Lecture 22: Additional concepts to know about

Question 1

Covariance

Answer 1

unstandardized correlation

Question 2

correlation in terms of the covariance

Answer 2

covariance adjusted for standard deviations

Question 3

cor(x, y) formula

Answer 3

cov(x,y)/sd(x)sd(y)

Question 4

cov(x,y) formula

Answer 4

cor(x,y)sd(x)sd(y)

Question 5

Does Covariance depend on units?
Is covariance confined to values -1-1?

Answer 5

It does depend on units and isn’t confined to values between -1-1 because it’s not adjusted for the sd

Question 6

Correlation is more/less useful for describing relationships between variables whilst covariance is used as a _________ ____ in some computations

Answer 6

More useful
Intermediate step

Question 7

Anova stands for

Answer 7

Analysis of variance

Question 8

ANOVA what is it?

Answer 8

a mean comprison test whose test statistic is called an F statistic

Question 9

Anova is a generalization of the _____ ________

Answer 9

t-test

Question 10

The t-test compares __ ______, whereas ANOVA is a ______ ____

Answer 10

2 mens
Omnibus test

Question 11

Omnibus test

Answer 11

A test that can compare many means at once

Question 12

If you’re using an ANOVA to compred 2 means it’s basically equivalent to a __-_____

Answer 12

t-test

Question 13

What is the F statistic

Answer 13

The t statistic squared

Question 14

The resulting p-value is ____ ______ as for the t-test

Answer 14

the same

Question 15

If you’re comprig more than 2 means then ANOVA gives you a single _-_____ testing the omnibus null hypothesis

Answer 15

p-value

Question 16

omnibus null hypothesis

Answer 16

All population means are equal

Question 17

If the p-value in ANOVA is statistically significant it means what?

Answer 17

We’re confident atleast one mean is different from the others but can’t specify which one’s or how many are different

Question 18

It’s good to do “_____ ___ _____” after a omnibus test

Answer 18

post hoc test

Question 19

post hoc tests

Answer 19

t tests of individual pairs of means –> omnibus test pretty pointless

Question 20

Types of ANOVA (4)

Answer 20

1. One way ANOVA
2. Factorial ANOVA
3. MANOVA
4. ANCOVA

Question 21

One way ANOVA is for designs with a ______ _____. It has ________- ________ and ______-__ versions just like the t-test does

Answer 21

Single factor
Between-subjects
Within-subjects

Question 22

Factorial ANOVA is for _______ designs. It has both _______-______ and ______-______ and also a “______” version

Answer 22

Factorial
Between-subjects
Within-subjects
Mixed

Question 23

“Two-way ANOVA” is for designs with _ ______, “three-way ANOVA” is for designs with ___ ______

Answer 23

2 factors
3 factors

Question 24

MANOVA stands for

Answer 24

Multivariate analysis of variance

Question 25

MANOVA is for designs with a _____ factor but ________ dependent variables

Answer 25

Single Multiple

Question 26

ANCOVA stands for

Answer 26

Analysis of covariance

Question 27

ANCOVA is like ANOVA but adjusts for one or more "_______" which are also _______ _________ ______

Answer 27

covariance potential confounding variables

Question 28

Linear regression

Answer 28

a type of correlational analysis in whcih we use one variable to predict another

Question 29

Regression line 2 other names

Answer 29

1. Least squares line 2. Line of best fit

Question 30

Formula for regression line

Answer 30

Y = intercept + slope*X or y=mx + c

Question 31

What can we use the regression line for?

Answer 31

To estimate the average value of Y for given value of X eg. predicting college GPA from SAT score

Question 32

In multiple linear regression, we use ______ predictor variables to predict Y

Answer 32

multiple

Question 33

False discovery rate

Answer 33

The percentage of false positives you allow in a study It's less than or equal to the familywise Type 1 Error rate

Question 34

FDR control is more/less strict thant the familywise Type 1 Error rate so adjustments (of alph levels and p-values) need to be more/ don't need to be as severe

Answer 34

Less Don't need to be as severe

Question 35

FDR is sometimes used when the number of tests in the family is _______ _____ and it's considered tolerable for some proportion of those tests to produce _____ __ _____

Answer 35

extremely large Type 1 errors