You are on page 1of 5

SHORT MANUAL

Two SPSS programs for interpreting multiple regression results

Urbano Lorenzo-Seva Pere J. Ferrando Eliseo Chico

CRAMC (Research Center for Behaviour Assessment), Rovira i Virgili University, Tarragona, Spain

August 2009

How do I compute a regression analysis when I only have the correlation matrix?
In this case, you have to use "mimr_cor.sps", and customize it to your own data. You must follow these steps:

Step 1
You must start SPSS program.

Step 2
You must open "mimr_cor.sps" as a syntax file.

Step 3
You must type your own correlation matrix: you must update the lines in "mimr_cor.sps" where the correlation matrix is defined. Please, remember that the first variable in the correlation matrix must be the criterion. In "mimr_cor.sps". In our example, the correlation matrix is:
***************** ENTER HERE YOUR CORRELATION MATRIX (Y X1 X2 ...Xk). ***************** First variable: criterion. ***************** The rest of variables: predictors. compute R={ 1.000, .564, -.423, -.166, .440, .420; .564, 1.000, -.356, -.081, .303, .527; -.423, -.356, 1.000, .198, -.241, -.244; -.166, -.081, .198, 1.000, .149, .014; .440, .303, -.241, .149, 1.000, .323; .420, .527, -.244, .014, .323, 1.000 }.

You must change these values for the ones related to your own data set. Please note that values in each row are separated by a coma mark, and that rows are separated by semicolon mark.

Step 4
You must type in "mimr_cor.sps" the label related to the criterion variable. In our example the criterion label is:
************************* ENTER HERE THE NAME OF YOUR DEPENDENT VARIABLE. compute vdname={ 'SWLS' }.

You must change the label for the one related to your own data set. Please note that label must be written between ' marks.

Step 5
You must type in "mimr_cor.sps" the labels related to the predictor variables. In our example the labels are:
*********************** ENTER HERE THE NAMES OF YOUR PREDICTOR VARIABLES. compute pnames={ 'OPTI'; 'PESI'; 'ATTEN'; 'CLARI'; 'REPAIR' }.

You must change labels for the ones related to your own data set. Please note that labels must be written between ' marks, and separated by semicolon marks.

Step 6
You must type in "mimr_cor.sps" the reliabilities related to the predictor variables. In our example, the values are:
*********************** ENTER HERE THE NAMES OF YOUR PREDICTOR *********************** ENTER HERE THE RELIABILITIES OF YOUR PREDICTORS VARIABLES. compute rb_p={ .63; .60; .89; .86; .83 }.

You must change values for the ones related to your own data set. Please note that values must be separated by semicolon marks. In case you do not have these values (or you want to have your regression analysis to be computed without taking into account measurement error), then the reliability of each predictor is set to one. For example, compute rb_p={1;1;1;1;1}.

Step 7
You must decide the degree of detail you want to have in the output. You must update the following lines:
************* ENTER HERE IF YOU WANT TO HAVE A DETAILED OUTPUT. ************* detailed = 0 PRODUCES A SIMPLE OUTPUT. ************* detailed = 1 PRODUCES A DETAILED OUTPUT. compute detailed =0.

Step 8
Even if it is not strictly necessary, we advise you to save the file "mimr_cor.sps" with the updates that you have done at this point. We would also suggest to use a file name related to your own data set.

Step 9
Finally, you must select all the code lines in the "mimr_cor.sps" file (for example, you can use CRTL+A command), and execute the code lines (for example, you can use CTRL+R command). SPSS will open an OUTCOME window with outcomes related to your own dataset.

How do I compute a regression analysis when I have the raw data?


In this case, you have to use "mimr_raw.sps", and to customize it to your own data. You must follow these steps:

Step 1
You must start SPSS program.

Step 2
You must open your own data contained in a SPSS data file, and "mimr_raw.sps" as a syntax file. Please, remember that the first variable in the data set must be the criterion.

Step 3
You must type in "mimr_raw.sps" the reliabilities related to the predictor variables. In our example, the values are:
*********************** ENTER HERE THE NAMES OF YOUR PREDICTOR *********************** ENTER HERE THE RELIABILITIES OF YOUR PREDICTORS VARIABLES. compute rb_p={ .63; .60; .89; .86; .83 }.

You must change values for the ones related to your own data set. Please note that values must be separated by semicolon marks. In case you do not have these values (or you want to have your regression analysis to be computed without taking into account measurement error), then the reliability of each predictor is set to one. For example, compute rb_p={1;1;1;1;1}.

Step 4
You must decide the degree of detail you want to have in the output. You must update the following lines:
************* ENTER HERE IF YOU WANT TO HAVE A DETAILED OUTPUT. ************* detailed = 0 PRODUCES A SIMPLE OUTPUT. ************* detailed = 1 PRODUCES A DETAILED OUTPUT. compute detailed =0.

Step 5
You must decide whether you want to have confidence intervals computed. Please note that the value of k also indicates the number of Bootstrap samples used during the computing. You must update the following lines:
****** ENTER HERE IF YOU WANT TO COMPUTE CONFIDENCE INTERVALS. ****** k = 0 CONFIDENCE INTERVALS ARE NOT COMPUTED. ****** k = 500 CONFIDENCE INTERVALS ARE COMPUTED, AND 500 BOOTSTRAP SAMPLES ARE USED (OTHER VALUES CAN BE USED: FOR EXAMPLE, k = 1000). compute k=500.

If Bootstrap samples are to be used, a minimum of 500 is required.

Step 6
Even if it is not strictly necessary, we advise you to save the file "mimr_raw.sps" with the updates that you have done at this point. We would also suggest to use a file name related to your own data set.

Step 7
Finally, you must select all the code lines in the "mimr_raw.sps" file (for example, you can use CRTL+A command), and execute the code lines (for example, you can use CTRL+R command). SPSS will open an OUTCOME window with outcomes related to your own dataset.

You might also like