Epidemiology Flashcards
What statistical test do you use for interval data for 3 or more groups of independent data? What is the non-parametric alternative? What do you use if a confounder is present?
- ANOVA (1 DV) or MANOVA (2 or more DVs)
- non-parametric alternative: Kruskai-Wallis
- if confounder is present: ANCOVA (1 DV) or MANCOVA (2 ore more DVs)
What statistical test do you use for interval data for 2 groups of paired/related data? What is the non-parametric alternative?
Paired t-test
*non-parametric alternative: Wilcoxon signed rank
What statistical test do you use for 3 or more groups of paired/related interval data? What is the non-parametric alternative? What if confounders are present?
Repeated measures ANCOVA (1 DV) or MANCOVA (2 or more DVs)
- non-parametric alternative: Friedman
- confounders present: repeated measures ANCOVA (1 DV) or MANCOVA (2 or more DVs)
What statistical test do you use for interval data for 2 groups of independent data? What is the non-parametric alternative?
- student t-test
* non-parametric alternative: Mann-Whitney
What type of statistical test do you use for membership prediction or association for interval data? What is the non-parametric alternative?
Linear regression
*non-parametric alternative: multinomial logistic regression
What type of statistical test do you use for time-to-event or survival for interval data? What is the non-parametric alternative?
Kaplan-Meier product limit estimate
*non-parametric alternative: Cox proportional hazards
What statistically test do you use for measure of correlation of interval data? What is the non-parametric alternative?
Pearson correlation
*non-parametric alternative: Spearman correlation
What are the post-hoc tests for 3 ore more group comparisons for interval data?
Student-newman-keul test Dunnett test Dunn test Tukey or scheffe test Bonferroni correction
What statistical test is used for 2 group of independent nominal data? What is expected cell count is less than 5?
Chi-square
*less than 5: Fisher’s exact
What statistical test is used for 3 or more groups of independent nominal data? What is expected cell count is less than 5?
Chi-square
*less than 5: Fischer’s exact
What statistical test is used for 2 groups of paired/related nominal data?
McNemar
What statistical test is used for 3 or more groups of paired/related nominal data?
Cochran
What statistical test is used for proportion of events (survival) of nominal data?
Log-rank
What statistical test is used for measures of correlation of nominal data?
Contingency coefficient
What type of statistical test is used for prediction/association (regression) of nominal data?
Logistic regression
What statistical test is used for 2 groups of independent ordinal data?
Mann-Whitney
What statistical test is used for 3 or more groups of independent ordinal data?
Kruskal-Wallis
What statistical test is used for 2 groups of paired/related ordinal data?
Wilcoxon signed rank
What statistical test is used for 3 or more groups of paired/related ordinal data?
Friedman
What statistical test is used for proportion of event (survival) of ordinal data?
Cox proportional hazards
What statistical test is used for measure of correlation for ordinal data?
Spearman correlation
What statistical test is used for prediction/association (regression) of ordinal data?
Multinomial logistic regression
Describe nominal data and give an example of this type of data
- Nominal data is dichtomous/binary; non-ranked name categories
- No magnitude/consistency of scale/rational zero
- ALL data that is forced into 2 categories is nominal (diseased/non-diseased, hypertensive/non-hypertensive)
- Ex: Race
Describe ordinal data and give an example of this data type
- Ranked categories; non-equal distances
- Yes magnitude, no consistency of scale/rational zero
- Class example is categorical in nature. Survey’s that ask “how satisfied are you” are examples of ordinal data. Many people believe that pain scales are ordinal
Describe interval data and give an example of this data type
- Yes magnitude/consistency of scale
- Any biological/physiological measurement that has units behind the number is classically interval data (BP, blood cholesterol, etc.)
What type of data can the mean value be useful for?
Interval data only
What type of data can the mode value be useful for?
All 3 types of data
What is a normally distributed curve and what are the standard deviation percentages that the data will fit inside?
- Normally distributed= symmetrical; occurs when the mean/median/mode are all close or near equal
- 68.27% of data will fit inside 1 SD
- 95.45% of data will fit inside 2 SD
- 99.73% of data will fit inside 3 SD
Described positively skewed data?
When the mean is higher than the median –> tail on the graph points to the right
Describe negatively skewed data
When the mean is lower than the median –> tail points to the left on the graph
What is the null hypothesis?
A research perspective that states that there will be no (true) difference between the groups being compared. This is from the most conservative perspective. Mandates that the data that you collect will prove your data wrong and you will have to drop the hypothesis
What are the required assumptions of interval data (for proper selection of a parametric test)?
- Normally distributed
- Equal variances (Levene’s test)
- Randomly-derived and independent
What is power and how can you calculate it?
- Power is the ability to detect a true difference if one truly exists between group-comparisons
- Can be found by: 1-beta. beta is usually set at 20% –> power= 80%. most well done studies have an 80% chance of finding a difference if there is one
What are type I and type II errors?
- Type I: authors claim that there is a difference when there is not one (alpha value)
- Type II: authors claim that there is not a difference when there is one (beta value)
Define p-value
- The probability of making a Type I error if the null hypothesis is rejected
- The probability of erroneously claiming a difference between groups when one does not really exist
- The probability of the outcome of the group’s difference occurring by chance
- The probability of obtaining group differences as great or greater if the groups were actually the same/equal
- The probability of obtaining a test statistic as high/higher if the groups were actually the same/equal
What are the general characteristics of lymph follicles (nodules)?
- Are not enclosed within a capsule
- Occur singly or in aggregates
- Are sites of B cell localization and proliferation
- Transiet
Describe primary follicles (nodules)
-Are spherical, tightly packed accumulations of virgin (have not been exposed to an antigen) B cells and dendritic reticular cells
Describe secondary lymph follicles (nodules)
- Are derived from primary follicles that have been exposed to non-self antigens
- Are not present at birth
- Contain corona (cortex): dark peripheral region; composed of densely packed B lymphocytes
- Germinal center is within the corona and is a lighter stained region. Composed of B lymphocytes, memory B cells, plasma cells, dendritic reticular cells which function as antigen-presenting cells
- vascular supply: arteriole and venule supply the cortex, another arteriole and venule supply the center, lymph capillaries are not present
Why would you choose a correlation test?
To determine the strength and direction of a relationship between variables
When would you use a survival test?
To compare the proportion of, or time-to, even occurrences between groups
When would you use a regression test?
- When you want to find the relationship between variables by allowing the PREDICTION about the DV, knowing the value/category of IVs
- Can calculate OR from this tests