) Verbal Score is the most representative of performing well on the paramedic exam. This is due to it having a very small P-value.
One can use the data to state that: every single point increase on student HOBET Verbal Score equates to a rise of .2106 points on the NREMT-Paramedic written exam.
Work Stress Level is also another variable that has a P-value of less than 5%. This is very significant to the understanding of what affects exam score. One can also understand that Social Interaction does not represent the exam score well. It is displayed through this that aggressive people actually tend to perform almost one point worse than passive individuals.
Our original belief that the HOBET Math Score would be the highest indicator of success on the paramedic exam score was proven wrong through this regression analysis. The Verbal Score was the best variable to predict the exam score. After consideration of this fact, one reason for this could include, since the nationally registered paramedic exam score is a written test (multiple-choice), it is important to have strong verbal test taking skills in order to pass the exam. It is still believed that math skills plays an important field in the actual field of emergency medicine, but that cannot be decided mathematically with this project data.
The HOBET exams break down each section of the exam parts with scores. In future studies of this regression analysis, it would be interesting to see the effects of the Verbal Section and Math Section of the test in further detail. Also in future studies, it would be interesting to study the outliers in greater detail. With any future study, it would be more informative with more student information, as well as analyzing further the effects of each variable.