topic Re: Levene's Test In SAS and SPSS in SAS Procedures

Levene's Test In SAS and SPSS

mmjohnson — Thu, 03 Mar 2016 17:41:03 GMT

Hi,

Has anyone encountered this - different Levene's test result for homogeneity of variance between SAS and SPSS?

Below is the results in SAS and after that SPSS. Everything else is the same excep the levene's test. Do you know why that may be?Thank you in advance.

SAS

Levene's Test for Homogeneity of Q4_CodedCount Variance ANOVA of Squared Deviations from Group Means
Source	DF	Sum of Squares	Mean Square	F Value	Pr > F
Condition	2	1.1994	0.5997	1.05	0.3493
Error	1409	802.6	0.5696

Source	DF	Sum of Squares	Mean Square	F Value	Pr > F
Model	2	4.5688012	2.2844006	4.35	0.0131
Error	1409	739.7109438	0.5249900
Corrected Total	1411	744.2797450

SPSS

Test of Homogeneity of Variances
Q4_CodedCount Count of correct benefits recalled
Levene Statistic	df1	df2	Sig.
8.768	2	1409	.000

ANOVA
Q4_CodedCount Count of correct benefits recalled
	Sum of Squares	df	Mean Square	F	Sig.
Between Groups	4.569	2	2.284	4.351	.013
Within Groups	739.711	1409	.525
Total	744.280	1411

Re: Levene's Test In SAS and SPSS

ballardw — Thu, 03 Mar 2016 17:54:37 GMT

Without knowing the options used between the two, i.e. program code, it is difficult to point towards specific possibilities.

The documentation shows that the SAS default in Proc Anova for HOVTEST=Levene defaults to using squared residuals (Type=Square).

I don't know about SPSS but that seems like something easy to investigate whether SPSS defaults to absolute residuals.

Re: Levene's Test In SAS and SPSS

Reeza — Thu, 03 Mar 2016 17:57:43 GMT

Or you could switch SAS to abs to check:

hovtest=levene(type=abs)

Re: Levene's Test In SAS and SPSS

mmjohnson — Thu, 03 Mar 2016 17:59:13 GMT

Thank you both for your comments. That is exactly what is happening, SPSS uses absolute and SAS uses squared by default. Thank you.

Re: Levene's Test In SAS and SPSS

Reeza — Thu, 03 Mar 2016 19:14:28 GMT

@mmjohnson Please mark the question answered 🙂