Page 74 - Azerbaijan State University of Economics

THE JOURNAL OF ECONOMIC SCIENCES: THEORY AND PRACTICE, V.78, # 1, 2021, pp. 66-85

RESEARCH INSTRUMENT

The questionnaire was designed to understand customers’ attitudes towards the
mentioned statements. For achieving the goal, participants answered 28 self-
reporting questions that were organized into two sections. Information regarding the
respondent’s age, occupation, gender as well as his/her smartphone model/brand was
included in the first section. Assumptions based on the scales included in the second
section (i.e., Brand awareness, Symbolic Brand Image, Social Influence, Hedonic
Motivation, Price Value, Habit, and Use/actual use).

ANALYTICAL PROCEDURES

A partial least squares-structural estimation technique (PLS-SEM) using SmartPLS
3.3.3 software was applied for measuring the association between variables.
Considering recommendations of Chin (1998) regarding formative and reflective
constructs as well as the suggestions of Venkatesh et al. (2003, 2012), actual use is
considered as a formative construct. The analysis was carried out in three stages.
Reflective measurement was tested in the first and the formative measurement
model in the second. In the third stage, a non-parametric version of partial least
squares-multi-group analysis (PLS-MGA) was applied. PLS-MGA is based on
independent bootstrap analysis for each group (Vinzi, Chin; W., Henseler, & Wang,
2011). For proving the significant difference between groups, p-value should be less
than 0.05 or higher than 0.95 (Henseler J., 2012).

RESULTS AND DISCUSSION
MEASUREMENT MODEL

Cronbach’s alpha (abbr. CA or α) is a popular tool in the case of measuring
reliability and it is accepted to be in a range from 0.6 to 1 where a higher value
expresses a better level of reliability (Malhotra & Birks, 2007, p. 358). Three or
more items needed to fulfill the main requirement for using Cronbach’s alpha
(Pallant, 2016), however, in the current survey most of the variables had only two
items. Therefore, the consistency of the constructs was measured by Spearman’s
correlation coefficient (Table 2).

Construct/ Composite Reliability is usually applied in the case of SEM. Generally,
CR higher than 0.7 is considered as a good level of reliability, while a value higher
than 0.6 is also acceptable (Joseph F. Jr Hair, Black, Babin, & Anderson, 2014, p.
619). The results illustrate that all constructs involved in the study have high
Composite Reliability scores.

74