Error in Post-hoc tests Univariate ANOVA?
Locked
Error in Post-hoc tests Univariate ANOVA?
 
|
Hello, I performed a univariate ANOVA in SPSS with two independent variables (factors A and B). The interaction effect of A*B was significant. Because I wanted to know which of the cell means of the interaction effect differed significantly, I performed a Bonferoni post-hoc procedure using the following additional syntax lines (not available in menu): /EMMEANS=TABLES(A*B) COMPARE (A) ADJ(BONFERRONI)
When looking at the output and comparing the results with the least significant difference (LSD) procedure, I discovered that the Bonferonni adjustment was only made for the second syntax line (comparing the levels of factor B within the levels of factor A), not for the first line (comparing the levels of factor A within the levels of factor B). Has anyone had any similar experiences? And could this be an error of SPSS or am I missing something? Best regards, Joost van Ginkel ********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. **********************************************************************
|
Re: Error in Post-hoc tests Univariate ANOVA?
|
|
Joost Just a quick thought but it could well be that SPSS will only take the last occurrence of the EMMEANS command – to check this out reverse the order of the
two lines and see if it does only give one set of output but this time is the opposite of what you got before. Best Wishes John S. Lemon Student Liaison Officer Directorate of Information Technology (DIT) - University of Aberdeen Edward Wright Building: Room G51 Tel: +44 1224 273350 Fax: +44 1224 273372 From: SPSSX(r) Discussion [mailto:[hidden email]]
On Behalf Of Ginkel, Joost van Hello, I performed a univariate ANOVA in SPSS with two independent variables (factors A and B). The interaction effect of A*B was significant. Because I wanted to know which of the cell means of the
interaction effect differed significantly, I performed a Bonferoni post-hoc procedure using the following additional syntax lines (not available in menu): /EMMEANS=TABLES(A*B) COMPARE (A) ADJ(BONFERRONI)
When looking at the output and comparing the results with the least significant difference (LSD) procedure, I discovered that the Bonferonni adjustment was only made for the second syntax line
(comparing the levels of factor B within the levels of factor A), not for the first line (comparing the levels of factor A within the levels of factor B). Has anyone had any similar experiences? And could this be an error of SPSS or am I missing something? Best regards,
Joost van Ginkel
********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. ********************************************************************** The University of Aberdeen is a charity registered in Scotland, No SC013683. |
Re: Error in Post-hoc tests Univariate ANOVA?
|
|
Dear John,
I tried that but the result stays the same (the
Bonferroni procedure is then carried out for the first line instead of the
second). I also tried to switch factors A and B in the syntax lines so that it
looks like this:
/EMMEANS=TABLES(B*A) COMPARE (B)
ADJ(Bonferroni)
/EMMEANS=TABLES(B*A) COMPARE (A) ADJ(Bonferroni) but that didn't work either. But thanks
anyway.
Best regards,
Joost van Ginkel
Joost R. Van Ginkel,
PhD From: Lemon, John S. [mailto:[hidden email]] Sent: 16 July 2009 10:20 To: Ginkel, Joost van; [hidden email] Subject: RE: Error in Post-hoc tests Univariate ANOVA? Joost Just
a quick thought but it could well be that SPSS will only take the last
occurrence of the EMMEANS command – to check this out reverse the order of the
two lines and see if it does only give one set of output but this time is the
opposite of what you got before. Best
Wishes John
S. Lemon Student
Liaison Officer Directorate of
Information Technology (DIT) - University of Aberdeen Edward Wright Building:
Room G51 Tel:
+44 1224 273350 Fax:
+44 1224 273372 From: SPSSX(r)
Discussion [mailto:[hidden email]] On Behalf Of Ginkel, Joost
van Hello,
I performed
a univariate ANOVA in SPSS with two independent variables (factors A and B). The
interaction effect of A*B was significant. Because I wanted to know which of the
cell means of the interaction effect differed significantly, I performed a
Bonferoni post-hoc procedure using the following additional syntax lines (not
available in menu):
/EMMEANS=TABLES(A*B) COMPARE (A) ADJ(BONFERRONI) When looking
at the output and comparing the results with the least significant difference
(LSD) procedure, I discovered that the Bonferonni adjustment was only made for
the second syntax line (comparing the levels of factor B within the levels of
factor A), not for the first line (comparing the levels of factor A within the
levels of factor B). Has anyone had any similar experiences? And could this be
an error of SPSS or am I missing something? Best
regards, Joost van
Ginkel ********************************************************************** This email
and any files transmitted with it are confidential and intended
solely for the use of the individual or entity to whom
they are
addressed. If you have received this email in error please
notify the system
manager. ********************************************************************** The University of Aberdeen is a charity registered in Scotland, No SC013683. |
Re: Error in Post-hoc tests Univariate ANOVA?
|
|
In reply to this post by Joost van Ginkel
How many levels does factor A and factor B have? I'd think
a Bonferroni correction wouldn't make sense if a factor had two levels but would
if a factor had three or more levels.
Gene
Maguin
|
Re: Error in Post-hoc tests Univariate ANOVA?
|
|
Dear Gene,
Factor A has two levels and factor B has three. Now that
you mention it, SPSS performs the Bonferroni correction for comparing factor B
within the levels of factor A as if three pairwise tests are performed (by
multiplying p by 3). Thus, if it's done the other way around (comparing the
levels of factor A within each level of factor B), only one pairwise test is
performed within each level of factor B, so a Bonferroni correction would make
no sense. However, when comparing the levels of factor B within the
levels of factor A, a total number of 2*3 = 6 pairwise tests are
performed. Shouldn't p be multiplied by 6 then (and by 3*1 =
3 when levels of factor A are compared within the levels of factor
A)?
Joost van Ginkel
Joost R. Van Ginkel,
PhD From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Gene Maguin Sent: 16 July 2009 15:40 To: [hidden email] Subject: Re: Error in Post-hoc tests Univariate ANOVA? How many levels does factor A and factor B have? I'd think
a Bonferroni correction wouldn't make sense if a factor had two levels but would
if a factor had three or more levels.
Gene
Maguin
********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. **********************************************************************
|
Re: Error in Post-hoc tests Univariate ANOVA?
|
|
Dear
Joost,
Each set of simple effects (A
at level 1 of B, A at level 2 of B, B at level 1 of A, etc.) is treated as its
own family for Type I error correction when doing simple effects tests using
EMMEANS COMPARE in SPSS.
David Nichols From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Ginkel, Joost van Sent: Thursday, July 16, 2009 9:25 AM To: [hidden email] Subject: Re: [SPSSX-L] Error in Post-hoc tests Univariate ANOVA? Dear Gene,
Factor A has two levels and factor B has three. Now that
you mention it, SPSS performs the Bonferroni correction for comparing factor B
within the levels of factor A as if three pairwise tests are performed (by
multiplying p by 3). Thus, if it's done the other way around (comparing the
levels of factor A within each level of factor B), only one pairwise test is
performed within each level of factor B, so a Bonferroni correction would make
no sense. However, when comparing the levels of factor B within the
levels of factor A, a total number of 2*3 = 6 pairwise tests are
performed. Shouldn't p be multiplied by 6 then (and by 3*1 =
3 when levels of factor A are compared within the levels of factor
A)?
Joost van Ginkel
Joost R. Van Ginkel,
PhD From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Gene Maguin Sent: 16 July 2009 15:40 To: [hidden email] Subject: Re: Error in Post-hoc tests Univariate ANOVA? How many levels does factor A and factor B have? I'd think
a Bonferroni correction wouldn't make sense if a factor had two levels but would
if a factor had three or more levels.
Gene
Maguin
********************************************************************** This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please notify the system manager. **********************************************************************
|
| Free forum by Nabble | Edit this page |