Given the following output of a Factor Analysis
Factor Eigenvalue Difference Proportion (%) Cumulative proportion (%)
1 7.938 5.207 46.7 46.7
2 2.731 0.506 16.1 62.8
3 2.225 1.187 13.1 75.9
4 1.038 0.174 6.1 82.0
5 0.864 0.179 5.1 87.1
6 0.685 0.226 4.0 91.1
… …. … ….
17 0.010 - 0.1 100.0
a) How many variables were included in the original dataset? 17 I fattori sono sempre uguali al
numero delle variabili
b) Based on the “eigenvalues greater than 1 rule”, how many factors should be retained and what
is the percentage of total variance explained by them? Vedo eigenvalue column e scelgo quelli
>1
Solutions:
a) 17
b) 4 factors; 82%
Given the following output of a Factor Analysis
Factor Eigenvalue Difference Proportion (%) Cumulative proportion (%)
1 3.080 2.118 38.5 38.5
2 0.962 0.110 12.0 50.5
3 0.853 0.119 10.7 61.2
4 0.734 0.073 9.2 70.4
5 0.660 0.030 8.3 78.7
… …. ….
8 0.497 - 6.2 100
1) Complete the table with the values of the cumulative proportion
2) Explain the reason why the “eigenvalues greater than 1” criterion for choosing the number
of factors is not satisfactory
Solutions:
1) 38.5; 50.5; 61.2; 70.4; 78.6;…;100.0
2) According to that criterion, we should retain just one factor which explains only 38.5 % of
the total variability
Given the following output of a Factor Analysis
Factor Eigenvalue Difference Proportion (%) Cumulative proportion (%)
1 6.822 5.632 48.73 48.73
2 1.190 0.221 57.23
3 0.969 0.228 64.15
4 0.741 0.112 69.44
5 0.629 0.171 73.93
… …. … ….
1) Complete the table with the values of the proportion (%)
2) If we want that the percentage of total variability explained by the factors is at least 70%,
how many factors should we retain?
Solutions:
1) 48.73; 8.50; 6.92; 5.29; 4.49
3) Five factors, since the corresponding cumulative proportion is the first to be greater than 70
When running a Factor Analysis, a communality of 0.88 shows that: communality= parte di
varianza della variabile di ingresso spiegata da tutti i nuovi fattori .
1) the variance of that factor is equal to 0.88
2) the factors explain 88% of the variance of that original variable
3) the covariance between a given original variable and a given factor is 0.88
4) the factors explain 88% of the total original variability
(no. 2 is correct)
When running a Factor Analysis, a factor loading of 0.75 shows that: Factor loading: coovarianza
tra variabile di ingresso e singola e singolo fattore. Communaliy= somma dei factor loadings
riferiti a una variabile di ingress (slide 541)
1) the variance of that factor is equal to 0.75
2) the factors explain 75% of the variance of that original variable
3) the covariance between a given original variable and a given factor is 0.75
4) the factors explain 75% of the total original variability
(no. 3 is correct)
Factor loading sono usati per dare un significato qualitativo ai fattori. Posso calcolarli per ogni variabile e
per ogni fattore vedendo il fattore a quali variabili è correlato (es. vedo che il fattore ha factor loading
alto con variabili: peso, altezza e capelli. Associo a quel fattore il significato di corpo).
Factor rotation: poichè il factor può avere molti factor loadings alti, io tramite la rotation faccio in modo
che ne abbia pochi così da identificarlo meglio ( se il fattore dell’es prima ha loadings alti anche con
titolo di studio e frequenza in palestra io non so come identificarlo…)
In Factor Analysis, the Factor Rotation is necessary for:
1) estimating the factor scores
2) make the communalities more interpretable
3) make the factors more interpretable
4) estimating the factor loadings
(no. 3 is correct)
Factor Analysis is useless when the original variables are
1) strongly correlated
2) almost uncorrelated
3) standardized
4) quantitative
(no. 2 is correct)