Page 456 - ISC PROCEEDINGS 21.4

The questionnaire used in this study consisted of three main sections. The first
section collected general information about respondents, including demographic
characteristics. The second section included measurement items related to green
marketing, green consumer culture, and green brand awareness. The final section focused
on measuring consumers’ green purchase behavior. All measurement items were
evaluated using a structured questionnaire adapted from previous studies.
In terms of sampling method, the study employed a convenience sampling
approach due to time and resource constraints. The survey questionnaire was distributed
online through social media platforms and online communication channels to reach
potential respondents. In addition, for the items referring to a specific brand, respondents
were asked to select an online cosmetics brand they were familiar with when answering
the questions. This approach allows respondents to provide evaluations based on their
actual perceptions and experiences with the chosen brand. After the data screening
process, a total of 303 valid responses were obtained and used for the subsequent
analysis.
3.3. Data analysis
The dataset was processed using SmartPLS 4, applying the Partial Least Squares
Structural Equation Modeling (PLS-SEM) approach. This technique was chosen as it is well-
suited for prediction-oriented frameworks and exploratory research, especially when the
objective is to investigate complex interrelationships among several constructs.
Furthermore, PLS-SEM is appropriate in contexts involving moderately sized samples and
data that may not follow a normal distribution (Hair et al., 2019). Hence, this approach is
regarded as suitable for assessing both the measurement model and the structural model
in the present study.
4. Result and discussion
4.1. Descriptive statistics
After the data screening process, a total of 303 valid responses were retained for
further analysis. In terms of gender, the sample consists of 118 males (38.9%) and 185
females (61.1%), indicating that female respondents account for a larger share of the
sample. Regarding age, the majority of respondents are between 18-25 years old with 182
participants (60.1%), followed by 26-30 years old with 79 participants (26.1%), 31-40
years
old with 28 participants (9.2%), and over 40 years old with 14 participants (4.6%),
suggesting that the sample is largely composed of young consumers.
4.2. Assessment of the measurement model
To verify the adequacy of the constructs prior to examining structural paths, an
assessment of the measurement model was conducted, with emphasis on establishing
both reliability and validity. This evaluation included key indicators such as factor loadings,
Cronbach’s alpha, composite reliability (CR), and average variance extracted (AVE). As
suggested by Hair et al. (2019), acceptable values require factor loadings above 0.7,
Cronbach’s alpha and CR exceeding 0.7 to demonstrate internal consistency, and AVE
surpassing 0.5 to ensure convergent validity. During the initial estimation, the item GPB4
failed to achieve the required loading level and was consequently excluded from the
construct. After that, the model was recalculated to confirm that the remaining indicators
satisfied the necessary reliability and validity criteria (Table 2).

455