Example:

41199 41219 41284 41338

So we want to create a variable with the 41284 as the value.

Thanks!

Christina:tup: ]]>

I have 6 items (questions) for all my 5 independent variable, IV, total of 36 items, for one dependent variable, DV. Hence my 5 IVs is testing the positive relationship for the one DV.

So I started with reliability analysis to get the cronbach alpha of more than 0.7, which is suggested by my research supervisor. It was tested that all items are good after the pilot study. So I'm all set and fine with some IV and the DV shows 0.6 ~ 0.8.

Next I did the factor analysis, with Varimax rotation and supress small coefficient (absolute value below 0.4). This way I have found the best items for each variable

by looking at the rotated component matrix.

Of course, before I perform my regression analysis, I make sure that I convert all the data into 'scale' since all my items are actually in likert-scale (ordinal). I went 'compute variable' and make sure to only get the average of the items that are 'clumped' together from the result of rotated component matrix.

For instance, if IndependentVariable_A has items M N O P Q R and the >0.4 in the same component has only M N O P, I will only take (M + N + O + P)/4 into conversion of scale, via Transform > Compute Variable

The next step, is that I perform the regression analysis where I put the DV and 5 IV all in scale form and what I obtained is a R of 0.310 and R-squared of 0.096. The unstandardized coefficient for the independent variable are 0.058, 0.077, 0.234, 0.039 and -0.082.

It seems ridiculous and I'm wondering if there's any steps that I missed or did wrongly, because by right, the R value and R-squared shouldn't be that ridiculously low, considering upon reliability analysis and factor analysis, I have did the correct way.

Funnily, even when I opened the sample SPSS file that my lecturer used in teaching the class, everything was considered flawless where his Cronbach and Factor analysis looked perfect, but when it comes to regression analysis, the R-value and R-squared value seemed to be awfully low.

Does anyone know what's going on? ]]>

The Command Syntax reference only states the following:

variance-inflation factors (VIF) displayed in the Coefficients table, and the

eigenvalues of the scaled and uncentered cross-products matrix, condition indexes,

and variance-decomposition proportions displayed in the Collinearity Diagnostics

table.

From the SPSS output in the table with eigenvalues/ condition indices and variance de-composition, I conclude that these statistics, the intercept is taken into account.

However, I am not sure whether the same is true for the VIF. Cohen, Cohen, Aiken and West (2003) state on p. 425 that "VIF, tolerance and the condition number in most statistical packages do not take multicollinerity involving the intercept into account", but they do not provide further information on the specific packages.

Many thanks for your help,

Franziska ]]>