Am I right in applying ANOVA?

[Pushpender Nath]

Dear Listers.

I am about to finalise my Ph.D thesis. I am testing 29 different hypothesis
in three different industries. Some hypothesis are:


H4 : Perceived market share of a brand is positively related to the brand
equity of services.

H5 : Price level is positively related with the brand equity of services.

H6 : Price policy, different from other players, is negatively related with
the brand equity of services.

As a part of hypothesis testing, other than multiple linear regression, I am
using one way ANOVA with Sceffe's post hoc analysis. Dependent variable in
all the hypotheses is Brand Equity (total score of a 21 item, five points
scale), and independent variables are different in all individual
hypothesis.

e.g. for  (H5: Price level is positively related with the brand equity of
services.), independent variable is price level and dependent variable is
brand equity. Price level is categorized in five groups *viz. *1,2,3,4,and
5; where 1 is lowest price level and 5 is highest price level.

While testing this hypothesis significant results (*p* less than 0.05), in
ANOVA. Descriptive statistics shows an increasing trend in dependent
variable with increasing values of independent variable. When I run post hoc
analysis with Scheffe's technique, I get significant differences in three
combinations of levels of independent variable *viz. *(1 & 2), (1 & 3), (4 &
2).

As an interpretation of the result, I write "*According to one way ANOVA,
Price level increases the brand equity and post hoc analysis indicates
significant differences in brand equity at only three combinations of levels
of independent variable viz. (**1 & 2), (1 & 3), (4 & 2)."*

I request you to suggest me please:
1. Am I doing it in a right manner?
2.Is the application of ANOVA correct in this case? Or should I look for
another hypothesis test?
3. What should I write as interpretation, when ANOVA produces *p*  values
less than 0.05 but no pairwise combination produces significant results in
Scheffe's post hoc test?

Kindly reply. Each word of your suggestion will be highly useful for me.


--
Regards

Pushpender Nath Saini
Univeristy Business School
Panjab Univeristy, Chandigarh
India
Ph.+91-9872819548

[Granaas, Michael]

If you are using ANOVA I think you want your post-hoc testing done utilizing polynomial contrast codes.

If the linear code is significant your results are consistent with your hypothesis.

If quadratic and/or cubic trends are significant you'll have an interesting finding to direct further research.

Michael
****************************************************
Michael Granaas             [hidden email]
Assoc. Prof.                Phone: [phone]
Dept. of Psychology         FAX:  [phone]
University of South Dakota
414 E. Clark St.
Vermillion, SD 57069
*****************************************************




-----Original Message-----
From: SPSSX(r) Discussion on behalf of Pushpender Nath
Sent: Tue 6/12/07 9:15 PM
To: [hidden email]
Subject: Am I right in applying ANOVA?
 
Dear Listers.

I am about to finalise my Ph.D thesis. I am testing 29 different hypothesis
in three different industries. Some hypothesis are:


H4 : Perceived market share of a brand is positively related to the brand
equity of services.

H5 : Price level is positively related with the brand equity of services.

H6 : Price policy, different from other players, is negatively related with
the brand equity of services.

As a part of hypothesis testing, other than multiple linear regression, I am
using one way ANOVA with Sceffe's post hoc analysis. Dependent variable in
all the hypotheses is Brand Equity (total score of a 21 item, five points
scale), and independent variables are different in all individual
hypothesis.

e.g. for  (H5: Price level is positively related with the brand equity of
services.), independent variable is price level and dependent variable is
brand equity. Price level is categorized in five groups *viz. *1,2,3,4,and
5; where 1 is lowest price level and 5 is highest price level.

While testing this hypothesis significant results (*p* less than 0.05), in
ANOVA. Descriptive statistics shows an increasing trend in dependent
variable with increasing values of independent variable. When I run post hoc
analysis with Scheffe's technique, I get significant differences in three
combinations of levels of independent variable *viz. *(1 & 2), (1 & 3), (4 &
2).

As an interpretation of the result, I write "*According to one way ANOVA,
Price level increases the brand equity and post hoc analysis indicates
significant differences in brand equity at only three combinations of levels
of independent variable viz. (**1 & 2), (1 & 3), (4 & 2)."*

I request you to suggest me please:
1. Am I doing it in a right manner?
2.Is the application of ANOVA correct in this case? Or should I look for
another hypothesis test?
3. What should I write as interpretation, when ANOVA produces *p*  values
less than 0.05 but no pairwise combination produces significant results in
Scheffe's post hoc test?

Kindly reply. Each word of your suggestion will be highly useful for me.


--
Regards

Pushpender Nath Saini
Univeristy Business School
Panjab Univeristy, Chandigarh
India
Ph.+91-9872819548