STATISTICS EXCLUSIVES PGQUESTIONS

STATISTICS  /PGQUESTIONS

Q. A screening test is performed and there are 10 positive results. Of these, six are false positives. Of note, there are exactly 10 people with disease in the population screened. Which of the following is true?

A. This test has a 40% sensitivity.

B. This test has a 60% sensitivity.

C. This test has a 100% sensitivity.

D. This test has a 60% specificity.

E. This test has a 40% specificity.

Ans:. A.

We know that there are six false positives. Thus, the remaining four positive tests must be true positives. Since there are exactly 10 people with disease, there must be six false negatives. Using this information we can calculate the sensitivity, but not the specificity. Sensitivity is a/(a + c)= 4/10 =0.4, or 40%.

Q. A screening test for diabetes is being evaluated. You know that 900 nondiabetics and 100 diabetics are being screened. There are 100 positive tests and of these, 90 have diabetes.

Which of the following is true in this population?

A. The NPV of the test is 0.9.

B. The NPV of the test is 0.5.

C. The PNV of the test is 0.1.

D. The PPV of the test is 0.1.

E. The PPV of the test is 0.9.

Ans: E.

Ninety of the 100 positive screens have diabetes, so they must be true positives. The remaining

10 must be false positives. Since there are 100 people with diabetes, there must be 10 false negatives as well. So we can calculate PPV but not NPV. The PPV is equal to 90/100 = 0.9.

Q. Given the following collection of data {1,2,4,6,7,8,9}, which is correct?
A. The mean is 5.0.
B. The median is 5.0.
C. The median is 5.5.
D. The median is 6.0.
E. The mean is 6.0.

Ans:. D.

 Given the set of numbers {1,2,4,6,7,8,9}, the sum of these numbers is 37 and there are seven values, so the mean is 37/7 =5and2/7, or 5.28. The median, which is the center of an odd number of values, is the number 6, with 1, 2, and 4 below and 7, 8, and 9 above.

Q. Given the following collection of data {1,1,1,1,2,3,5,5, 6,6,7,7,9,9}, which of the following is correct?
A. The mode is 1.
B. The mean is 4.5.
C. The median is 6.
D. A and B are correct.
E. A and C are correct

Ans. D.

Given the set of numbers {1,1,1,1,2,3,5,5,6,6,7,7,9,9}, there are 14 numbers, and their sum is 63. Thus the mean is 4.5; clearly, the median is 5; and the mode is 1. Thus, the answer is (D), which is a mode of 1 and a mean of 4.5.

Q. You are told that your patient has a cholesterol result that is the mean of that in the population. Which of the following distributions would also mean that his result is necessarily
greater than half of the population?
A. Skewed to the left
B. Skewed to the right
C. Uniform
D. Normal
E. Bimodal

Ans:  B.

The question asks, “In what type of distribution is the mean greater than the median?”The mean will equal the median in normal and uniform distributions. In a symmetric bimodal distribution this will also be true. In a skewed-to-the-left distribution, the median will be greater than the mean, whereas in a skewed-tothe- right distribution, the mean will be greater than the median

Q. Which of the following statements is true about a 95% confidence interval?
A. There is a 95% chance that the true value is in the interval.
B. Ninety-five percent of the time, the true value will fall in the interval.
C. If you repeated the study 20 times, the true value would be in the interval at least once.
D. If the interval does not contain the null hypothesis, you can be 95% certain that your result is the true value.
E. If you repeat the study an infinite number of times, the results would fall in the interval 95% of the time

Ans;E.

What a 95% CI tells you is that if you repeated the same experiment that you did to get your summary statistic (e.g., the mean or proportion), an infinite number of times, 95% of the time you would get a value that falls within the 95% CI. It does not tell you that the true value you are seeking is going to fall within your CI, particularly because your experiment may be faulty or biased; thus, the true value may be missed by the CI. Even with a perfectly performed experiment, the CI is not used to tell you about the true value; it gives just the results of that

particular experiment.

Q. You are designing a study to investigate whether ethnicity is associated with type II DM. You have a cohort of 1000 patients of Asian, African American, Caucasian, and Latino ethnicity. Which of the following tests would be the best to compare the results of the study?
A. Student t test
B. Power analysis
C. z test
D. Chi-square test
E. Bonferroni correction

Ans. D.

In this study design, you are attempting to determine whether there is a difference in any of these groups of patients with respect to the risk for type 2 diabetes. The test that is needed must be able to compare proportions between multiple groups. The only one that can do that is the chi-square test. The student t test is to compare two means; the z test is to compare the proportions of two groups. A power analysis is performed to measure the power of a particular study to reject the null hypothesis. A Bonferroni correction is performed when making more than one comparison, using, for example, a student t test. Using a cutoff of the test that allows a 5% chance of type I error goes awry when you make many comparisons of outcomes in the same study. For example, if you compared 100 different means, on average you would expect five of them to meet the p _ .05 requirement. Thus, a Bonferroni or, more commonly, the student- Newman-Keuls test can be used to adjust the cutoff when making multiple comparisons.