Lecture 12 9/18/24

Question 1

What is the null hypothesis?

Answer 1

there is no difference between/among groups

Question 2

What is the alternative hypothesis?

Answer 2

there is a difference between/among groups

Question 3

What is random error?

Answer 3

-absence of precision
-may result in underestimation or overestimation of true value
-measurement device has precision limitations

Question 4

What is sampling error?

Answer 4

the difference between the measured value and the true value when random error occurs

Question 5

What is systematic error/bias?

Answer 5

-absence of validity
-error having a nonzero mean, so that its effect is not reduced when observations are averaged
-deviations from true value all occur in the same direction

Question 6

What are the characteristics of chance?

Answer 6

-random error occurs by chance
-not predictable
-always present
-increased sample size decreases random error

Question 7

What is a type 1 error with respect to the null hypothesis?

Answer 7

-rejecting the null hypothesis when it is true
-we find a difference where none exists

Question 8

What is a type 2 error with respect to the null hypothesis?

Answer 8

-accepting the null hypothesis when it is false
-we do not find a difference where one exists

Question 9

What is a p-value?

Answer 9

-probability that the observed differences could be due to chance
-probability of making a type 1 error

Question 10

What is power?

Answer 10

-probability that you will find a statistically significant difference when it exists and is of a certain magnitude
-probability of being able to identify an effect if one exists
-analogous to sensitivity

Question 11

Why is a large sample size preferred?

Answer 11

it will always increase statistical power and allow for the detection of both large and small effect sizes regardless of population variation

Question 12

Why is it important to calculate sample size?

Answer 12

to know the sufficient number of patients needed to detect a clinically important difference if it exists

Question 13

Which aspects of sample size calculation are decided by the investigator?

Answer 13

-minimum effect size
-probability of type 1 error
-probability of type 2 error

Question 14

Which aspects of sample size calculation are determined by the data?

Answer 14

-variability of the data
-rate of events

Question 15

What non-statistical considerations must be taken into account when determining sample size?

Answer 15

-resources
-objectives of the study

Question 16

What statistical considerations must be taken into account when determining sample size?

Answer 16

-precision of the estimate
-effect size
-expected variation in the data
-level of confidence
-significance level/type 1 error
-power to detect real effects

Question 17

Why do continuous outcomes possess the most precision and accuracy?

Answer 17

due to the “true zero” that allows for measurement of both distance and magnitude

Question 18

Why do categorical outcomes and ordinal outcomes lack precision and accuracy?

Answer 18

due to observation biases

Question 19

What must be done to increase precision?

Answer 19

use a larger sample size

Question 20

What are the benefits of a larger effect size?

Answer 20

-increased statistical power
-decreased sample size

Question 21

What is effect size?

Answer 21

difference between means or proportions due to treatment

Question 22

How does population variation impact sample size?

Answer 22

the more variation expected of the outcome, the larger the sample size required to draw accurate conclusions

Question 23

What is a confidence interval?

Answer 23

estimated range of values which is likely to include an unknown population parameter (typically 95% likely)

Question 24

How does a confidence interval differ from a point estimate?

Answer 24

a point estimate is a single value given to estimate a population parameter, while a confidence interval provides a range of values that is likely to contain the point estimate

Question 25

What is suggested when the confidence interval contains the null value?

Answer 25

suggests that the parameter is not statistically significant from the null

Question 26

What are the characteristics of underpowered studies?

Answer 26

-small sample size
-statistically non-significant results and low power make studies inconclusive