Flashcards
Unit 11.B
Cards in this deck (14)
- 1a. In a simple regression equation, perceived social support is predicted by scores on an extraversion scale. The standardized beta is .4. From this information, can you determine R2 for the regression equation, and can you determine r for the relationship between extraversion and social support? If so, what are these values?i. r=standardized beta =.4; R^2 =r^2 = .16
- 2b. In a multiple regression equation, perceived social support is predicted by scores on an extraversion scale and scores on an introversion scale. The standardized betas are .6 for extraversion and .4 for introversion. From this information, can you determine R2 for the regression equation, and can you determine r for the relationship between extraversion and social support? If so, what are these values? Also, from this information, can you determine the correlation between extraversion and introversion, and can you determine what the standardized beta would be for extraversion if introversion were removed from the equation? If so, what are these values?i. cannot determine r or R2 from beta weights.
- 3c. A principal components analysis is conducted on items from a personality scale assessing extraversion and introversion and two, unrotated components are extracted. In a multiple regression equation, perceived social support is predicted by scores on the first component and scores on the second component. The standardized betas are .6 for the first component and .4 for the second component. From this information, can you determine R2 for the regression equation, and can you determine r for the relationship between the first component and social support? If so, what are these values? Also, from this information, can you determine the correlation between the first component and the second component, and can you determine what the standardized beta would be for the first component if the second component were removed from the equation? If so, what are these values?i. r=standardizedbeta
ii. R2 =Σr2 =.62 +.42 =.52
iii. Correlation between components = 0
iv. Betas are unchanged if one predictor is removed - 4a. What would increase the eigenvalue for the first factor?If the factor loadings are larger
If more items (similar to the existing items) are added to the analysis - 5b. What would increase the proportion of variance explained by the first factor?If the factor loadings are larger
- 6c. What would increase the total proportion of variance explained by all the factors?If the factor loadings are larger
If more factors are extracted - 7d. What would increase the average size of the communalities?If the factor loadings are larger
If more factors are extracted - 8e. What would decrease the average size of the residual correlations?If the factor loadings are larger
If more factors are extracted - 9You conduct a principal components analysis on a set of 15 items. In this analysis, the sum of the eigenvalues is equal to 15. Based on this information, answer the following questions.a. What can you say about how many factors must have been extracted?
15
b. What can you say about what the residual correlation matrix looks like?
All the residuals will be zero
c. What can you say about the proportion of variance that is explained?
100% of variance is explained
d. What can you say about the communalities for the items?
They each equal 1 for a total of 15 - 10You conduct a principal components analysis on a set of 12 items. In this analysis, the sum of all the communalities is to 12. Based on this information, answer the following questions.a. What can you say about how many factors must have been extracted?
All 12 factors have been extracted
b. What can you say about what the residual correlation matrix looks like?
All the residuals will be zero
c. What can you say about the proportion of variance that is explained?
100% of the variance is explained
d. What can you say about the sum of the eigenvalues for the factors?
It equals 12 - 11You conduct a principal components analysis on a set of 8 items. In this analysis, all the residual correlations are zero. Based on this information, answer the following questions.a. What can you say about how many factors must have been extracted?
8
b. What can you say about the proportion of variance that is explained?
100%
c. What can you say about the communalities for the items?
Each equals 1
d. What can you say about the sum of the eigenvalues for the factors?
Sum = 8 - 1210. In principal components analysis, what are four ways of deciding how many factors to extract?Theoretical judgement
-Eigenvalues greater than 1
-Scree Plot
-Parallel Analysis - 1311. How do you create and interpret a scree plot?-it is a plot for your eigenvalues for each successive factors
-you take all of your items on the questionnaire and you plot the values as they decrease in eigenvalue as you extract them
-you are looking for a break in the plot, look for where the cliff occurs and extract the factors before the cliff, and reject the ones that are at the bottom, lay a pencil over the line and any factor above you can accept (subjective?) - 14. How do you conduct a parallel analysis?1. Simulate population where all item correlations = 0
2. Draw samples that parallel the observed sample
• same n
• same number of items
3. Conduct EFA on each simulated sample
4. Compute the average eigenvalue
5. Keep a factor if its eigenvalue exceeds the parallel analysis average eigenvalue
Information
- Description
- graduate measurement Flashcards on Unit 11.B, created by abbi gutierrez on 08/04/2021.
- Created by
- abbi gutierrez
- Views
- 0
- Rating