Statistics Flashcards
Reliable test
Consistent result
Valid test
Reflective of the true value
The mean
Average no
The median
Middle no in the data set
The null hypothesis
nothing of statistical significance happened
The alternative hypotheisis
Something of statistical significance did happen and we can therefore reject the null hypothesis
When to use z statistic
Examining a sample taken from a population with a known standard deviation
When to use the t statistic
Sample of a population with unknown standard deviation (usual case)
P value
The chance of finding the value in the control population
A p value of <0.05 means that there is less than a 5% chance of rejecting the null hypothesis when it is actually true
i.e <5% chance the test statistic is due to chance alone rather than statistically different from the control
Type 1 error
Rejecting the null hypothesis when it is true
= False positive
Type 2 error
Rejecting the alternative hypothesis when it is true
= False negative
Changes that lead to an increase in power
Increase in sample size
Increase in significance level
Increase in the detected difference
A confidence level of 95%
investigators are 95% confident that the mean value of the population is between this interval (5% of time it will not!)
When to use mean
parametric data
When to use median (middle no)
Non-parametric data
eg. ordinal data, binary date
The issue with post hoc analysis
Inflation of p value and risk of type 1 error
Relative risk
risk of outcome in intervention group/risk of outcome in control
Relative risk reduction
1 - RR
Absolute risk reduction
Risk of outcome in control minus risk of outcome in the intervention group
NNT
Number of people need to undergo the intervention in order to prevent the outcome in one person
NNT = 1 / ARR
Cross-sectional study
Examines the relationship between diseases (or other health- related characteristics) and other variables of interest as they exist in a defined population at one particular time i.e. exposure and outcomes are both measured at the same time.
Case-control study
Identifies those with the disease and without the disease and compares how they differ in terms of an explanatory variable
Cohort study
Follows exposed and unexposed people
Reason for testing mortality rate over survival time in screening test
Leading bias and length bias
Allocation bias
Researches allocate which intervention the participant is to receive
Channeling bias
When a patient’s prognosis influences which group they are allocated to
Ascertainment bias
Type of sample bias, the study population does not represent true population
Interviewer bias
A systematic difference in how information is obtained, recorded or interpreted
Chronological bias
Difference between those recruited earlier in the process and those recruited later
Recall bias
Paticipants do not remember previous events
Attrition bias
Participants leave the study
Placebo
improved performance due to intervention
Nocebo
Worsened performance due to intervention
Hawthorne effect
participants report improvement because they know they are on the drug/ or being observed
Likelihood ratio of a positive test
sensitivity / (1-specificity)
likelihood ratio of a negative test
(1-sensitivity) / specificity
False positive rate = (type I error)
1 − specificity
False negative rate = (type II error)
1 − sensitivity
Effect of decreased prevalence
increases NPV & decreases PPV
Effect of increased prevalence
decreased NPV ad increased PPV
True positive rate
sensitivity
True negative rate
specificity
Focus of phase 1 trials
pharmacology
Focus of phase 2 trial
safety
Focus of phase 3 trial
efficacy
Focus of phase 4 trial
long term safety
Population attributable factor
PAFI = [Pe (RR - 1)] divided by [1 + Pe (RR - 1)].
Pe= % of the population exposed to the risk factor
Hazard ration
Time from recruitment until event
