Intro to Biostatistics Flashcards
What are the 3 primary levels (groupings) for variables?
- Nominal
- Ordinal
- Interval or Ratio
What are the 3 attributes of data measurement that determines the groups?
- Order/Magnitude
- Consistency of scale/ Equal distances
- Rational absolute zero
Nominal
- Dichotomous/Binary
- Non-ranked
- Named categories
Ordinal
- Ordered
- Rank-able categories
- Non-equal distance
Interval/Ratio
- Order and magnitude
- Equal distances
Which levels of measurement are known as discrete?
- Nominal
- Ordinal
Which level is known as continuous?
Interval/Ratio
After data is collected, we can appropriately go ___________ in specificity/detail of data measurement (levels) but never _________!
Down,
Up
Mode
Most common
Median
Middle when numbers are in order
Mean
Average
Interquartile Range
Range of the middle, innermost 50%
What is most useful in ordinal data?
- Mode
- Median
What is most useful in nominal data?
- Mode
What is most useful in interval?
- Mean
But all three can be used
Variance
The average of the squared differences in each individual measurement and the mean.
the mean - measurement divided by the number of groups.
Standard Deviation (SD)
Square root of variance value
What are stats tests that are useful for normally-distributed data called?
Parametric tests
Positively Skewed
- When the mean is higher than the median
- Tail pointing to the right
Negatively skewed
- Mean lower than the median
- Tail pointing to the left
Skewness
A measure of the asymmetry of a distribution
What is the skewness value of the perfectly-normal or symmetric distribution?
Zero
Kurtosis
A measure of the extent to which observations cluster around the mean
What is the kurtosis value for a normal distribution?
Zero
Wha does a positive kurtosis mean?
More cluster
What does a negative kurtosis mean?
Less cluster
What are the required assumption of Interval/Ratio data?
- Normally-distributed
- Equal variances
- Randomly-derived and independent
What is the purpose of Levene’s test?
To assess equal variance between groups
How is data that is not normally-distributed handled?
- Use a statistical test that does not require the data to be normally distributed
- Transform data to a standardized value (z-score or log transformation)
Null Hypothesis
Researcher perspective which states there will be no true difference between the groups being compared
Type 1 Error
- Alpha
- rejecting null hypothesis when it is true and should be accepted
- claiming difference when there is none
Type 2 Error
- Beta
- Accepting the null hypothesis when it is actually false and you should not have accepted it
- claiming there is no difference when there is one
1 SD above and below the parameters will represent ___ % of all people in the study.
68%
2 SD above and below the parameters will represent ___ % of all people in the study.
95%
3 SD above and below the parameters will represent ___ % of all people in the study.
99.7%
Power
- 1 - beta
- the statistical ability of a study to detect a true difference if one truly exists
- 80% by default
What percentages of type 2 error is acceptable?
20%
Sample size and power
the larger the sample size, the greater the likelihood of detecting a difference if one truly exists
increase in power
What percentages of type 1 error is acceptable?
5%
If a p value is ______ than the pre-selected alpha, (0.05, 5%), then we say that it is statistically significant
lower
Interpretations of a p value
- the probability of making a type 1 error if the null hypothesisi is rejected
- the probability of wrongly claiming a difference between groups when one doesn’t really exist
- the probability of obtaining group differences as great or greater if the groups were actually the same
- the probability of obtaining a test statistic as high/higher if the groups were actually the same/equal
Confidence Intervals
the range in which you are a certain percent confidant that the real difference lies
What are the 4 key questions to select the correct statistical test?
- What data level is being recorded
- What type of assessment is desired?
- How many groups
- Is the data independent or related
What is the name for the correlation test for nominal data?
Contingency Coefficient
What is the name for the correlation test for ordinal data?
Spearman Correlation
What is the name for the correlation test for inteval data?
Pearson Correlation
What key words should you look for in a survival test?
- time to event
- event occurrence
What is the name for the survival test for nominal data?
log-rank
What is the name for the survival test for ordinal data?
Cox-proportional Hazards test
What is the name for the survival test for interval data?
Kaplan-Meier test
What are the key words for a regression test?
- prediction
What is the name for the regression test for nominal data?
Logistic Regression
What is the name for the regression test for ordinal data?
Multinomial Logistic Regression
What is the name for the regression test for interval data?
Linear Regression
What is the nominal test for independent data with 2 and 3 or more groups?
- Chi square
- Fisher’s exact for cells less than 5
What is the post hoc test for nominal data
bonferonni test
What is the nominal test for related data with 2 groups?
McNemar test
What is the nominal test for related data with 3 or more groups?
Cochran
What is the ordinal test for independent data with 2 groups?
Mann-Whitney test
What is the ordinal test for independent data with 3 or more groups?
Krusal-Wallis test
What is the ordinal test for related data with 2 groups?
Wilcoxin Signed Rank test
What is the ordinal test for related data with 3 or more groups?
Freidman
What are the post hoc tests for ordinal data with 3 or more groups?
- Student-Newman-Keul test
- Dunnett
- Dunn
What is the interval test for independent data with 2 groups?
Student t test
What is the interval test for independent data with 3 or more groups?
- ANOVA
- MANOVA ( 2 or more dependent variables)
What is the interval test for independent data with 3 or more groups with confounders?
- ANCOVA
- MANCOVA
What is the interval test for related data with 2 groups?
Paired t test
What is the interval test for related data with 3 or more groups?
- RM ANOVA
- RM MANOVA
What is the interval test for related data with 3 or more groups with confounders?
- RM ANCOVA
- RM MANCOVA
What are the post hoc tests for interval data with 3 or more groups?
- Student-Newman-Keul test
- Dunnett
- Dunn
- Tukey or Sceffe tests
- Bonferroni
Student-Newman-Keul test
- compares all pairwise comparisons possible
- all groups must be equal in size
Dunnett test
- compares all pairwise comparisons against a single control
- all groups must be equal in size
Dunn test
- compares all pairwise comparisons possible
- useful when all groups are not equal in size
Kappa statistic
correlation test showing relationshio of agreement between evaluators
-