The impact of environmental concern on green product purchase behavior

¶ … Environmental Concern on the Purchase of Green Products: An Empirical Study of German Retailers

Respondents

The verification of the conceptual framework and hypotheses was done using questionnaire survey. Primary data was gathered from a sample population of 270 respondents who were residents in Germany and who had had bought electronic goods before. One of the major industries dedicated to decreasing their impact on the environment is the "consumer electronics" industry. The industry has applied several green initiatives that focus on the environmental issues in manufacturing, clean delivery systems, energy efficiency and design.

The survey was done in 2016 between the months of August and September. Researchers collected the data and the interviewees were approached directly and they guaranteed that there was no scale interference to lower the possibility of influenced or biased answers.

The demographic composition of the survey's respondents was as follows:

Age

20-25

5(1.8%)

26-30

86 (31.9%)

31-35

112 (41.5%)

36 and above

67 (24.8%)

Gender

Female

148 (54.8%)

Male

122 (45.2%)

Education

Undergraduate

1 (4%)

Graduate

84 (31.1%)

Post graduate

166 (61.5%)

Doctoral degree

19 (7%)

Occupation

Private services

188 (69.6%)

Business

38 (14.1%)

Government job

21 (7.8%)

Self-employed

23 (8.5%)

Table 1: Demographic composition of the respondents

As the table shows, respondents were included from a wide range of age groups. Most of them, 41%, were aged between 31 and 35. The minority age group was the 20 to 25 age group that accounted for only 1.8% of the sample. Generally, those aged 36 and over accounted for 24.8% while those aged between 26 and 30 accounted for 31.9%.

When considering the education levels of the respondents chosen, those respondents who had finished their masters made up 61.5% of the sample while those who had only an undergraduate degree made up only 4% of the sample. The remainder respondents in the sample had other qualifications that feel between a bachelors and a doctoral degree.

Measurements

The respondents used the Likert scale to evaluate the aspects of self expressive benefits, purchase intention, attitude, environmental knowledge and environmental concern. The scale was from 1 to 5, with 5 being 'strongly agree' and 1 being 'strongly disagree'.

This model gives quantitative values on the validity and reliability of the aspects utilized (highlighted above) in the study. To assess convergent validity, factor loading, composite reliability the formulas presented by cronbach's alpha and average variance extracted were utilized (Ahmad & Thyagaraj, 2015).

On the validated scale, Environmental Concern had a total of seventeen items, Declared Purchase had 13 items while Purchase Intention had 15 items. To evaluate the application of these scales, we used models that were similar to Likert and used a 1 to 5 agreement/disagreement range with 1 representing 'complete disagreement' and 5 representing 'complete agreement'.

For the OTHERS column, participants were also requested to score their overall concerns from the same range of 1 to 5. Data analysis of frequency tests was done using SPSS 15.0 while structural equation modeling was evaluated using smart PLS 2.0 -- M3 and frequency tests (Ringle, Wende & Will, 2005).

SEM was used as the main data analysis method. The model evaluated the cause and effect of the relationships that existed between the various constructs and the hypotheses tests that would follow through the evaluation of coefficients.

The construct of Measure model, Partial Least Square -- Path Modeling (PLS -- PM), was used. On doing a Mardoa PK test to check the balance and similarity with the normal multivariate distribution (Hair et. Al., 2013), the results showed significance (p

The SEM measurement models used for the calculation of SEM were those that had Free Asymptotic Distribution. The models that might be applied include Weighted Least Square (WLS), PLS-PM and Diagonally Weighted Least Square (Hair et al., 2013).

Environmental Concern

Lab

Items

EC_1

There should be punishment for those firms that disrespect or damage the environment.

EC_2

Toxic substances from agricultural inputs cause harm to the environment.

EC_3

I comprehend that organic products cause no harm to the environment.

EC_4

Environmental declarations indicate that manufactures have some level of concern for the state of the environment.

EC_5

My town's pollution concerns me.

EC_6

I get concerned when I spot people polluting the streets.

EC_7

I dump organic waste and inorganic wastes separately at my home.

EC_8

Humanity's survival may be threatened by deforestation.

EC_9

I prefer biking or public transport.

EC_10

By saving energy and water, I feel that I am playing a part in solving the issues we have with natural resources.

EC_11

By buying environmentally-safe products, I feel that I am protecting the environment.

EC_12

The atmosphere is damaged by carbon (IV) oxide emissions.

EC_13

Paper and plastic bags are harmful to the environment.

EC_14

All paper and plastic bags should be recycled.

EC_15

Chemical products like detergents designed for home use may harm the environment

EC_16

I reuse wrappers whenever I can.

Table 2: The Scales utilized in research environmental concern

Purchase Intention

Lab

Items

PI_1

I choose those products that lead to the least pollution whenever possible.

PI_2

I try to not use manufactured products that cause damage to the environment.

PI_3

I purchase food lacking in toxic environmental products.

PI_4

I pay a premium for those foods that lack toxic agricultural chemicals which can lead to environmental damage.

PI_5

Price differences influence my purchase when buying environmentally friendly products.

PI_6

I am willing to pay a premium for environmentally friendly products.

PI_7

I may prefer those products that show the environmental certificates of the manufacturer over those that don't.

PI_8

Before purchase, I verify whether a product is environmentally friendly or not.

PI_9

I have made a decision to purchase concentrated products.

PI_10

I have made a decision to purchase compacted products so as to help lower gas emissions.

PI_11

I have made a decision to purchase products with fewer wrappings so as to reduce impact on the environment.

PI_12

I have made a decision to avoid products whose wrappings are non-biodegradable.

PI_13

I have made a decision to purchase home chemicals such as detergents that are non-biodegradable.

PI_14

I have made a decision to purchase refill products.

PI_15

I have made a decision to buy goods in larger sizes and bundles so as to reduce frequency of purchase.

Table 3: Scales utilized in research purchase intention

Declared Purchase

Lab

Items

DP_1

When purchasing a product, I verify whether the producer respects the environment or not.

DP_2

I usually purchase food lacking in harmful chemicals as I know that by doing so, I am helping conserve the environment.

DP_3

I pay a premium to purchase products that respect the environment.

DP_4

I purchase organic products as they are healthier.

DP_5

I pay a premium for organic products.

DP_6

I purchase goods with environmental certificates as they are environmentally conscious.

DP_7

When making a purchase decision, I choose the product that does the least damage to the environment.

DP_8

I usually purchase concentrated goods as they help in water and energy conservation.

DP_9

I purchase compacted products to reduce emission of gases and also because they are easier to transport.

DP_10

I usually purchase products that have the least number of wrappings.

DP_11

I usually purchase environment-friendly home chemicals.

DP_12

I purchase re-fill products so as to put past wrappings to use.

DP_13

I usually purchase products that have non-traditional packing design to minimize solid waste.

DP_14

I have switched from non-environmentally conscious products to those that respect the environment.

Table 4: Scales utilized in research declared purchase

As was shown in the data analysis, PLS 2.0 M3 software, was used. The model was tested for validity and was also adjusted in accordance to the requirements. Even then, the fact remains that the overall standard deviations, coefficients of variation and averages of the respondents that had been given by the sample population with regards to the correction for discarded items could lead to the overall variability to be low.

The R2 looks at a specific part or section of the variables that explains the constructs and shows the adjustment model's quality. The values 0.25, 0.50 and 0.75 were considered weak, moderate and strong respectively (Hair et al., 2013). The Average Variance Expected (AVE) had to be larger than 0.5 in order to meet the model's convergence. The presence or absence of bias from the respondents was tested using the formulas of Cronbrach's alpha and Composite Reliability.

The communality (f2) was used to evaluate if each of the constructs were "useful" to the adjustment of the model. The values 0.35, 0.15 and 0.02 are deemed large, medium and small respectively and the adjustment model's accuracy was evaluated using the model of Redundancy.

AVE

Composite Reliability

R Square

Cronbach's Alpha

Redundancy

Communality

DP_Ind

0,511349

0,893248

0,408761

0,863878

0,200363

0,373152

DP_Other

0,527591

0,869925

0,323307

0,820418

0,160964

0,327942

EC_Ind

0,55498

0,918076

0,899902

0,436833

0,436833

IP_Ind

0,521184

0,883828

0,514629

0,846564

0,257748

0,350896

IP_Other

0,527084

0,847779

0,129625

0,775651

0,06523

0,292177

Reference Value

>0,50

> 0,70

0.02 as small, 0.13 as medium, and 0.26 as large

> 0,60

Positive

Positive

Table 5: Quality criteria of model adjustments -- SEM specification -- Average Variance Extracted (AVE), Composite Reliability

The validity and reliability of the construct and aspects that were evaluated in the study had been based on the suggestions made by Lacker and Fornel. All the aspects indicated a standardized loading factor that was greater than 0.7 while the appropriate range is considered to be between 0.85 to 0.95. This basically means that there was a good level of similarities and balance in the validity of the chosen variables. Cronbach's α was utilized in evaluating consistency within the relationships of the variables that fell within the range of 0.78 to 0.9. This evaluation also showed similarity and balance between variables. Interestingly all the composite reliability values easily crossed the 0.7 minimum requisite. The AVE ranged from 0.52 to 0.55 and thus it also met the 0.5 minimal required limit. Furthermore, Square Multiple Correlation (SMC) was utilized in ensuring all the items had discriminant validity. Table 5 lists the SMC values that were found.

Finally, since the Average Variance Extracted was higher than the relationship that existed amongst each construct, the validity among the constructs was also validated as a necessity. The relationship existing between the constructs were then compared to the AVE values' square roots using the method dictated in the Fornell-Larker criterion. The result expected is that the AVE's square roots must be higher than the correlation or relationships that currently exist between constructs. Discriminant validity is also used to indicate the extent of the independence between the chosen variables and aspects or formulas used (Hair et al., 2013). This is shown in the table below

DP_Ind

DP_Other

EC_Ind

IP_Ind

IP_Other

DP_Ind

0,715086708

DP_Other

0,638885

0,726354597

EC_Ind

0,430985

0,178868

0,744969798

IP_Ind

0,638183

0,338611

0,717376

0,721930745

IP_Other

0,419288

0,567938

0,360035

0,526907

0,72600551

Table 6: Comparison of the AVE's square roots and the constructs' correlation

Nonetheless, there was still a requirement to complete calculations using the Goodness-of-Fit aspect or construct in order to evaluate the overall quality of the adjustment made. Gof indicator is shown by using the mean taken from the average of AVE and the average r2 (Ringle, Wende & Will, 2005). The value that was gotten was 0.426 and it showed that the model was actually adjusted well. Any value higher than 0.36 is usually considered a good value for the fields of behavioral and social sciences (Hair et. al., 2013).

The researchers then used the constructs and aspects chosen and included all of their indicators to make appropriate adjustments and create a new model which is shown in figure 1. It shows the interviewees' perceptions of how their purchase decisions had an impact on the overall environment, and how they view the behavior of other members of society as an influencer on the buying decisions and environmental conservation. It is important to note that one's intention to buy an environmentally-friendly product is determined by the environmental consciousness and concern of the individual or the society as a whole. This is confirmed when an analysis of the model is done.

Figure 1: Model adjusted in research (adopted from Braga Junior et al., 2015).

It is important to note that all the coefficients that were used as foundations in this study were significant as they were greater than 0.05. The Bootstrap method was used to estimate the significance with the value of N. equaling to 600 and replications being 1000 (Ringle, Wende & Will, 2005).