An analysis of consumer behavior regarding green product purchases in Semarang, Indonesia: The use of SEM-PLS and the AIDA model

2.1 Green awareness

The findings of Bamberg and Möser [40] indicate a substantial correlation between psychological characteristics and pro-environmental behavior. Environmental stewardship is crucial in assessing an individual’s degree of environmental awareness. Pro-environmental activity necessitates moral and social direction and comprehension of origins and process characteristics. Likewise, Zsóka et al. [41] discovered that enhancing awareness via education results in markedly elevated environmental knowledge and behavior levels. The development of awareness immediately affects consumers’ inclination to buy environmentally friendly items, as evidenced by a study conducted [42], which emphasizes that both knowledge and awareness propel green purchasing behavior.

Notwithstanding numerous eco-innovation initiatives, like eco-print and eco-enzymes, Semarang continues to encounter significant obstacles in efficiently minimizing trash. This indicates that merely possessing sustainable alternatives is inadequate. A profound comprehension and consciousness among customers are crucial elements for facilitating significant behavioral transformation. Valenti et al. [43] assert that awareness constitutes the preliminary phase in decision-making processes, consistent with the AIDA model. Dlamini and Mahowa (2024) identified that perceived price and attitude strongly affect consumers’ intentions to purchase green cosmetic items, highlighting the complexity of green behavior.

In the realm of eco-friendly product marketing, cultivating consumer response is essential. The awareness stage is essential since it establishes the groundwork for further engagement, encompassing desire and action. This sequence reflects the stages of the AIDA paradigm, wherein enhanced awareness cultivates desire and culminates in action. Nevertheless, despite an improved understanding of environmental concerns, other studies, like those [44–47] demonstrate that enhanced knowledge does not consistently result in sustainable purchasing behavior. This gap underscores the necessity of synchronizing consumer awareness with implementable measures to attain significant environmental effects.

Hence, the concluded hypothesis that:

H1: Green Awareness expresses a positive influence on Purchase Intention

H2: Green Awareness expresses a positive influence on Purchase Decision

2.2 AIDA model

The government should conduct a more thorough analysis of sales of environmentally friendly items using the AIDA Model. This theory may be used for micro, small, and medium enterprises (MSMEs) since [48] suggests that this model offers advantages in terms of profitability when marketing products and services. The difference between consumer needs will lower AIDA’s conversion rate, making companies consider the AIDA model as a marketing process [49]. In the buying process, the trust of potential customers is used with the AIDA Model as a strategy. Understanding of communication is possessed by the AIDA Model, which has 4 processes such as awareness, interest, desire, and action [19,50].

It was first researched by E. St. Elmo Lewis in 1989 in the same sense [51]. Industry to reduce waste is inseparable from intellectual in terms of marketing, promotion and sales, and advertising. Studies of one of the variables of the AIDA model explain that the level of consumer purchase awareness can be explained by the four 4 s in the AIDA model. Products are made conscious; products are made to attract attention, including the facilities and benefits they get, offers, and profits generated in buying products, until finally, they buy products. This means that intellectual property was obtained in this AIDA Model test [52]. The findings in the AIDA Model by Li and Yu, 2013 are very efficient from the perspective of marketing and advertising to help purchase decisions [53].

Hence, the concluded hypothesis that:

H3: The AIDA Model expresses a positive influence on the purchase Intention

H4: The AIDA Model expresses a positive influence on the purchase Decision

2.3 Green advertising

Advertising influences customer behavior by shaping perceptions, attitudes, and purchasing decisions [54]. As a viable strategy in competitive marketplaces, green marketing emphasizes packaging, processing, and a successful promotion that coincides with ecological ideals [3,55]. Contrary to popular belief, advertising functions not solely to produce profit but also to educate consumers about sustainable products [56]. Nonetheless, demographic variables and satisfaction do not substantially influence the effect of advertising on purchasing decisions. Environmental concerns foster green awareness, although they do not consistently translate into sustainable purchasing behavior, revealing a disparity between intention and action [45,57] Green advertising must cultivate cognitive and emotional engagement with customers [58] and highlight perceived value to bridge this gap and enhance purchase intent [59].

Consumer innovation further influences the impact of green advertising. Innovative consumers exhibit heightened responsiveness to advertisements that emphasize eco-friendly attributes; nevertheless, the efficacy of advertising diminishes when innovation is not prioritized. Valenti et al. [43] assert that mere awareness – aligned with the AIDA model – is inadequate without subsequent actionable measures. [60] emphasize that price and attitude influence green purchasing behavior, stressing the necessity for advertising to harmonize rational and emotional appeals. Ultimately, attaining significant environmental results necessitates synchronizing consumer awareness with new, implementable tactics. Mass media advertising generates awareness, shapes behavior, and promotes sustainable consumption via word-of-mouth endorsements [61].

Hence the concluded hypothesis that:

H5: Green Advertising expresses a positive influence on purchase Intention

H6: Green Advertising expresses a positive influence on purchase Decision

2.4 Purchase intention

Environmental degradation may be mitigated by acquiring environmentally friendly items (Chan, 2001). Purchase intention and behavior facilitate the acquisition of ecologically sustainable items [62]. Green product services are acquired based on the customer’s goal to make environmentally friendly purchases [63,64]. Awareness of environmental concerns is a crucial initial step in shaping attitudes, including purchasing decisions [65]. The AIDA model influences purchasing decisions by shaping the purpose behind them [66]. Customers exhibit a correlation between their decision-making process and loyalty when they desire to purchase based on perceived quality [67].

Consumer green purchasing behavior is influenced by the willingness of customers to buy green items, which is driven by motivating reasons [62,68]. An individual’s cognitive capacity will influence the point at which they cease searching for a product before purchasing without retaining any recollection of the search process [69]. Buy intention influences buy choice favorably, and organizations may enhance consumer purchase decisions by effectively managing factors such as benefit, trust, and brand [70]. Attitude significantly impacts purchase intentions, which in turn positively affect purchasing decisions [71]. Furthermore, Attitude towards behavior is a factor that strongly and directly affects the intention to buy sustainable, environmentally friendly products [4].

Hence, the concluded hypothesis that:

H7: Purchase Intention expresses a positive influence on purchase Decision

2.5 Consumer innovativeness

Consumer innovation refers to customers becoming the first to purchase and use new items, driven by their aims and priorities [72]. Consumer innovativeness is the inclination to acquire and utilize new items faster and more frequently than others, indicating significant personality characteristics [72]. Previous research has established a clear connection between consumer innovation and the desire to acquire new goods. Li et al., added, however, that there is limited understanding of the specific processes that drive this link, particularly in the context of sustainable products. Nevertheless, behavior mediates the connection between innovation and purchase intention.

Consumer innovativeness, driven by functional motives, positively affects behavioral intentions [73]. Moreover, the consumer’s orientation denotes their purchasing intention, which impacts the decision-making process [74]. Individuals’ attitudes towards green advertising impact their purchasing interest in eco-friendly items, which significantly mitigate environmental issues [75]. The impact of advertising on a city’s contribution may be observed through the behavior of its community. The findings of the study conducted by academics [76] indicate that when the market share of a particular sector grows, customers in other categories exhibit a reduced inclination toward adoption. Moreover, the findings [8] suggest that social influences can impact attitudes, but they may not always lead to tangible buying decisions within the circumstances.

Hence, the concluded hypothesis that:

H8: Consumer Innovativeness strengthens the influence between Green Awareness and Purchase Decision

H9: Consumer innovativeness strengthens the influence between Green Advertising and Purchase Decision

3.1 Total population in Semarang City

The research only covered people in Semarang, Central Java. To determine which sub-districts represented Semarang City, a multistage random sampling technique was used to meet criteria such as the largest population. As shown in Table 1, five sub-districts were selected based on their population size: Pedurungan, Tembalang, West Semarang, Banyumanik, and Ngaliyan.

3.2 The rationale for selecting respondents’ profiles using stratified random sampling

The stages involved in implementing a stratified random sampling technique are as follows:

• [77] recommend estimating structural equation models (SEM) with 100–300 samples. This sample selection is expected to provide representative and dependable outcomes, taking into account the intricacy of the utilized model. A total of one hundred and sixty-eight responses that fulfilled the specified criteria were included.

• 835,849 people were male, and 852,284 were female. Systematic sampling identified Semarang City’s five main sub-districts. The research included people from the Pedurungan, Tembalang, West Semarang, Banyumanik, and Ngaliyan sub-districts. Since a stable population was anticipated, 168 samples were evenly distributed across 62 sub-districts. Two samples were randomly assigned to 56 sub-districts, and each of the six selected sub-districts received three samples.

• Online methods offered faster accessibility, while offline surveys included those with limited internet access.

3.3 Method of sample determination

Data collection in this study is a survey method by filling out a questionnaire. The same criteria are used to purposefully sample the village with the largest population. Researchers will select limited responders:

1. Are you familiar with or have you purchased eco-friendly products?

2. Gaining broad knowledge about eco-friendly products. Age limits the validity of responses to 15 years.

Researchers used a Likert scale to quantify factors by giving numbers to responses. The Likert scale measures people’s opinions and impressions of social issues [78]. Quantitative analysis and 5-point scoring are used.

1 = Strongly disagree

2 = Disagree

3 = Ambivalent (uncertain)

4 = Agree

5 = Strongly agree

3.4 Indicator each of the variables

The researchers initially examined the elements of the AIDA technique. Subsequently, they employed the structural equation modeling (SEM) approach to analyze the findings obtained from testing the AIDA method. The AIDA model was treated as the dependent variable in this analysis. All variables underwent testing to determine whether they had direct, indirect, or moderate relationships in the variable relationships. Each variable was observed using the following indicators: Table 3 outlines the indicators for green awareness, while Table 4 presents the indicators and relationships within the AIDA model. Table 5 details the indicators for green advertising, and Table 6 shows the indicators for purchase intention. The comprehensive results, including the variables’ relationships and effects on purchase decisions and consumer innovativeness, are summarized in Tables 7 and 8.

• X1 = Green Awareness (exogen);

• X2 = AIDA Model (exogen);

• X3 = Green Advertising (exogen);

• Y1 = Green Purchase Intention (endogen);

• Y2 = Green Purchase Decision (endogen); and

• M1 = Consumer Innovativeness (moderate).

SEM PLS required reliability and validity indications before internal and external tests. Additionally, the indicator helps interviewers ask reliable questions.

4.1 Respondents profile

A total of 168 respondents were included in the sample. Out of the 168 respondents, the distribution based on the sub-district of origin is as follows: 53 respondents (17.9%) came from Pedurungan, 38 respondents (22.6%) came from Tembalang, 31 respondents (18.5%) came from West Semarang, 39 respondents (23.2%) came from Banyumanik, and 30 respondents (17.5%) came from Ngaliyan. The subsequent information provides the population statistics for each of the top 5 sub-districts in Semarang City, categorized by sub-district. The Tembalang and Pedurungan neighborhoods exhibit notable engagement in environmental awareness initiatives, including the Green Warmindo Program, which has effectively promoted waste segregation in food establishments. The local community, particularly in Tembalang, has a commendable environmental consciousness. Warmindo managers have effectively reduced plastic consumption and continuously implemented waste segregation. Table 9 provides a detailed overview of the respondents’ sub-district distribution.

4.2 Respondents gender

Out of the 168 respondents, there were 53 men, accounting for 31.5% of the total, and 115 females, accounting for 68.5% of the total. The discovery that 68.5% of participants in this study were female aligns with other studies, indicating that women exhibit greater engagement in surveys about environmentally conscious conduct. Studies conducted by Alfia et al. [94], as well as Karp [95], have demonstrated that women tend to exhibit higher levels of care towards environmental matters. Tien and Huang [96] additionally verified that women consistently exhibit more significant concern and engage in more environmentally friendly behavior than men. In contrast, Tobler et al. [97] discovered that gender and environmental concerns impacted individuals’ willingness to pay for eco-friendly products. Table 9 shows the gender distribution among the respondents.

4.3 Characteristics of respondents based on age

Out of the 168 respondents, 55 were under 18 years old, accounting for 32.7% of the total. There were 82 respondents aged 18 to 25 years, making up 48.8% of the total; 16 respondents were between 26 and 31 years old, representing 9.5% of the total; 4 respondents fell in the age range of 32 to 37 years, accounting for 2.4% of the total. 6 respondents were between 38 and 43 years old, making up 3.6% of the total. Lastly, 5 respondents were over 43 years old, representing 3.0% of the total. Straughan and Roberts [98] proposed that younger generations may exhibit greater sensitivity towards environmental issues through digital media, whereas older generations tend to rely on conventional sources of information. Table 9 illustrates the age distribution of the respondents.

4.4 Characteristics of respondents based on education

According to the educational background of 168 respondents, there are 3 junior high schools/MTs, accounting for 1.8% of the total. There are 129 high schools/vocational schools, representing 76.6% of the total. Additionally, there are 30 graduates, making up 17.9% of the total, and 6 post-graduates, accounting for 6% of the whole. This finding supports the results of a study by Lee et al. [99], which showed that individuals with secondary education were more likely to participate in environmental surveys. This may be due to greater exposure to environmental campaigns conducted in secondary education institutions, which, as Tikka et al. [100] noted, are often the primary medium for raising environmental awareness among students.” Table 9 presents the educational background of the respondents.

4.5 Characteristics of respondents based on education

According to the survey conducted with 168 participants, the distribution of occupations was as follows: 126 students (75%), 10 civil servants (6%), 10 lecturers/teachers (6%), 12 private employees (7.1%), 3 homemakers (1.8%), 5 unemployed individuals (3.0%), and 2 individuals in other occupations (1.2%). Table 9 summarizes the respondents’ characteristics, including subdistrict, gender, age, education, and type of work. This discovery aligns with the research conducted by Tamar et al. [101], which similarly shows that college students frequently constitute the largest demographic in such investigations. This implies that educational programs are crucial in increasing young people’s environmental consciousness, as noted by Tamar et al. [101]. They argue that formal and informal education in educational institutions can significantly impact students’ environmental attitudes and behaviors.

4.6 Partial least square analysis

4.6.1 Outer model analysis

Figure 1 depicts the path analysis of the outer model, illustrating the connections between the indicators that measure the latent variables. The figure presents values such as factor loadings, which indicate convergent validity, and correlation coefficients between latent variables, which indicate discriminant validity. Composite reliability, average variance extracted (AVE), and Cronbach’s alpha are utilized to assess the internal consistency and reliability of these metrics. This graphic provides a visual representation of the relationship between these indicators and each of the latent variables in the research model.

Figure 1

Path analysis (Outer model).

4.6.1.1 Convergence validity analysis

The value generated from the loading indicator contains the magnitude of the factor in each construct. The tolerance level in construct loading is 0.5–0.60 which can be used in development.

Table 10 indicates that the Green Awareness construct, as assessed by 13 indicators, had the lowest loading value of 0.725 on the X3 indicator and the maximum loading value of 0.865 on the X5 indicator. The AIDA Model construct, as assessed by 14 indicators, exhibited the lowest loading value of 0.722 on indication X16 and the maximum loading value of 0.867 on indicator X20. The Green Advertising construct, consisting of 12 indicators, had the lowest loading value of 0.721 for the X28 indication and the maximum loading value of 0.841 for the X34 indicator. The Purchase Intention contract had the lowest loading value of 0.734 on the Y6 indicator and the highest loading value of 0.862 on the Y11 indicator. The Purchase Decision contract had the lowest loading value of 0.714 for the Y16 indicator and the highest loading value of 0.860 for the Y18 indicator. The Consumer Innovativeness construct, assessed using 12 indicators, had a loading value of 0.707 for the M5 indication, which was the lowest, and a loading value of 0.790 for the M3 indicator, which was the highest. Among all the indicators within each construct, the indicator loading value exceeds 0.7, indicating its validity as a measure of the construct.

Table 10

Loading indicator (Outer model)

	Consumer innovativeness	Green advertising	Green awareness	Model AIDA	Purchase decision	Purchase intention
M1	0.762
M2	0.739
M3	0.790
M4	0.764
M5	0.707
M6	0.772
M7	0.751
M8	0.774
M9	0.747
M10	0.755
M11	0.733
M12	0.740
X1			0.745
X2			0.815
X3			0.725
X4			0.803
X5			0.865
X6			0.757
X7			0.741
X8			0.856
X9			0.817
X10			0.819
X11			0.861
X12			0.834
X13			0.835
X14				0.773
X15				0.758
X16				0.722
X17				0.772
X18				0.809
X19				0.794
X20				0.867
X21				0.810
X22				0.757
X23				0.758
X24				0.732
X25				0.828
X26				0.740
X27				0.736
X28		0.721
X29		0.772
X30		0.815
X31		0.736
X32		0.764
X33		0.792
X34		0.841
X35		0.822
X36		0.784
X37		0.820
X38		0.778
X39		0.791
Y1						0.749
Y2						0.760
Y3						0.836
Y4						0.815
Y5						0.857
Y6						0.734
Y7						0.814
Y8						0.774
Y9						0.769
Y10						0.786
Y11						0.862
Y12						0.829
Y13						0.767
Y14						0.828
Y15					0.834
Y16					0.714
Y17					0.741
Y18					0.860
Y19					0.806
Y20					0.773
Y21					0.849
Y22					0.841
Y23					0.788
Y24					0.772
Y25					0.733

4.6.1.2 Discriminant validity analysis

Table 11 shows the correlation of the value of the indicator must be higher than the construct of the other so-called indicator to measure the validity of the discriminant meets on the outer model. In the outer discriminant value of this model, each construct has a high variable value compared to the other or it can be said that the discriminant validity is high.

Table 11

Cross loading value

	Consumer innovativeness	Green advertising	Green awareness	Model AIDA	Purchase decision	Purchase intention
M1	0.762	0.536	0.561	0.535	0.602	0.624
M2	0.739	0.441	0.478	0.440	0.529	0.536
M3	0.790	0.562	0.570	0.580	0.640	0.638
M4	0.764	0.366	0.367	0.379	0.446	0.437
M5	0.707	0.262	0.315	0.298	0.346	0.337
M6	0.772	0.531	0.526	0.502	0.534	0.572
M7	0.751	0.344	0.394	0.384	0.443	0.437
M8	0.774	0.358	0.313	0.358	0.427	0.419
M9	0.747	0.337	0.293	0.353	0.374	0.381
M10	0.755	0.244	0.240	0.295	0.337	0.329
M11	0.733	0.401	0.384	0.396	0.421	0.408
M12	0.740	0.422	0.412	0.393	0.466	0.473
X1	0.375	0.631	0.745	0.648	0.632	0.653
X2	0.526	0.663	0.815	0.726	0.742	0.723
X3	0.530	0.548	0.725	0.639	0.676	0.621
X4	0.401	0.669	0.803	0.718	0.702	0.719
X5	0.507	0.732	0.865	0.778	0.759	0.763
X6	0.439	0.638	0.757	0.632	0.615	0.650
X7	0.440	0.622	0.741	0.667	0.669	0.699
X8	0.412	0.725	0.856	0.746	0.722	0.713
X9	0.395	0.671	0.817	0.730	0.711	0.699
X10	0.491	0.756	0.819	0.765	0.755	0.746
X11	0.487	0.747	0.861	0.770	0.751	0.716
X12	0.447	0.764	0.834	0.745	0.746	0.706
X13	0.442	0.708	0.835	0.724	0.730	0.709
X14	0.410	0.693	0.771	0.773	0.720	0.712
X15	0.375	0.645	0.624	0.758	0.662	0.628
X16	0.388	0.639	0.617	0.722	0.637	0.643
X17	0.372	0.682	0.693	0.772	0.706	0.686
X18	0.417	0.713	0.703	0.809	0.757	0.720
X19	0.473	0.683	0.671	0.794	0.712	0.742
X20	0.518	0.784	0.790	0.867	0.799	0.823
X21	0.441	0.714	0.766	0.810	0.747	0.727
X22	0.537	0.634	0.717	0.757	0.710	0.655
X23	0.497	0.674	0.662	0.758	0.687	0.665
X24	0.448	0.645	0.616	0.732	0.644	0.625
X25	0.421	0.754	0.716	0.828	0.735	0.735
X26	0.401	0.603	0.636	0.740	0.676	0.640
X27	0.419	0.616	0.636	0.736	0.646	0.627
X28	0.326	0.721	0.523	0.560	0.579	0.514
X29	0.393	0.772	0.585	0.623	0.646	0.609
X30	0.364	0.815	0.603	0.652	0.632	0.646
X31	0.392	0.736	0.629	0.624	0.624	0.614
X32	0.389	0.764	0.576	0.607	0.604	0.610
X33	0.466	0.792	0.708	0.681	0.712	0.689
X34	0.511	0.841	0.728	0.737	0.753	0.757
X35	0.523	0.822	0.720	0.752	0.711	0.745
X36	0.443	0.784	0.734	0.738	0.744	0.746
X37	0.413	0.820	0.776	0.768	0.764	0.756
X38	0.497	0.778	0.681	0.730	0.721	0.715
X39	0.483	0.791	0.683	0.727	0.719	0.738
Y1	0.456	0.659	0.695	0.692	0.722	0.749
Y2	0.557	0.642	0.667	0.677	0.660	0.760
Y3	0.528	0.721	0.702	0.743	0.725	0.836
Y4	0.477	0.733	0.715	0.703	0.695	0.815
Y5	0.477	0.760	0.760	0.778	0.779	0.857
Y6	0.499	0.590	0.589	0.609	0.635	0.734
Y7	0.566	0.689	0.728	0.754	0.744	0.814
Y8	0.440	0.764	0.749	0.727	0.700	0.774
Y9	0.421	0.715	0.677	0.697	0.664	0.769
Y10	0.535	0.685	0.646	0.695	0.707	0.786
Y11	0.517	0.712	0.779	0.769	0.767	0.862
Y12	0.575	0.684	0.683	0.703	0.727	0.829
Y13	0.572	0.674	0.630	0.679	0.709	0.767
Y14	0.587	0.681	0.698	0.701	0.760	0.828
Y15	0.494	0.744	0.746	0.763	0.834	0.741
Y16	0.453	0.637	0.628	0.610	0.714	0.637
Y17	0.478	0.670	0.610	0.694	0.741	0.673
Y18	0.574	0.783	0.799	0.795	0.860	0.809
Y19	0.496	0.663	0.686	0.693	0.806	0.666
Y20	0.510	0.643	0.708	0.702	0.773	0.655
Y21	0.535	0.773	0.814	0.823	0.849	0.825
Y22	0.576	0.748	0.748	0.772	0.841	0.788
Y23	0.454	0.678	0.671	0.723	0.788	0.668
Y24	0.503	0.664	0.650	0.682	0.772	0.677
Y25	0.483	0.602	0.579	0.626	0.733	0.630

The cross loading value in bold shows the highest loading for the indicator on its own construct, compared to the loading on other constructs.

4.6.1.3 Reliability of studies variables

Composite reliability (rho_c): Composite reliability measures the internal consistency of indicators that measure a particular construct. Values above 0.70 are generally considered good, indicating that the indicators consistently measure the intended construct.

Average Variance Extracted (AVE): AVE measures the amount of Variance captured by a construct relative to the total variance, including error variance. A value above 0.50 is considered adequate, indicating that the measured construct can explain more than half of the Variance in the indicator.

Cronbach’s alpha: Cronbach’s alpha is another reliability or internal consistency measure. A value above 0.70 is considered acceptable, while a value above 0.80 is considered good.

All evaluated constructs exhibited Composite Reliability and Cronbach’s Alpha values surpassing 0.90, indicating exceptional internal consistency. Furthermore, all AVE values surpassed 0.50, indicating that the constructs evaluated were capable of elucidating over 50% of the variation in their respective indicators. Hence, the evidence showcased in Table 12 substantiates the soundness and dependability of the constructs employed in this investigation.

Table 12

Reliability of studies variables

Construct	Composite reliability (rho_c)	Average variance extracted (AVE)	Cronbach’s alpha
Consumer innovativeness	940	567	932
Green advertising	951	620	944
Green awareness	960	651	955
AIDA model	955	603	949
Purchase decision	949	630	941
Purchase intention	961	639	956

Diagonal values: The bolded diagonal values represent each build’s roots of the Average Variance Extracted (AVE). A value exceeding 0.50 is required to demonstrate satisfactory construct validity. Table 13 displays the AVE root values. The AIDA model has a score of 0.776. Consumer innovativeness score is 0.753. The Green Advertising score is 0.787. Level of green awareness is 0.807, Purchase Decision is 0.793, and Buying inclination is 0.799. All AVE root values in this table exceed 0.50, indicating that each concept demonstrates strong validity.

The off-diagonal: numbers represent the correlations between distinct constructs. These correlations illustrate the connections between the different components in the model. The correlation coefficient between the AIDA Model and Consumer Innovativeness is 0.563. The association coefficient between Green Advertising and Purchase Intention is 0.869. To establish discriminant validity, it is necessary for the average variance extracted (AVE) value of each construct to exceed the correlation between that construct and other constructs. This table shows that the root AVE value in each diagonal cell surpasses the correlation value found in the associated column and row. Table 13 further substantiates these findings by displaying the specific AVE and correlation values across different constructs, reinforcing the overall discriminant validity of the model.

Table 13

AVE root value and correlation between constructs

	AIDA model	Consumer innovativeness	Green advertising	Green awareness	Purchase decision	Purchase intention
AIDA model	0.776
Consumer innovativeness	0.563	0.753
Green advertising	0.874	0.555	0.787
Green awareness	0.887	0.562	0.848	0.807
Purchase decision	0.907	0.638	0.875	0.880	0.793
Purchase intention	0.889	0.643	0.869	0.870	0.895	0.799

The cross loading value in bold shows the highest loading for the indicator on its own construct, compared to the loading on other constructs.

4.6.1.4 PLS predict

Q²_predict is a valid prediction metric. Values greater than 0 imply that the model possesses strong predictive capability. A greater value indicates a superior predictive ability of the model for the variable. Root Mean Square Error (RMSE) quantifies the average of the squared differences between the anticipated value and the actual value. Smaller numbers indicate the superior prediction ability of the model. As indicated by a Q²_predict value of 0.875, the purchase decision model demonstrates a high level of predictive accuracy for purchasing decisions. The root mean square error (RMSE) value of 0.363 suggests a relatively small prediction error. The purchase intention model, represented by Q²_predict = 0.838, has a high level of prediction accuracy. A root mean square error (RMSE) value of 0.414 suggests a relatively small prediction error. Table 14 demonstrates the metrics for different constructs, emphasizing the strong predictive powers of the model and the small margins of error in the predictions.

Table 14

PLS predict

Construct	Q² predict	RMSE
Purchase decision	0.875	0.363
Purchase intention	0.838	0.414

Construct	Q² predict	PLS-SEM_RMSE	PLS-SEM_MAE	LM_RMSE	LM_MAE
Y1	0.501	0.540	0.399	0.649	0.512
Y2	0.471	0.531	0.450	0.606	0.472
Y3	0.566	0.507	0.410	0.658	0.505
Y4	0.553	0.496	0.373	0.592	0.458
Y5	0.635	0.494	0.394	0.591	0.468
Y6	0.375	0.671	0.55	0.824	0.647
Y7	0.563	0.501	0.399	0.538	0.437
Y8	0.596	0.504	0.393	0.583	0.464
Y9	0.522	0.571	0.432	0.718	0.555
Y10	0.496	0.614	0.482	0.746	0.565
Y11	0.611	0.479	0.373	0.601	0.458
Y12	0.510	0.555	0.447	0.678	0.514
Y13	0.472	0.583	0.481	0.650	0.524
Y14	0.511	0.555	0.452	0.665	0.518
Y15	0.598	0.488	0.370	0.613	0.465
Y16	0.431	0.548	0.436	0.680	0.541
Y17	0.476	0.565	0.473	0.723	0.575
Y18	0.679	0.466	0.365	0.608	0.470
Y19	0.519	0.569	0.460	0.750	0.572
Y20	0.513	0.574	0.445	0.680	0.532
Y21	0.696	0.440	0.352	0.508	0.399
Y22	0.631	0.459	0.362	0.617	0.469
Y23	0.524	0.498	0.372	0.578	0.443
Y24	0.502	0.565	0.463	0.738	0.564
Y25	0.409	0.607	0.508	0.786	0.634

Q ²predict: Most constructs exhibit Q²_predict values above 0.5, indicating a high predictive capability level. As an illustration, Y18 (with a Q²_predict value of 0.679) has exceptional predictive capability.

PLS-SEM_RMSE dan PLS-SEM_MAE: The results for PLS-SEM are often lower than those for LM_RMSE and LM_MAE, suggesting that the PLS-SEM model has superior predictive performance compared to the linear model.

LM_RMSE dan LM_MAE: The results in PLS-SEM tend to be higher, which suggests that the linear model has more prediction errors.

PLS-SEM models typically exhibit superior predictive ability compared to linear models, as seen by lower root mean square error (RMSE) and mean absolute error (MAE) values and higher Q²_predict values. This model demonstrates efficacy in accurately predicting the factors assessed in this study.

4.6.2 Inner model analysis

The evaluation of the structural model involves examining the R ² (R-Square) and f2 (effect size) values, which are fully shown below:

4.6.2.1 R square (R ² )

The test model’s goodness of fit can be assessed by examining the R-square value. The buy intent construct achieved an R-square value of 0.841, indicating that 84.1% of the variation in purchase intent can be explained by green awareness, the AIDA model, and the green advertising construct. By contrast, constructions not included in the study accounted for the remaining 15.9% (100–84.1%). A purchase decision R-square value of 0.892 indicates that 89.2% of the variation in purchasing decisions can be accounted for by factors such as green awareness construction, AIDA models, green advertising, and purchase intent. Table 15 displays the complete R-square values.

Table 15

R-square value

	R-square
Purchase decision	0.892
Purchase intention	0.841

4.6.2.2 F square (F ²)

Table 16 below displays the independent variables’ path coefficients (Consumer et al. Model and Purchase Intention) on the dependent variables (Purchase Decision and Purchase Intention). The route coefficient value indicates the magnitude and direction of the association between these variables. Changes in the value of F-squares can be used to explain the influence of exogenous constructs on endogenous constructs and whether they have a substantive effect. The F-squares assessment criteria are 0.02 small influence, 0.15 medium influence, and 0.35 large. The complete result of the F-squares (effect size) value is presented in the table below.

Table 16

F-square value

	Purchase decision	Purchase intention
Consumer innovativeness	0.040
Green advertising	0.070	0.122
Green awareness	0.045	0.094
AIDA model	0.172	0.141
Purchase intention	0.043

Table 16 shows that the value of F-square (effect size) of the customer innovativeness construct on purchase decisions is 0.040, from the green advertising construct 0.070, green awareness 0.045, AIDA model 0.172, and purchase intention only 0.043. The value of the effect size of the green advertising construct on purchase intention is 0.122, from the green awareness construct, is 0.094, and the AIDA model is 0.141. Only the effect size value from the AIDA model to the purchase decision has a medium/medium influence, while other effect sizes are in the small category. Independent and dependent constructs need to be tested for 5% significance and t-statistics. If the significance level is 5% if the t-statistic value >1.96, the null hypothesis (H0) is rejected. The t-statistical value of the coefficient of influence of the latent construct is obtained from bootstrapping PLS. The results of the bootstrapping PLS Model are presented in the figure below. The AIDA Model is the primary determinant of purchasing decisions. However, Green Advertising and Awareness significantly impact the intention to purchase, as indicated in Table 17 and Figure 2.

Table 17

Value of coefficient (Original sample), standard error and T-statistics

Path analysis	Original Sample (O)	Standard deviation (STDEV)	T Statistics (|O/STDEV|)	p values
Consumer innovativeness → Purchase decision	0.090	0.044	2.034	0.042
Green advertising → Purchase decision	0.257	0.074	3.455	0.001
Green advertising → Purchase intention	0.304	0.066	4.598	0.000
Green awareness → Purchase decision	0.225	0.087	2.604	0.009
Green awareness → Purchase intention	0.280	0.080	3.496	0.000
AIDA model → Purchase decision	0.433	0.079	5.487	0.000
AIDA model → Purchase intention	0.375	0.074	5.074	0.000
Purchase intention → Purchase decision	0.180	0.091	1.972	0.049
Consumer innovativeness × Green awareness → Purchase decision	−0.160	0.069	2.330	0.020
Consumer innovativeness × Green advertising → Purchase decision	0.185	0.074	2.648	0.008

Figure 2

Path analysis (Inner model).

The parameter coefficient’s value can be observed in the original sample value, standard error (standard deviation), and the t-statistics and p-values in Table 17.

4.7 Partial square analysis discussion

4.8 Limitation

One significant constraint of this study is the broad use of the phrase “environmentally friendly products.” Without naming specific items or categories, the findings may fail to include the distinct qualities and customer behaviors associated with various types of environmentally friendly products. Future studies should focus on examining certain eco-friendly items to offer more precise insights and suggestions.

5.1 Conclusion

To address those objectives, we will examine the impact of environmental awareness, the AIDA model, and green advertising on the intentions to purchase and the actual purchasing decisions of environmentally friendly products in Semarang City. Here is the explanation:

1. Hypothesis 1: The results indicate that Green Awareness has a substantial and positive impact on Purchase Intention, with a coefficient of 0.280, a t-statistic of 3.496, and a p-value of 0.000. This underscores the significance of environmental consciousness in influencing buying intentions.

2. Hypothesis 2 states that Green Awareness has a notable and favorable effect on Purchasing Decisions, with a coefficient of 0.225, a t-statistic of 2.604, and a p-value of 0.009. Environmental consciousness significantly influences shopping choices.

3. Hypothesis 3 states that the AIDA model has a substantial and positive impact on Purchase Intention. The coefficient is 0.375, the t-statistic is 5.074, and the p-value is 0.000. This highlights the efficacy of the AIDA approach in enhancing purchase intention.

4. Hypothesis 4: The AIDA model has a statistically significant and positive effect on purchase decisions, with a coefficient of 0.433, a t-statistic of 5.487, and a p-value of 0.000. This demonstrates the efficacy of the AIDA model in influencing purchase decisions.

5. Hypothesis 5 states that Green Advertising has a noteworthy and favorable effect on Purchase Interest, with a coefficient of 0.304, a t-statistic of 4.598, and a p-value of 0.000. This validates the significant impact of green advertising on stimulating purchasing intention.

6. Hypothesis 6 states that Green Advertising has a notable and favorable impact on Purchasing Decisions. The coefficient is 0.257, the t-statistic is 3.455, and the p-value is 0.001. This demonstrates the significant impact of green advertising on purchasing decisions.

7. Hypothesis 7: The AIDA model has a substantial impact on purchasing decisions, with a coefficient of 0.433, a t-statistic of 5.487, and a p-value of 0.000. These findings highlight the significance of the AIDA model in buying choices.

8. Hypothesis 8 states that Consumer Innovativeness has a notable influence on the connection between Green Awareness and Purchase Decisions. The coefficient is −0.160, the t-statistic is 2.330, and the p-value is 0.020. This demonstrates that consumer innovation has an impact on this connection.

9. Hypothesis 9 states that Consumer Innovativeness has a notable and favorable influence on the connections between Green Advertising and Purchase Decisions. The coefficient value is 0.185, and the t-statistic is significant.

5.2 Policy and practice implications of the findings

5.3 Suggestion

1. To enhance education programs: Consumer Innovation influences purchase decisions, but thorough educational campaigns showcasing environmentally friendly product benefits are essential. Forward-thinking buyers may be more likely to buy eco-friendly items if they are aware of their benefits.

2. Engaging in partnerships with influencers and media: Using celebrities in green advertising efforts may change attitudes and increase the appeal of eco-friendly products among customers. Eco-friendly product information may be communicated through reliable media.

3. Development of ecologically sustainable inventive products: Innovation significantly impacts customer choices. Thus, companies should emphasize innovative, sustainable products. These products should appeal to eco-conscious consumers and attract forward-thinking consumers.

4. Segmented marketing strategy: Developing segmented marketing tactics to effectively engage varied customer categories. To attract imaginative consumers, companies may promote environmentally friendly products’ technology and inventive aspects. However, they may highlight the product’s environmental benefits and sustainability to attract eco-conscious buyers.