Vous êtes sur la page 1sur 2

2009/2010 Exam Paper Question 1 a. What guidelines can you use to determine the number of factors to extract?

Explain each briefly. i. Total variance explained by the factors. Each factor contributes to the variance that explains total variation of original variables in the model. General guidelines is to cover 70-90% of total variation ii. Easy interpretation In addition to total variation, the numbers of factors extracted should also be easy to interpret. This is useful in identifying important interrelationship among measured variables b. How would you interpret a canonical correlation analysis? What are the limitations associated with CCA? We can interpret the results by quantifying the strength of association between the dependent and independent sets of variables. The derived canonical variables show which combination of the original variables best exhibit this association. Limitations: Difficulty in interpreting the results, and it may be useless if there is zero correlation between the dependant variables. c. How do you use the factor loading matrix to interpret the meaning of factors? A factor is easily interpreted by one or more variables with high loadings (near 1) on that particular factor. The remaining variables with small loadings on this factor may be of high loading on another factor. Hence, these remaining variables can explain the interpretation of that factor. d. How and when should you use factor scores in conjunction with other multivariate statistical technique? e. How does a researcher decide the number of clusters to have in solution? One way is by observing the distance between clusters at successive steps. We can stop when the distance exceeds a specified value or when the successive differences in distances between steps make sudden jump f. When would you employ logistic regression rather than discriminant analysis? What are the advantages and disadvantages of this decision? Logistic regression is used we dont know and dont assume the distribution of independent variables. On the other hand, discriminant analysis is used only when we know the distribution of all variables in the model. Advantage of logistic regression: We can use independent variables from any distribution

Disadvantage: Less powerful especially if the distribution of the variables is known. g. When would the researcher use an oblique rotation instead of an orthogonal rotation in factor analysis? What are the basic differences between them Oblique rotation is used to permit greater flexibility in rotated factors. This is done when degree of correlation allowed among factors is, in general, small and these two highly correlated factors are better interpreted as only one factor. In that situation, it will make the interpretation to be simpler. The basic difference between oblique and orthogonal is that in Oblique rotation, the two or more rotated factors are not perpendicular to each other, unlike orthogonal rotation. h. How does the researcher know whether to use hierarchical or nonhierarchical cluster technique? Under which conditions would each approach be used? i. Under what circumstances would you select canonical correlation analysis over multiple regressions as the appropriate statistical technique? If there is more than one dependant variables and the dependant variables are moderately intercorrelated, so it does not make sense to treat them separately What is the difference between cluster analysis and discriminant analysis? Cluster analysis: technique of grouping individuals or objects into unknown groups. The number and characteristics of the groups are to be derived from the date and are not usually known prior to analysis Discriminant analysis: technique of grouping individuals into one of two/more known groups on the basis of set of measurement. It is a parametric method which is based on the knowledge of the distribution of all the variables involved in the models

j.

Vous aimerez peut-être aussi