Biostats Flashcards
What are the three groupings for data variables?
Order/Magnitude
Consistency of scale/equal distances
Rational Absolute Zero
____ determines the appropriate statistical test
Data Type
What does order or magnitude mean?
Can you quantify the difference?
Is something bigger or smaller than something else?
What does consistency of scale or equal distances mean?
Does the value have units
Can you show the difference between values consistently
Can you calculate the range
If you see a dichotomous data set, automatically think its what type of data?
Nominal
Ex: Smoking/Non-Smoking
2 Age groups
Give an example where nominal data can have multiple categories
Car Brands: Ford, Dodge, Honda etc
Ice Cream Flavors: Chocolate, Strawberry, Cookies n’ Cream
In healthcare, a number association with a blood test or outcome is still nominal since there is no relationship of magnitude between the test and the number associated with it
What are the attributes for nominal data?
No order or magnitude (no quantitative association)
No consistency of scale or equal distances
Explain Ordinal data and include the two elements
Order, rankable data
There is order or magnitude but NO consistency of scale or equal distances. No way to quantitatively separate pain scale elements
Ex: Pain Scale, Income Brackets, SES
Explain Interval data (2) with include the two elements
Interval data can have negatives
Ratio has an absolute zero (physiological parameters)
Does have order/magnitude and does have consistency of scale
Ex: A1C, Height, LDL, Speed
____ and ____ data are discrete while ____ data is continious
Nominal and Ordinal
Interval/ratio
What are the two measures of central tendency to identify spread or dispersion?
Variance
SD
What type of test is used for normally-distributed data?
Parametric tests
Describe a positively skewed data distribution
Asymmetrically distribution with one tail longer, right sided.
Mean > Median
Describe a negatively skewed data distribution
Asymmetrically distribution with one tail longer, left sided.
Mean < Median
Outliers (do/dont) influence mode
don’t
Outliers (do/dont) influence mean and median
do
If you have skewed data, you must rule out what kind of statistical test?
Interval data
What are the values from 1 SD, 2SD and 3SD?
68%
95%
99.7%
Kurtosis
Measure of the extent to which observations cluster around the mean
Positive Kurtosis
Narrow, tight bell curve with more clustering of mean
Negative Kurtosis
Fat bell curve with less clustering of mean
What are the 4 ways to determine if a data set is normally distributed?
Levene’s Test
Skewiness
Draw and Analyze a graph
Analyze Mean and Median
What are ways to deal with non-normally distributed data
Convert it into a standardized value: Z-Score “log” transferred the data
Drop the stat test to a ordinal test (non-parametric test)
Skewiness
Measure of the asymmetry of a distribution
What descriptive statistical elements are useful for nominal data? ordinal data? interval data?
Nominal: Mode
Ordinal: Mode and Median
Interval: Mode, Median and Mean
Type I Error
Also known as Alpha
Rejecting the null hypothesis when it is true (should have accepted it)
Type II Error
Also known as Beta
Accepting the null hypothesis when it is false (shouldn’t have accepted it)
What is power?
The statistical ability of a study to detect a true difference, IF one truly exists between group-comparisons, and therefore the level of accuracy in correctly accepting/not accepting the null hypothesis.
Greater the Power, greater the ____ ____
Sample Size
What are the elements that determine a sample size?
Minimum difference between groups deemed significant
Expected variation of measurement
Alpha (type I) and Beta (type 2) error rates and confidence interval (usually ranges from 90% to 99%).
P-value is associated with what kind of error? Type I or Type II?
Type I
Levene’s Test checks for? What is null hypothesis?
Equality of variance
Null: There is no difference and are equal
If Levene’s Test P-Value is less than .05 then
variance are not equal and do not use t-test
What is the benefit of using confidence intervals over p-value?
Can infer p-value after knowing the variance and spread of data from mean.
For a confidence interval to be statistically significant, the values MUST be
One one side of the ratio of 1. In other words, directional words must both be greater/greater above 1.0 or less than/less than below
What is the relationship between confidence interval and precision?
If you increase the confidence interval, the precision decreases.
What is the purpose of the correlation?
It provides a quantitative measure of the strength and direction of a relationship
What are the first two question we must ask for picking a statistical test?
What kind of data are you collecting?
What kind of comparison/assessment is desired?
What is the correlation test for nominal data?
Contingency coefficient
What is the correlation test for ordinal data?
Spearman Correlation
What is the correlation test for interval data?
Pearson Correlation
What are the key words for correlation?
“Correlate” “Compare”
What is the comparison or assessment type for “event or “Time-to-event”?
Survival Test
What is the nominal survival test?
Log-Rank Test
What is the ordinal survival test?
Cox-Proportional Hazard test
What is the interval survival test?
Kaplan-Meier test
Time with respect to a clock is only pertaining to which type of survival test?
Interval: Kaplan-Meier Test
What is the keyword for survival test?
“Changes over time” “every 2 months over 36 months”
What is the comparison or assessment for “outcome prediction or association”? It uses a prediction by utilizing multiple variables to predict an outcome
Regression
What data can we obtain from a regression test?
Odds Ratio
What is the nominal regression test called?
Logistic Regression
What is the ordinal regression test called?
Multinomial logistic regression
What is the interval regression test called?
Linear Regression
What is the most common type of comparison/assessment?
Frequencies, Counts and Proportions
What two additional questions need to be answered when dealing with frequencies, count and proportions?
How many groups?
Is the data independent or related?
What type of test is best for frequencies, counts and proportions with 2 groups of independent data?
Pearson’s Chi square test
What type of test is best for frequencies, counts and proportions with 3 or more groups of independent data?
Chi-square test of independence
If you have independent data with a expected cell count of less than 5 for frequencies, counts and proportions we should use?
Fischer’s Exact Test
What are buzzwords for related or paired frequencies, counts and proportions tests?
“pre vs post”
“baseline vs end”
“before vs end.”
For statistically significant findings in 3 or more comparisons, one must perform ____ to determine which groups are different. Specifically, ____ to avoid _____.
Post-hoc testing
Bonferroni test of inequality
Risk of increasing type I error
For 2 groups of paired/related frequencies, counts and proportions we would use?
McNemar
For 3 or more groups of paired/related frequencies, counts and proportions we would use?
Cochran