Dear forum,

I am currently working on my thesis. A central part is a regression model with the following variables:

Dependent variable: support for nuclear energy (continuous/scale variable)

Independent variables:

1. Positive (dummy variable which is equal to 1 if respondent has received a positive frame of nuclear energy in his survey);
2. Negative (dummy variable which is equal to 1 if respondent has received a negative frame of nuclear energy in his survey);
3. Dual (dummy variable which is equal to 1 if respondent has received both a positive and a negative frame of nuclear energy in his survey);
4. Positive * level of value endorsement (the level of value endorsement is a continuous variable);
5. Negative * level of value endorsement;
6. Female (1 if respondent is female);
7. Self-placement on political left-right scale (continuous variable).

The problem is that I have detected heteroscedasticity in my regression model. The ZRESID - ZPRED plot has the characteristic cone shape which indicates heteroscedasticity. I want to fix this by using heteroscedasticity robust standard errors.

How do I get SPSS to calculate these for me? I am unfamiliar with the GLM setting of SPSS, and when I do try to use the GLM, I still can't find the option to calculate parameters with robust standard errors. Please help me! Many thanks.

------------------------------
Marnix van Thiel
------------------------------

Dear forum,

I am currently working on my thesis. A central part is a regression model with the following variables:

Dependent variable: support for nuclear energy (continuous/scale variable)

Independent variables:

1. Positive (dummy variable which is equal to 1 if respondent has received a positive frame of nuclear energy in his survey);
2. Negative (dummy variable which is equal to 1 if respondent has received a negative frame of nuclear energy in his survey);
3. Dual (dummy variable which is equal to 1 if respondent has received both a positive and a negative frame of nuclear energy in his survey);
4. Positive * level of value endorsement (the level of value endorsement is a continuous variable);
5. Negative * level of value endorsement;
6. Female (1 if respondent is female);
7. Self-placement on political left-right scale (continuous variable).

The problem is that I have detected heteroscedasticity in my regression model. The ZRESID - ZPRED plot has the characteristic cone shape which indicates heteroscedasticity. I want to fix this by using heteroscedasticity robust standard errors.

How do I get SPSS to calculate these for me? I am unfamiliar with the GLM setting of SPSS, and when I do try to use the GLM, I still can't find the option to calculate parameters with robust standard errors. Please help me! Many thanks.

------------------------------
Marnix van Thiel
------------------------------

Dear forum,

I am currently working on my thesis. A central part is a regression model with the following variables:

Dependent variable: support for nuclear energy (continuous/scale variable)

Independent variables:

1. Positive (dummy variable which is equal to 1 if respondent has received a positive frame of nuclear energy in his survey);
2. Negative (dummy variable which is equal to 1 if respondent has received a negative frame of nuclear energy in his survey);
3. Dual (dummy variable which is equal to 1 if respondent has received both a positive and a negative frame of nuclear energy in his survey);
4. Positive * level of value endorsement (the level of value endorsement is a continuous variable);
5. Negative * level of value endorsement;
6. Female (1 if respondent is female);
7. Self-placement on political left-right scale (continuous variable).

The problem is that I have detected heteroscedasticity in my regression model. The ZRESID - ZPRED plot has the characteristic cone shape which indicates heteroscedasticity. I want to fix this by using heteroscedasticity robust standard errors.

How do I get SPSS to calculate these for me? I am unfamiliar with the GLM setting of SPSS, and when I do try to use the GLM, I still can't find the option to calculate parameters with robust standard errors. Please help me! Many thanks.

------------------------------
Marnix van Thiel
------------------------------

Dear forum,

I am currently working on my thesis. A central part is a regression model with the following variables:

Dependent variable: support for nuclear energy (continuous/scale variable)

Independent variables:

1. Positive (dummy variable which is equal to 1 if respondent has received a positive frame of nuclear energy in his survey);
2. Negative (dummy variable which is equal to 1 if respondent has received a negative frame of nuclear energy in his survey);
3. Dual (dummy variable which is equal to 1 if respondent has received both a positive and a negative frame of nuclear energy in his survey);
4. Positive * level of value endorsement (the level of value endorsement is a continuous variable);
5. Negative * level of value endorsement;
6. Female (1 if respondent is female);
7. Self-placement on political left-right scale (continuous variable).

The problem is that I have detected heteroscedasticity in my regression model. The ZRESID - ZPRED plot has the characteristic cone shape which indicates heteroscedasticity. I want to fix this by using heteroscedasticity robust standard errors.

How do I get SPSS to calculate these for me? I am unfamiliar with the GLM setting of SPSS, and when I do try to use the GLM, I still can't find the option to calculate parameters with robust standard errors. Please help me! Many thanks.

------------------------------
Marnix van Thiel
------------------------------