Midterm 3: Week 11,12,13,14 Flashcards
Give an example of a Point Estimate Value
An example of a Point Estimate Value is a mean, the parameter this estimates is the real mean; μ (myu)
Define a Point Estimate
A point estimate is a single numerical value that is used to estimate a corresponding population parameter
What are Point Estimates usually accompanied with?
Point Estimates are usually accompanied with other numerical descriptors
Why is it important that Point Estimations are accompanied with other numerical descriptors?
It is important that Point Estimations are accompanied with other numerical descriptors because it allows to see if the Point Estimation is a reasonable approximation
What are the 2 Numerical Descriptors that accompany Point Estimations?
- The strength of an estimation; p-value
2. The possible range of our estimation; confidence interval
What are the 2 tests that we are interested in when we’re looking at the relationship between 2 variables
- Measure of association: how strong the hypothesized association is between 2 variables
- Significance test: how likely the relationship between the 2 variables is due to chance
What does the Null Hypothesis state?
The Null Hypothesis states that there is no difference in the exposed and unexposed groups; there is no association between the 2 variables
What does the Alternate Hypothesis state?
The Alternate Hypothesis states that there is a difference between the exposed and unexposed group; there is an association between the 2 variables
Define a Test Statistic
A value calculated from the data that is used to evaluate the evidence in support of the null hypothesis.
Define a Significance Level
A Significance level is a standardized level of probability that we compare our test statistic to (α-value). If the test statistic is smaller than the significance level, then we reject the null hypothesis.q
What does the P-Value give a measure of?
The P-Value gives a measure of how confident we are about a point estimate
What does it mean to get a P-Value of 0.05?
A P-Value of 0.05 would mean that there is less than a 5% chance that outcome we observed is due to chance; 95% chance that the outcome we observe is not due to chance.
How do we know what an appropriate p-value to use is?
We must look at what the possible gains, losses and circumstances are
Why is it better to have a large sample size?
It is better to have a large sample size because the larger the sample size the more likely it is that the random error of one person will cancel out the random error of another person, meaning there is less net random error
Define a Confidence Interval
A confidence interval is a measure of uncertainty about the true value of a the estimate; we can never be 100% about the true value of a estimate
Is it better to have a wide confidence interval of narrow?
Narrow
What does a Confidence Interval of 1 or greater indicate?
A Confidence Interval of 1 or greater indicates that the cause-effect relationship may not be accurate
Are significance tests always black and white? ie: if a Significance Test shows that there is not a causal relationship, or there is, that is correct?
No, in a very large sample size the test may be very sensitive to differences and mislead the overall relationship and make it loook like there is no relationship when there really is one.
Define Statistical Significance
Statistical Significance is when we mathematically test the outcomes of a study
Clinical Significance
The relevance of research to individuals; builds an understanding of a person
Public Health Significance
The relevance of a research to the population as a whole; the government etc
Define Statistical Power
The ability to show association between 2 variables if it exists
What are 3 ways Random Error can occur?
- Poor Precision
- Sampling Error
- Variability of Measurement
What is Poor Precision?
Not having accurate measurements
What is Sampling Error?
Sampling Errors occur when the samples are not representative of the population, this can happen because of poor study design, chance, or small sample size
What can minimize random error?
We can minimize random error by having a large sample size, and repeating measurements/training interviewers
Can random error be completely eliminated?
No, it can not
Error graphs: Big box, dead centre means what?
Random error is high, but systematic error is low
Error graphs: big box, skewed left means what?
Random error and Systematic error
Error graphs: Small box, skewed left means what?
Low random error, but high sampling error
Error graphs: Small box, dead centre means what?
Low random error, and low sampling error
Define Systematic Bias
Systematic Bias is error that results from poor study design, protocol, analysis of the study
If Random Error is due to chance, what is Systematic Bias due to?
Systematic Bias is do to the researchers fault
What are the 2 main types of Bias?
Selection bias, and Information Bias
Define Selection Bias
Selection Bias is a bias that occurs due to the way the exposed and unexposed (cases and controls) were selected
How does Selection Bias most likely happen?
Selection Bias most likely happens when subjects are chosen from a sub-group and therefore are not representative of the population. Ex: it would not be appropriate if most of your subjects were chosen from a gym you frequent, if the population of interest was all of Canada
Define Information Bias
Errors made in measuring or classifying subjects
Name 6 examples of Information Bias?
- Recall Bias; subject misremember
- Misclassification Bias; mix up exposed and unexposed etc
- Wish Bias; change memories because of affection
- Surrogate Interview Bias; getting information from a 2nd source
- Surveillance Bias; Look as one group closer, or less close
- Hawthorne Bias; Those in studies act more cooperatively than the population would
Case-control studies are more likely to be affected by what kind of bias?
Case-control studies are more likely to be affected by selection bias
Cohort studies are more likely to be affected by what kind of bias?
Cohort studies are more likely to be affected by information bias
Define Confounding
Confounding is the distortion of an effect measure because of an other factor that is not part of the causal pathway
What is an important note to remember about confounders?
The third variable (the suspected confounder) must be related to both the exposure and the outcome
What is a good example of a confounder?
Alcohol consumptions effect on risk of lung cancer, the confounder here would be smoking because those who consume alcohol also commonly smoke
Why would hours underground, not be a confounder for lung cancer caused by asbestos mining?
Hours underground would not be a confounder because it is part of asbestos mining; part of the causal pathway
What are 5 ways to deal with confounding?
- Randomization
- Restriction
- Matching
- Stratification
- Multivariate analysis
How is Randomization used to decrease the effect of confounders?
Randomization more equally distributes confounders in the exposed and unexposed group, cancelling out their net impact.
How is Restriction used to decrease the impact of confounders?
Restriction restricts the differences in characteristics of the subjects they study in the exposed and unexposed groups so that if everyone has that characteristic, everyone is affected equally. Ex: If age is a confounder, subjects in the exposed and unexposed groups would all be of similar age
How is Matching used to decrease the effect of confounders?
Matching forces confounder to be equally distributed on the exposed and unexposed sides by matching a unexposed individual almost exactly to a exposed individual
How is Stratification used to decrease the impact of confounders?
Stratification analyzes the exposure and outcome for every confounder, to understand the effect the confounder has on the exposure and outcome.