Bias And Random Error

Question 1

Error is defined as

Answer 1

The difference between the true value of a measurement and the recorded value of the measurement.

Question 2

Random error definition

Answer 2

Variability. Also known as random variation or noise in the system.

Question 3

Bias or systemic error definition

Answer 3

Deviations that are not due to chance alone.

Question 4

Example of a systemic

Answer 4

Measuring device that is improperly calibrated.

Question 5

Key difference between random error and bias.

Answer 5

Random error has no preferred direction, so increasing observation number decreases random error. Bias has a direction so increasing observation number does not reduce bias.

Question 6

Random error corresponds to

Bias corresponds to

Answer 6

Random error corresponds to imprecision.

Bias corresponds to inaccuracy.

Question 7

What’s a null hypothesis

Answer 7

Typically the null hypothesis reflects the lack of an effect in the alternative hypothesis reflects the presence of an effect.

Question 8

What is a two-sided alternative?

Answer 8

Two sided alternative only indicates that the two outcome groups are different not that one is better or worse specifically. Ie hypothesis is that treated (group A) is different than untreated (group b). Says nothing about better or worse.

Question 9

Two sample t test, what does it do?

Answer 9

It creates a signal to noise ratio. T= (Xa - Xb) / (standard error of Xa -Xb). Where (Xa - Xb) is the difference in means of control vs exp group.

Question 10

How is the t value useful?

Answer 10

We can use the t value to generate the P value.

Question 11

Define P value

Answer 11

the p-value is the probability of observing a t value as extreme or more extreme than the t value actually observed if the null hypothesis is true.

Question 12

Type 1 error vs type 2 error

Answer 12

Type 1 error is rejecting the null hypothesis when it is true.type II error is failing to reject the null hypothesis when it is false.

Question 13

How to determine appropriate sample size for a study?

Answer 13

1) determine the effect size of interest. Ie how much change in cholesterol levels do we care about? Must be clinically meaningful
2) sample size should have good statistical power ie B=0.1 or 0.2.
3) should have significance level a= 0.05. (p value)
4) na=21ä^2/b2. Ie n= 21 (standard deviation squared) / (meaningful difference squared).
Standard deviation would be taken from another study.

Question 14

Low statistical power is type 1 or type 2 error?

Answer 14

Type 2 error.

Question 15

How to calculate a confidence internal with 95% confidence

Answer 15

It’s the difference between the 2 groups (eg 2.5 mg/ml cholesterol levels difference between treated and noon treated group) +/- (1.96 x standard error).

Question 16

What is a confidence interval?

Answer 16

Confidence intervals provide a plausible range of values for a population measure.

Question 17

List five sources of bias in clinical studies

Answer 17

Selection bias,
Procedure selection bias, 
post-entry exclusion bias, 
bias do to selective loss of data, 
assessment bias.

Question 18

Selection bias diminishes internal or external validity?

Answer 18

External

Question 19

Selection bias definition

Answer 19

Selecting a sample that is not representative of the population.

Question 20

Procedures selection bias definition

Answer 20

Investigators decide on treatment assignments to specific types of patients.

Question 21

Post entry exclusions is

Answer 21

Changing exclusion criteria

Question 22

Selective loss of data bias and how to combat it.

Answer 22

Excluding data due to non compliance or decisions. Intent to treat analysis analyses all randomized subjects.

Question 23

6 tools statisticians have to combat bias and (which type of bias they diminish).

Answer 23

1. Randomization (procedure selection bias)
2. Masking (assessment bias)
3. Concurrent controls (minimizes confounding results)
4. Objective assessments (assessment bias)
5. Active follow up and endpoint ascertainment (assessment bias)
6. No post hoc exclusion (post entry exclusion bias)

Question 24

Statistical bias definition

Answer 24

Difference between the parameter to be estimated and the mathematical expectation of the estimator.