Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
Hierarchical multiple regression
Message posted by Robert on September 17, 1999 at 12:00 AM (ET)
I am a student conducting a research project. I have one dependent variable and six independent variables. I am using SPSS 9.0 and would like information on how to do a hierarchical regression analysis. Any help or instruction is appreciated.
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: Hierarchical multiple regression
Message posted by JG on September 18, 1999 at 12:00 AM (ET)
You are probably looking for step by step regression, where one independent variable is introduced at each step.
Re: Hierarchical multiple regression
Message posted by JG on September 18, 1999 at 12:00 AM (ET)
You are probably looking for step by step regression, where one independent variable is introduced at each step.
Re: Hierarchical multiple regression
Message posted by Bill on September 20, 1999 at 12:00 AM (ET)
The hierarchical approach is one which the one or more independent "variables are entered cumulatively according to some specified hierarchy which is dictated in advance by the purpose and logic of the research." (Cohen and Cohen, 1975) The order of entry is determined by the researcher; this is not a step-wise procedure. DOS versions of SPSS allowed the user to indicate a hierarchical method. The current version (9.0.0) does not explicitly allow for this, but I think that the "Block x of y" entry on the linear regression dialog box using the Method = Enter allows for this approach. That is, Age and Sex might be entered on Block 1 of 1 and an attitudinal variable entered on Block 2 of 2. Maybe Phil or Jack can verify that this is the proper way to do the hierarchical method.
Re: Hierarchical multiple regression
Message posted by Bill on September 20, 1999 at 12:00 AM (ET)
The Method command in SPSS for the instructions I gave previously (if you paste your syntax and look in the syntax window) is
/METHOD=ENTER faca /METHOD=ENTER facb. where faca is Factor A and facb is Factor B (both independent variables).
Upon further reading, this method works if there are not dummy variables in the analysis. If facb were a drug dose (A vs B vs C) that had to be coded into 2 dummy variables (d1 and d2) then the above would become: /METHOD=ENTER faca /METHOD=ENTER d1, d2 .
This would give you the individual effect of d1 and d2 at stage 2, not the effect of dose at stage 2. To solve the problem, replace the method line with:/METHOD=TEST (faca) /METHOD=TEST (facb). for the first example, and with
/METHOD=TEST (faca) /METHOD=TEST (d1, d2). where the dummy variables replace facb. Is seems that the TEST command is not available in the dialog box. Therefore, use the Paste command and edit the syntax in a syntax window as needed.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
