Bias And Random Error Flashcards
Error is defined as
The difference between the true value of a measurement and the recorded value of the measurement.
Random error definition
Variability. Also known as random variation or noise in the system.
Bias or systemic error definition
Deviations that are not due to chance alone.
Example of a systemic
Measuring device that is improperly calibrated.
Key difference between random error and bias.
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.
Random error corresponds to
Bias corresponds to
Random error corresponds to imprecision.
Bias corresponds to inaccuracy.
What’s a null hypothesis
Typically the null hypothesis reflects the lack of an effect in the alternative hypothesis reflects the presence of an effect.
What is a two-sided alternative?
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.
Two sample t test, what does it do?
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.
How is the t value useful?
We can use the t value to generate the P value.
Define P value
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.
Type 1 error vs type 2 error
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.
How to determine appropriate sample size for a study?
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.
Low statistical power is type 1 or type 2 error?
Type 2 error.
How to calculate a confidence internal with 95% confidence
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).
What is a confidence interval?
Confidence intervals provide a plausible range of values for a population measure.
List five sources of bias in clinical studies
Selection bias, Procedure selection bias, post-entry exclusion bias, bias do to selective loss of data, assessment bias.
Selection bias diminishes internal or external validity?
External
Selection bias definition
Selecting a sample that is not representative of the population.
Procedures selection bias definition
Investigators decide on treatment assignments to specific types of patients.
Post entry exclusions is
Changing exclusion criteria
Selective loss of data bias and how to combat it.
Excluding data due to non compliance or decisions. Intent to treat analysis analyses all randomized subjects.
6 tools statisticians have to combat bias and (which type of bias they diminish).
- Randomization (procedure selection bias)
- Masking (assessment bias)
- Concurrent controls (minimizes confounding results)
- Objective assessments (assessment bias)
- Active follow up and endpoint ascertainment (assessment bias)
- No post hoc exclusion (post entry exclusion bias)
Statistical bias definition
Difference between the parameter to be estimated and the mathematical expectation of the estimator.
