statistics: 6.3 LINEAR REGRESSION IN JASP

assumption checks for simple linear regression on jasp: linearity? 7 pts

through a scatterplot:

1. open the dataset on jasp

2. click on regression

3. click on classical-correlation

4. select BMI then variables

5. select PA then variables

6. click on scatterplots

7. see the results on the right side and check for a correlation

assumption checks for simple linear regression on jasp: outliers? 7 pts

through a scatterplot:

1. open the data set on jasp

2. click on regression

3. click on classical correlation

4. click on BMI then variables

5. click on PA then variables

6. click on scatterplots

7. check the results on the right side and find any outliers using the max and min standardized residual statistics

what are standardized residual statistics? 2 pts

it is a measure of the strength of the differences between observed and expected results that can be interpreted like z-scores; the min and max residuals should not exceed -3.0 to +3.0 and if they do then there are outliers

assumption checks for simple linear regression on jasp: homoscedasticity?

through a scatterplot that represents the level in the DV and the size of the residual:

1. open the data set on jasp

2. click on regression

3. click on linear regression

4. select BMI then dependent variable

5. select PA then covariate

6. click on plots

7. select residual vs predicted

8. check the results on the right side and see if the data points are approx symmetrical around the line of best fit

assumption checks for simple linear regression on jasp: normality of residuals? 8 pts

through histograms and Q-Q plots:

1. open the data set on jasp

2. click on regression

3. click on linear regression

4. select BMI then dependent variable

5. select PA then covariate

6. click on plots

7. select residuals histogram, standardized residuals, and Q-Q plot standardized residuals

8. check if it looks like a normal distribution

what is Q-Q plot? 1 pt

a plot that shows the quantiles of the residuals against those expected for a normal distribution

how to interpret a Q-Q plot? 3 pts

1. if the residuals are normally distributed the points will be close to the diagonal line

2. if the points concentrate above or below the line there is a problem with kurtosis

3. if the points snal around the line then there is a problem with skewness

assumption checks for simple linear regression on jasp: multicollinearity? 9 pts

through a collinearity diagnostic:

1. open the data set on jasp

2. click on regression

3. click on linear regression

4. select BMI then dependent variable

5. select PA then covariates

6. select variable 3 then covariates

7. click on statistics

8. select collinearity diagnostics

9. check teh results on the right side and see if tolerance is above 0.1 and if VIF is below 10

what is tolerance? 5 pts

it indicates the amount of variance of each IV that is shared with the rest of the IVs

1. tolerance = 1→ perfect non-collinearity (IV shares no variance with other IVs)

2. tolerance </= 0.1→ problems of collinearity (IV shares </= 90% variance with other IVs)

3. tolerance = 0→ perfect collinearity (IV shares all variance with other IVs)

tolerance > 0.1 is considered okay

what is VIF? 3 pts

VIF- variance inflation factor

1. VIF < 10→ low collinearity : non-problematic (IV shares no variance with other IVs)

2. 2. VIF > 10→ high collinearity : problematic estimation of the regression coefficient (IV shares </= 90% variance with other IVs)

how to report assumption checks? 3 pts

1. assumption of linearity- a scatterplot that shows the relations beyween IV and DV was positive/negative and linear

2. assumptionof no outliers- an analysis of standard residuals showed that the data contained no outliers

3. assumptions of normality of errors and homoscedasticity- residual plots showed homoscedasticity and normality of the residuals