Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
20 Cards in this Set
- Front
- Back
kurtosis
|
- without changing the mean
- vertical distortion |
|
leptokurtosis
|
- high middle
|
|
platykurtosis
|
- lower middle
- can be flat |
|
negative skew
|
higher data values predominate
|
|
postivie skew
|
lower data values predominate
|
|
statistics
|
- values calculated from sample data used to estimate a value or parameter in a larger population
|
|
point estimate
|
a single statistic calculated from sample observations used to estimate the numerical value of a particular population parameter
|
|
statistical significance testing
|
tells us how much statistical determinations are expected to vary due to chance variations in random sample
|
|
two-tailed test
|
- investigator can't be sure that the population's parameter is greater or less than the null hypothesis
- most common |
|
one-tailed test
|
- used when the direction of the relationship being studed is known and analysis is only concerned with determining the strength of the relationship
- greater statistical power |
|
dependent variables
|
- the study outcome or endpoint
- usual approach: analyze one such variable at a time |
|
independent variable(s)
|
- variable(s) measured to estimate the corresponding measurement of the dependent variable
- defines the conditions under which the dependent variable is to be measured |
|
univariable analysis
|
- 1 dependent variable, no independent variable
- estimation of the probability of a disease without regard to any characteristic such as age, gender, diet |
|
bivariable analysis
|
- 1 dependent variable, 1 independent variable
- determining the if there is an assciation between birth control pill use and stroke |
|
multivariable analysis
|
- 1 dependent variable, >1 independent variable
- determining the likelihood of stroke for femails, of different ages and smoking status - also used to adjust for confoudning variables |
|
continuous data
|
- the range of measurement is any number of equally spaced numerical values between any two points
- measure on ratio or interval scales - a great number of possible values exist - ex. serum cholesterol, age, weight - wide range of values |
|
ordinal data
|
- discrete data
- the interval between consecutive measurements is not necessarily known or constant - ordered sequence - ex. number of pregancies, scoring systems, mammogram interpretation, cancer staging - limited number of possible variables |
|
nominal data
|
- data which is not ordered is measured on a nominal scale
- ex. eye color, gender - the number of nominal variables is the number of categories minus one |
|
rescaling of data
|
- age to age range to young, old description
- shifting down results in the loss of information - statistical power reduced, increasing the likelihood of type II error |
|
Uses of univariable analysis
|
1.) Descriptive studies - clinical and lab results on a group of diseased patients
2.) Descriptive measurements of the study sample 3.) Comparison, using 2 measurements, with the same or paired individuals: a special situation - a hypothesis may be tested in some cases, no independent variable *ex. a blood pressure determination made twice on the same individual, or on two paired similar individuals |