quiz 5

Question 1

heart disease description

Answer 1

• is the leading cause of death in the US
• risks vary by sex, age, and ethnicity
• linked to physical activity levels and diet

Question 2

when do you use a bivariate linear regression?

Answer 2

• when you have 2 variables
• should fit a generally linear relationship

Question 3

bivariate linear regression

Answer 3

• explains the sample by describing a line
• model is built by parameterizing, or calculating m and b

Question 4

what are some ways to model relationships between variables

Answer 4

• measure samples to understand population
• build model to explain relationship for sample
• use model to predict or extrapolate the values of untested individuals

Question 5

what are two linear regression techniques?

Answer 5

• bivariate
• multiple linear regression

Question 6

bivariate

Answer 6

• one predictor
• BMI
• used together to predict variation in a “response”

Question 7

multiple linear regression

Answer 7

• more than one predictor
• Age, BMI, LDL-C

Question 8

what are the 3 questions to ask about a regression model?

Answer 8

• What is the calculated model?
• How much variation in the data does it explain? (R square value)
• does it explain a significant amount of the variation? (F&p-value)

Question 9

more parameters=

Answer 9

model is better at predicting the noise in your data

Question 10

what does each parameter have associated with it?

Answer 10

error

Question 11

Real world dirty approach

Answer 11

includes only significant predictors

Question 12

real world good approach

Answer 12

Akaike Information Criterion (AIC)
- runs from 1 to -1
-1=did something wrong
1= great
0= not as efficient

Question 13

regression

Answer 13

• a model building technique that uses sample to build a mathematical model
• linear regression produces a model 0of a straight line

Question 14

R^2

Answer 14

a measure of model fit
- should use adjusted form if you have more than one predictor

Question 15

why would we want to compare two or more scalar variables?

Answer 15

observational Studies

Question 16

what kind of design is this example: Independent samples t-test comparing acid soil to basic soil

Answer 16

Experimental design

Question 17

what kind of design is this example: Independent Samples t-test comparing contaminated and uncontaminated lake

Answer 17

Experimental designs

Question 18

what kind of design is this example:repeated measures ANOVA comparing baseline, subtraction, and recovery

Answer 18

Experimental designs

Question 19

what kind of design is this example: measure size of single paternity clutches

Answer 19

observational designs

Question 20

what kind of design is this example:measure territory size

Answer 20

observational Designs

Question 21

what kind of design is this example: measure males circulating and bound testosterone

Answer 21

Observational

Question 22

correlation

Answer 22

simplest measure of relationship between two scalar variables.

Question 23

what does correlation measure?

Answer 23

how much variability is shared between two variables.

Question 24

Example of correlation

Answer 24

• height and weight
• you know that they aren’t independent

Question 25

what kind of graph would correlation have?

Answer 25

scatter plot; one variable on each axis

Question 26

how do we quantify shared variability?

Answer 26

pearson's coefficient of correlation, r

Question 27

what is the value range for pearson's coefficient of correlation?

Answer 27

- 1 (perfectly negative) to 1 (perfectly positive)

Question 28

what does a perfect positive slope look like?

Answer 28

increasing slope from left to right

Question 29

what does a perfect negative slope look like?

Answer 29

decreasing slope from left to right

Question 30

what does a pearson's coefficient of correlation of zero slope look like?

Answer 30

straight with no slope or all over the place

Question 31

what are the limitations to pearson's correlation?

Answer 31

- parametric (normal distribution and meets assumption) - Assumes a linear relationship - sensitive to outliers

Question 32

what can outliers do to pearson's correlation?

Answer 32

can weaken a relationship or make it look like there is a relationship that wasn't there

Question 33

spearman's correlation description

Answer 33

- non-parametric - ranks points for each variable and calculates the difference between ranks - if correlated, the ranks for each variable should be similar (tallest person is also the heaviest)