amj Australasian Marketing Journal - ANZMAC

Fashion Clothing Involvement
are an optimum linear combination forming the latent
construct (Chin, 1998a,b). Therefore, given the theoretical
formulation and the research context (including sampling),
PLS is particularly suitable as a method for analysis and
model evaluation.
The PLS computer program as implemented by Lohm ller
(1981) and further developed by Chin and Fry (2000) was
used to systematically evaluate the properties of the model as
formulated in Figure 1. The results for the structural model
are shown in Figure 2 which depicts all constructs without the
measured variables for clarity. Evaluation of complex models
can not be made on the basis of any single, general fit index
but rather involves multiple indices which are characterized
Gender
Age
-.021
.091**
.024
R 2 .078*
-.259**
R 2 .009*
Selfmonitoring
.109**
Materialism
.498**
.077**
-.225**
.143**
.127**
54 Australasian Marketing Journal 9 (1), 2001
.031
R 2 .263*
Motives
.692**
by many aspects regarding their quality, sufficiency to
explain the data, congruence with substantive expectations,
precision and confidentiality (Lohm ller, 1989 p. 49).
Hence, a systematic examination of a number of fit indices for
predictive relevance of the model is necessary (Fornell and
Cha, 1994; Lohm eller, 1989). As no distributional assumptions
are made these indices provide evidence for the existence
of the relationships rather than a definitive statistical
tests which may be contrary to the philosophy of soft modeling
(Falk and Miller, 1992).
In the hypothesised model the reflective latent variables were
self-monitoring, materialism, and involvement, while motives
were formulated in the formative sense.
R 2 .708*
Fashion
clothing
involvement
Figure 2: Results for hypothesised model
.924**
.944**
Note: Path coefficients are standarised; R 2 indicates variance explained; ** indicates significant path
Fashion
clothing product
involvement
Fashion
clothing purchase
decision
involvement