Bivariate Analysis: Correlation/Association

Question 1

What is Correlation?

Answer 1

Describes the degree to which the change in value of one variable corresponds to a change in value of a second variable

Question 2

What is the Pearson Correlation Coefficient?

Answer 2

Measure of the linear relationship between two continuous variables

Question 3

What is the parameter and range for the Pearson Correlation Coefficient?

Answer 3

Rho
-1 to +1

Question 4

0 < r < 1 Indicates what in the Pearson Correlation Coefficient?

Answer 4

Positive correlation, as one variable increases the second variable increases

Question 5

0 > r > 1 Indicates what in the Pearson Correlation Coefficient?

Answer 5

Negative correlation, as one variable increases the second variable decreases

Question 6

r = 0 Indicates what in the Pearson Correlation Coefficient?

Answer 6

No Linear Relationship

Question 7

A positive correlation is seen with what line tilt in a Pearson Correlation Coefficient?

Answer 7

With the higher end of the line pointed towards the positive numbers

Question 8

What type of data is used in Pearson Correlation Coefficient?

Answer 8

Continuous data, a mean must be calculated BUT it does NOT need to be on the same scale

Question 9

What is the hypothesis for Pearson’s Correlation Coefficient?

Answer 9

Ho = p=0
p is rho, the population parameter for r

Question 10

The larger the what, = a larger t-statistic and a greater chance to be significant

Answer 10

1. r
2. N

Question 11

What is the Spearman Rho Coefficient?

Answer 11

Used with ORDINAL/RANKED data, or continuous data that are not normally distributed
-Continuous data can be transformed into ranks then used to calculate Spearman Rs

Question 12

Values of Rs can vary between what for Spearman Rho Coefficient?

Answer 12

-1 and +1

Question 13

Pearson and Spearman are very similar, how so?

Answer 13

1. Both have values that range between -1 and +1
2. Values close to 1, indicate greater correlation; positive or negative signs indicate positive or negative correlation

Question 14

When can Spearman Rs be used?

Answer 14

1. Continuous data is not available
2. Continuous data is skewed

Question 15

What is the Phi Coefficient?

Answer 15

Estimates the extent/strength of the relationship between 2 dichotomous variable

Question 16

What is the range for the Phi Coefficient?

Answer 16

-1 to +1

Question 17

How is the Phi Coefficient related to Chi Square?

Answer 17

Chi Square is equal to the pho coefficient squared

Question 18

If variables are POSITIVELY associated, the data falls along what cells?

Answer 18

DIAGONAL: a and d

Question 19

If variables are NEGATIVELY associated, the data falls along what cells?

Answer 19

NON-DIAGONGAL: b and c

Question 20

What are the ranges of relationship for Phi Coefficient?

Answer 20

1. Weak or NO Relationship: -0.3 to +0.3
2. Moderate Relationship: +/- 0.3 to +/- 0.7
3. Strong Relationship: +/- 0.7 to +/- 1

Question 21

How do you find the test statistic with Chi Square and Phi Coefficient?

Answer 21

X2 test stat = (sample size)(phi coefficient^2)

Question 22

What is the Kappa Coefficient?

Answer 22

Examine inter-rater agreement
Only covers 2x2

Question 23

What is the range for Kappa Coefficient?

Answer 23

0 to 1
No such thing as a negative

Question 24

What does 0 mean in Kappa Coefficient?

Answer 24

No more agreement between two raters as can be expected on the basis of chance aka CHANCE

Question 25

What does 1 mean in Kappa Coefficient?

Answer 25

Perfect agreement, all observations on the diagonal

Question 26

What are the ranges of relationship for Kappa Coefficient?

Answer 26

1. Poor Agreement lower than 0.4
2. Fair to Good Agreement 0.4-0.75
3. Excellent Agreement greater than 0.75

Question 27

Correlation does NOT indicate Causality, it what?

Answer 27

Indicated if the value of one variable increases or decreases while a second variable increases or decreases

Question 28

Variables with casual relation will be correlated but

Answer 28

Variables that are correlated do not have to have a causal relationship