SAMPLE QUESTIONS FOR HSM 470 FINAL EXAM

For each of the variables below, indicate their level of measurement using the following (1 point each):

N = Nominal

O = Ordinal

I = Interval or ratio

1. __I___ ___ Age measured in years
2. __I__ ____ Systolic blood pressure measured in mm Hg
3. __O______ Age measured as < 10, 11-20, 21-30, 31-40, > 40
4. __N______ Gender measured as 0 = female, 1 = male
5. __N______ Application of MAST measured as 0 = not applied, 1 = applied
6. __I___ ___ Daily urine output measured in ml/day

For each of the following, indicate which statistical test is appropriate (2 points each):

1. Pearson Correlation You wish to measure the association between blood pressure (measured in mm Hg) and age measured in actual years (assume all data are normally distributed).
2. Independent T test You wish to compare the average daily urine output between experimental drug A and experimental drug B (assume all data are normally distributed).
3. ANOVA You are interested in the difference between the on-scene times (measured in minutes) between A, B, C, and D shifts (assume all data are normally distributed).
4. Paired T test You wish to measure the difference between pre-test and post-test scores of 25 paramedics who just completed an ACLS course (assume all data are normally distributed).
5. Mann-Whitney You wish to compare the Glasgow Coma Scores of head-injured patients who received hyperventilation and those who did not (assume data are NOT normally distributed).
6. Wilcoxon You wish to compare the Glasgow Coma Scores of head-injured patients before and after receiving hyperventilation (assume data are NOT normally distributed).
7. Chi square w/ Yates You wish to compare the differences in the frequencies of students who passed their National Registry Practical exam (0 = fail, 1 = pass), based upon the gender of the student (0 = female, 1 = male). (assume F_e > 5 for all cells)
8. Independent T test You wish to compare the differences in the ages of patients enrolled in your cardiac arrest survival study. The groups you wish to compare are the experimental group, who received Amiodorone (coded 1), and the control group who received Lidocaine (coded 0). Age was measured in years. Assume all data are normally distributed.
9. ANOVA You want to determine the statistical difference in the systolic blood pressures of patients who are taking different antihypertensives. Group 1 is taking a beta blocker, group 2 is taking a calcium channel blocker, and group 3 is taking a diuretic. Systolic blood pressure is measured in mm Hg. Assume all data are normally distributed.
10. Pearson Correlation You wish to measure the association between the test scores (measured between 0 and 100) on the ACLS final exam and the level of experience of the paramedics taking the exam, measured in years (assume all data are normally distributed).

Please comment on each of the following vignettes and interpret their statistical output (5 points each):

1. Using the statistical output below, please comment on the association between years of field experience and test scores on the National Registry written exam. The variable YEARS is measured in actual years of field experience and SCORE, is the written test score measured between 0 and 100%. Assume the data were normally distributed. Assume that alpha is 0.05.

- - Correlation Coefficients - -

YEARS SCORE

YEARS 1.0000 -.8586

( 25) ( 25)

P= . P= .000

SCORE -.8586 1.0000

( 25) ( 25)

P= .000 P= .

• ANSWER: Because the data are measured on an interval scale, are assumed to normally distributed, and we wish to measure the strength of association, the Pearson Correlation procedure is the appropriate statistical test. In the output we see that the test score is negatively correlated with years of field experience. Given the correlation coefficient of –0.8586, this means that for every additional year of field experience, the test score can be expected to decline by 0.8586 points. In looking at the P value for this coefficient, we see that P = 0.000 which is less than alpha. From this we conclude that we cannot accept the null hypothesis that the correlation coefficient is zero (meaning no correlation between field experience and test scores). Therefore we conclude that the correlation coefficient of –0.8586 is significantly different from zero and that there is a true negative correlation between these variables.

2. Using the statistical output below, please comment on the difference in the pass rates on the National Registry Practical Exam between certificate paramedics and AAS degreed paramedics. The data are not normally distributed. The variable PASS is coded as 0 for failure, and 1 signifies a passing score on the practical exam. The variable EDUCATION is coded as 0 for certificate paramedics and 1 for AAS degree paramedics. Assume that alpha is 0.05.

EDUCATIO by PASS

Chi-Square Value DF Significance

Pearson .37085 1 .54254

Continuity Correction .04340 1 .83497

Likelihood Ratio .37188 1 .54198

Linear-by-Linear .35602 1 .55073

Association

Fisher's Exact Test:

One-Tail .41791

Two-Tail .69510

Minimum Expected Frequency - 5.760

Number of Missing Observations: 1

• ANSWER: Because the data are measured on an ordinal level, the chi square test is appropriate. Even though the data are not normally distributed, the chi square test can still be used. This is a 2x2 contingency table so we must use either the Yates continuity correction or the Fisher Exact Test. In looking at the minimum expected frequency of 5.76, this is greater than the minimum expected frequency of 5 required for the chi square. Therefore, we do not need the Fisher’s exact test and can use the Yates continuity correction. This yields a P value of .83497 which is greater than alpha of 0.05. We cannot reject the null hypothesis and conclude then, that there is no difference in the pass rates based upon whether the paramedic holds the AAS degree or a certification.

3. Using the statistical output below, please comment on the difference in the pretest and post-test scores of the paramedics completing the BTLS course. The variable PRETEST is the score on the pretest and is measured on an interval scale between 0 and 100%. The variable POSTTEST is the score on the post-test and is measured on an interval scale between 0 and 100 Assume the data are normally distributed and that alpha is 0.05.

t-tests for Paired Samples

Number of 2-tail

Variable pairs Corr Sig Mean SD SE of Mean

-------------------------------------------------------------------------------

POSTTEST 84.9600 7.877 1.575

25 .706 .000

PRETEST 77.5600 11.373 2.275

-------------------------------------------------------------------------------

Paired Differences

Mean SD SE of Mean t-value df 2-tail Sig

----------------------------------------------------------------------------

7.4000 8.052 1.610 4.60 24 .000

95% CI (4.076, 10.724)

• ANSWER: Because the data are measured on an interval scale, are dependent, and are normally distributed, the paired T test is appropriate. The average score on the pretest was 77.56 and the average score on the post-test was 84.96. The P value is 0.000, which is less than alpha. We reject the null hypothesis and conclude that the observed difference between pretest and post-test scores is a true difference and not the result of random sampling error.

4. Using the statistical output below, please comment on the differences in the test scores on the state written exam, between paramedics graduating from an AAS degree program, and registered nurses who have challenged the paramedic exam through the nurse-to-paramedic bridge option. The variable POSTTEST is the score on the written exam, measured on the interval scale between 0% and 100%. The variable BRIDGE is coded as 0 for the paramedics, and 1 for the registered nurses. Assume all data are normally distributed and that alpha is 0.05.

t-tests for Independent Samples of BRIDGE

Number

Variable of Cases Mean SD SE of Mean

-----------------------------------------------------------------------

POSTTEST

BRIDGE 0 13 83.6154 9.394 2.606

BRIDGE 1 12 86.4167 5.885 1.699

-----------------------------------------------------------------------

Mean Difference = -2.8013

Levene's Test for Equality of Variances: F= 2.692 P= .114

t-test for Equality of Means 95%

Variances t-value df 2-Tail Sig SE of Diff CI for Diff

-------------------------------------------------------------------------------

Equal -.88 23 .386 3.168 (-9.354, 3.751)

Unequal -.90 20.36 .378 3.110 (-9.282, 3.680)

-------------------------------------------------------------------------------

• ANSWER: Because the data are measured on an interval scale, are independent, and are normally distributed, the independent (unpaired) T test is appropriate. The mean test score of the paramedics is 83.61 and the mean test score of the nurses is 86.41. To test the significance of this difference, we must first determine if the variances are equal. The null hypothesis is that there is no difference in the variances between the nurses and the paramedics. We test this hypothesis using Levene’s test for equality of variances. The P value for this test is 0.114, which is greater than alpha. We therefore accept the null hypothesis and conclude that the variances are equal and that the observed difference is purely the result of random sampling error. Now that the issue of variances has been settled, we can test for the statistical significance of the observed differences in test scores. Reading from the line for equal variances, the P value is 0.386, which is greater than alpha. We accept the null hypothesis that there is no difference in the test scores between new graduates of an AAS program and nurses challenging the paramedic exam.