Exam 3 : Biostatistics Flashcards
A research perspective which states there will be NO difference between the groups being compared
Null hypothesis
What are the three primary level for variables based on answering 2 key attributes in stats?
Nominal
Ordinal
Interval or ratio
That are the three key attributes of data measurment?
- Order/ magnitude
- Consistency of scale/ equal distances
- Rational absolute zero
What level of measurement can be described as dichotomous/ binary; Non-ranked named categories?
Nominal
What level of measurement can be described as Ordered, rank-able categories; non-equal distance?
Ordinal
What level of measurement can be described as having order & magnitude & equal distances?
Interval/ ratio
Which of the two, interval or ratio, has an arbitrary zero value ( 0 doesnt mean absence)
Interval
Which of the two, interval or ratio, has an absolute zero value? ( o means absence of measurment value)
Ratio
T/F
After data is collected, we can appropriately go up in specificity/ detail of data measurements, but not down.
False
What is descriptive statistics?
Non-comparative, simple description of various elements of the study’s data such as : Age, education, living situation, native, number of people..ext..
The average of the squared-differences in each individual measurement value and the groups’ mean is what calculation? (From mean)
Variance
The square root of variance value is what calculation?
Standard deviation
T/F
Normally distributed graphical data is what symmetrical
True
What do the mean/ median/ mode look like in a normally distributed dataset?
Equal/ near equal
What are stats test useful for normally-distributed data called?
Parametric tests
What are parameters?
Mean/ median/ mode
1 standard deviation on both sides of the mean (up and down) is approximately what percentage?
68%
2 standard deviations on both sides of the mean is approximately what percentage?
95%
3 standard deviations on both sides of the mean are approximately what percentage?
99.7%
What data NEEDS to be normally distributed?
Interval
When the mean is higher than median what type of skew does this make?
Positive skew
What does “skew” mean graphically?
The graph represents an asymmetrical distribution with one ‘tail’ longer than the other
When the mean is lower than the median the graphical data is known as what?
Negatively skewed
When the tail is on the left what type of skew is this?
Negative skew
When the tail is on the right, what type of skew is this?
Positive skew
Outlier can move the mean, median, mode?
Mean
What is a skewness value?
A measure of the asymmetry of a distribution
What does a graph with a skewness value of zero look like?
Evenly distributed
What would a graph with -2.0 skewness value look like?
Negative skew, so tail on left
What is known as the measure of the extent to which observations cluster around the mean?
Kurtosis
When the kurtosis value is 0, what does the graph look like?
Normally distributed
What does a positive kurtosis imply
More clusters
What does a negative kurtosis imply?
Less clusters
What data would be considered discrete?
Nominal
Ordinal
What data would be considered continuous?
Interval
What test tell us if interval data is normally distributed/ of equal variances between the groups?
Levene’s test
What do we do when data is NOT normally-distributed?
- Use statistical test that doesn’t require parametric test (non parametric tests)
- Transform data to a standardized value ( z-score or log)
What do we always run, even if the data isn’t normally distributed?
Descriptive statistics & graphs
What error is it when the researcher rejects the null hypothesis when it is actually TRUE?
Type 1 error ( alpha error)
What error is it when the researcher accepts the null hypothesis when it is actually FALSE?
Type 2 error ( beta error)
The statistical ability of a study to detect a true difference, IF one truly exists between group-comparisons. The larger the sample size, the greater likelihood hood. What is this?
Power
T or F
The smaller the difference between groups necessary to be considered “significant”, the greater sample size needed.
True
What are the three sample size determiners?
- Minimum difference between groups deemed significant
- Expected variation of measurment
- Type 1 & 2 error rates & confidence intervals
What are the steps to getting a p value?
- Critical value calculated
- Compared to appropriate table of probabilities
- Probability value is obtained
T or F
P value is based on the probability of observing, due to chance alone, a test statistic value as extreme or more extreme than actually observed if groups were similar
True
T/ F
P value is picked after the study is run
False
When is p value statistically significant?
When it is lower than the picked value (5%)
If the p value is lower than the preselected p value, making the information statistically significant, then what do we do with the null hypothesis?
Reject the null
What are 4 interpretations of a pre-set (a priori) p value?
- Probability of Type 1 error
- Probability of erroneously claiming a difference
- Probability of obtaining group differences as great or greater if the groups were actually the same/equal
- Probability of obtaining a test statistic as high/higher if the groups were actually the same/equal
What are the 3 required assumptions of interval/ratio data?
- Normally distributed
- Equal variances
- Randomly-derived & independent
What are the most common selections for a confidence interval?
90%
95%
99%
What are the two things that confidence intervals are based on?
- Variation in sample
2. Sample size
T/F
If we hope over 1.0 in the confidence interval ( go from neg to pos) then it is statistically significant
False ( if it crosses over 1.0, it ISN’T statistically significant)
What does a 95% confidence interval mean?
We are 95% confident that the “true” difference or relationship between the groups is contained within the confidence interval range
T/F
Statistical significance doesn’t confer meaningful clinical significance
True, it doesn’t. Statistical significance really has no magnitude, it only tells us that there is or isn’t a probability of a difference between groups
What is the first key question when determining the correct statistical test?
What data level is being recorded?
- order/ magnitude
- equal, consistent distances
What is the second question to ask to correctly select a statistical test?
- What type of comparison/assessment is desired?
- correlation
- event/occurrence/ time to event
- outcome prediction/ association (OR)
- frequencies/counts/proportions
What is the third question to ask to correctly select a statistical test?
- How many groups are being compared
- 2 0r 3
What is the 4th question to ask to select the correct statistical test?
- Is the data independent or related?
This test provides a quantitative measure of the strength (magnitude) & direction of a relationship between variables
Correlation
What is the nominal correlation test?
Contingency coefficient
What is the ordinal correlation test?
Spearman correlation
What is the interval correlation test?
Pearson correlation
Greater than 5% for a Pearson correlation just means there is no linear correlation; what may still be present though?
Non-linear correlation
This correlation test controls for confoundings
Partial correlation
What happens when a p value is greater than 5%?
It’s not linear
What is the time to event/ event occurrence test ?
Survival test
This test compares the proportion of events over time, or time to events, between groups
Survival test
What is a nominal survival test?
Log-rank test
What is the ordinal survival test?
Cox-proportional hazards test
What is the interval survival test?
Kaplan-Meier test
How can we represent all survival tests graphically?
By a Kaplan Meier curve
What is the outcome PREDICTION/ association test?
Regression
This test provides a measure of the relationship between variables by allowing the prediction about the dependent, or outcome, variable knowing the value/category of independent variables
-also able to calculate OR for a measure of association
Regression
What is the nominal regression test?
Logistic regression
What is the ordinal regression test?
Multinomial logistic regression
What is the interval regression test?
Linear regression
For nominal data of greater than 5 cell counts and 2 or more groups, what test should we use?
Chi square
For nominal data of less than 5 cell counts and 2 or 3 groups what test should we use?
Fisher’s exact test
T/F
For statistically significant findings in nominal data of 3 or more comparisons, one must perform subsequent analysis to determine which groups are different
True
What subsequent tests should be run on a nominal data that 3 or more comparisons are found to be statistically significant?
Bonferroni test of inequality
Bonferroni correction
This test adjusts the p value for number of comparisons being made. It is very conservative
Bonferroni test of inequality
When using nominal related data in 2 groups, what test should we run?
McNemar test
When using nominal related data in 3 or more groups, what test should we use?
Cochran
When using ordinal independent data in 2 groups, what test should we run?
Mann-Whitney test
When using ordinal independent data in 3 or more groups, what test should we use?
Kurskal-Wallis test
When running a Kruskal Wallis test, when 3+ group comparisons are significant then we must perform a ______________ to determine where differences are.
Post-hoc test
When using ordinal paired data of 2 groups, what test do we run?
Wilcoxon signed rank test
When using ordinal paired data with 3 or more groups, what test should we run?
Friedman test
What are 3 post hoc tests for 3 or more group comparisons in ordinal data?
Student Newman Keul Test
Dunnett test
Dunn test
Which post hoc test compares all to a single control
Dunnett test
What post hoc test compares all possible comparisons and the groups must be equal in size?
Student Newman Keul test
What post hoc test compares all comparisons possible but the groups are not of equal size?
Dunn test
When using independent interval data from 2 groups, what test should we run?
Student t-test
When using independent interval data from 3 or more groups, what tests should we run?
ANOVA
MANOVA
T/F
ANOVA test compares the means of all groups against a single DV
True
T/F
MANOVA test compares the means of all groups against multiple DVs?
True
When using related interval data from 2 groups, what test should we run?
Paired t test
When using related interval data from 3 or more groups, what tests should we run?
ANOVA
MANOVA
T/F
Turkey test is a post hoc test for interval dat that is slightly more conservative then the Stu.N.K test.
True
T/F
The scheffe test is a post hoc test for interval data that is less affected by violations informality and homogeneity of variances. It is considered the most conservative.
True
This correlation test shows relationship or agreement between evaluators (consistency of decisions, determinations)
Kappa statistics
what does a kappa test of +1 mean?
The observers perfectly classify everyone exactly the same way
What does a kappa test of 0 mean?
There is no relationship at all between the observers classifications, above the agreement that would be expected by chance