Test 2 Flashcards
Assess the strength of linear association between variables
Correlation
One is not dependent on the other
equation that best describes linear relationship between variables. Manipulate X to see what Y does
one variable is assumed to vary linearly depending on the level of the other
Linear regression
can be (-1 - 1). the closer to 1 or -1 = a strong linear association. its the magnitude and direction
Correlation Coefficiant
How two variables move together
(-1 to 1) positive or negative
Corrleation
How two variables vary from their means
Also how two variables move together
Covariance
- assumes that each variable follows a normal distribution
- Also follow a bivariate normal distribution ( its a moultivariate distribution)
- both variables are continuous
Pearson’s Correlation Coefficient
- Does not follow a normal distribution
- Alternative Nonparametric measure of correlation
- Assess linear and non-linear association(curves)
usually works off of ranks
Spearman’s
Y is always ______ on x
Dependent
x - independent
y - dependent
The equation of the line is completely dependent on what
The slope and in the y-intercept
What determines the values for the estimates of the slope and the y-intercept? It determines the best fit line. The sum of all the distances is what we are trying to minimize
Least Squares
what measures the expected rate of change in the dependent variable for one unit increase in the independent variable?
The SLOPE
situation in which the variance of the dependent variable is the same for all the data. The variability of BMI is the same for a person that is 5’10’’ and for a person that is 5’5’’.
Homoscedasticity
What does the null say with correlation coefficients?
There is no linear association N=0
The relationship between two variables
One is always dependent on the other
Simple linear regression
The measure of the rate of change in response variable for a unit increase in the independent variable
Slope in linear regression
What helps us estimate the mean value of y at a given value of X?
Population regression line - this is what we want to know so we can estimate it for the entire population
what method uses the equation of the line - y=a+bx - to describe the relationship between two variables
Regression
linear regression
the sampling variability - how the slope will vary from sample to sample is _____
standard error
what two values are used to calculate the std error of the slope
MSE and sum of squares for X
Which of the following statements best describes the slope parameter in a linear regression equation
measures the rate of change in the response variable for a unit increase in the independent variable
The relationship between two variables
One is always dependent on the other
Simple linear regression
Which of the following is used to test for the association between two categorical variables?
a single sample from same population and measure hair and eye color (2 variables) and test how these are associated
– Chi square
What helps us estimate the mean value of y at a given value of X?
Population regression line - this is what we want to know so we can estimate it for the entire population
Which of the following test is most appropriate for categorical data obtained from paired samples?
Before and after characteristics!! - is the proportion different on two occasions?
McNemar’s test
What test is used to measure one variable in two different populations?
depression in Males and Females
Homogeneity
what two values are used to calculate the std error of the slope
MSE and sum of squares for X
Which of the following statements best describes the slope parameter in a linear regression equation
measures the rate of change in the response variable for a unit increase in the independent variable
what is the strength of association; how reliable is the the association?
Degree of association
What is the function of the observed and expected concordance rates? did they respond the same on the same exact survey given two months apart. it the proportion of agreement between the 2 surveys. The reproducibility
Kappa Statistic
What test is used when samples are obtained from two populations, a single categorical variable is measured, and interest is in determining whether the populations are the same across levels of the categorical variable?
Chi square test of homogeneity
Which of the following test is most appropriate for categorical data obtained from paired samples?
McNemar’s test
a 2d cross tabulation of frequencies of occurrence for two categorical variables
Contingency table
what test would test whether the proportion of depression is similar in the two populations
Homogeneity
what is the strength of association; how reliable is the the association?
Degree of association
What is the function of the observed and expected concordance rates? did they respond the same on the same exact survey given two months apart. it the proportion of agreement between the 2 surveys. The reproducibility
Kappa Statistic
a test that is done on data that is ranked and continuous and does not have to be normally distributed
NonParametretic test
a two-independent sample t-test that is a function of (uses) the sample size and the ranks of the observations
Mann-Whitney U Test
A test that uses the + or -. the outcome is measured in matched or paired samples
Before and after data, data is from the same person
Sign test
what test always uses ranked data, is continuous and never normally distributed
Nonparametric test
A test that test whether two populations are equal
2 independent samples
Mann-Whitney U test
what test has a test statistic based on the counts of + and -‘s obtained from the differences in before and after scores
Sign test
What test is when the null is rejected if the median difference is zero
Sign test
a nonparametric test that test more than two independent samples and also test is the population medians are equal
Kruskal-Wallis test
a type of sign test that is measure in matched or paired samples. does not follow a normal distribution, test statistic is W
Wilcoxon signed rank test
When would you reject the null in a Kruskal-Wallis test.
you reject if the evidence shows that there is a difference in the median of the data among the three populations
Most nonparametric method perform analysis using
rank test
What method allows the researcher to remove the effect of the confounding variable? the one variable that might alter the true relationship between two variables (smoking/lung cancer)
confounding = drinking
Cochran-Mantel-Haenszel
what method adjusts for the confounding variable. the researcher computes with the confound variable and without and then compares the two to see if it makes a difference
Cochran-Mantel-Haenszel
y hat =
estimated regression of a line
what is the estimate of the part of the total variability in the response that can be explained by the linear association between X and Y
MSregression
How are the results from a regression analysis usually presented
ANOVA table
In regression, what is referred to as the “residual?”
slide 55 in the corrleation/regression pp
The variance that is not explained
MSE
In a two-factor ANOVA invovling factors A and B and their interaction, which hypothesis should be tested first
Interactin
What does the F-statistic in ANOVA represent
The ratio of between group to within group measures of variability
Which of the following post-hoc tests is best used when the goal is to compare each treatment with a control group
Dunnett’s
In post-hoc analyses, what is the name of the test that adjusts the two-sample t-test by dividing the type I error rate by number of pairwise comparisons.
Bonferroni
In the analysis of variance, the test statistic is based on which of the following distributions (statistical tables)?
F
Which of the multiple comparison methods presented do not adjust (unadjusted) for inflation of the experimentwise error rate
Fisher’s
The hypothesis test used in the analysis of variance involves a comparison of
Means of several groups
A main goal of hypothesis testing is to?
Simultaneously minimize the chance of making either a type 2 or type 1 error
What would we expect the F value to be near when the null hypothesis is true
1
The error rate used to describe the overall Type 1 error rate when conducting post-hoc multiple comparison tests is called
Experimentwise error rate
When visually assessing interaction between two factors in ANOVA, parallel mean profile lines are an indication of significant interaction
FALSE
What statistical test measure the variability of mean between groups and also with the groups
ANOVA (F)
What is the estimated variance - the measure of variability within the groups
MSE
What is the total amount of variability, NOT the average
Sum of squares
what is the measure of proportion of total variation that can be accounted for by the indepedent factor; it ranges from (0 and 1)
R2 - coefficient of determination
what is the same as Bonferroni but has more power and controls the error rate no matter how many times you test the data
Tukey’s
What compares groups means with control group means
Dunnett’s
In ANOVA, when comparing means, what hypothesis test is done first
Interaction
