The Impact of Cause-Focused Messages on the Purchase and Consumption Behavior for Services Supporting Environmental Conservation: A Field Experiment in Yamanashi, Japan

In this study, we will conduct an intervention using a touchscreen vending machine that potential consumers will operate to purchase services supporting environmental conservation. The vending machine includes six different screens: the initial screen, consent screen, service explanation screen, payment interface, usage instructions, and a photo & card printing feature. Our intervention will focus on the service explanation screen.

There are two versions of the service explanation screen: a Product-focused screen and a Cause-focused screen. The Product-focused screen highlights the service details at the top and the major portion of the display using images and text, while the use of the environmental conservation funds is described only with text at the bottom and a minor portion. Conversely, the Cause-focused screen emphasizes the use of the environmental conservation funds prominently at the top with images and text, relegating the service information to a smaller text section at the bottom.

Although the textual content is identical on both screens, they differ in the following three aspects: 1) the visual content displayed (service details or the use of conservation funds), making one more visually compelling; 2) which content occupies the top, prominent position; and 3) the proportion of the screen each content type occupies (dominance).

Based on the type of service explanation screen displayed, potential consumers will be divided into two groups:
Product-focused group (control): those who view the Product-focused screen
Cause-focused group (treatment): those who view the Cause-focused screen

2024-10-26

2024-10-27

1) Dropout rate (or number of dropouts) at the service explanation screen
2) Purchase rate (or number of purchases) of the services

To examine whether the displayed service explanation screen affects the purchasing behavior of potential consumers, we will use the following two outcomes. These metrics are derived from the operational log data of vending machines selling the services.

Dropout Rate = (Total number of logs where the "Return to Top Page" button was touched on the service explanation screen + Total number of logs that automatically returned to the top page after a certain period on the service explanation screen) / Total number of logs that reached the service explanation screen

Purchase Rate = Total number of logs where the payment process was fully completed / Total number of logs that reached the service explanation screen

Note: The reason for not considering the purchase rate as the sole primary outcome is the potential influence of screens following the service explanation screen. After the service explanation screen, the vending machine displays a payment screen with a QR code. This payment screen also includes a "Return to Top Page" button and an automatic return feature after a certain time, which may affect the results. For example, if it takes time to launch the QR payment app, the screen may unintentionally return to the top page after the set period. Since this cannot be solely attributed to the intervention's effect, we also use the dropout rate at the service explanation screen to measure its pure impact. Therefore, the effect of the intervention is primarily examined using the dropout rate from the service explanation screen, with the purchase rate serving to complement the results.

1) Usage log of the dedicated smartphone application
2) Survey results

Secondary outcomes will be analyzed as exploratory research rather than hypothesis-testing research. The reason for this is that the randomization in this experiment applies only to potential buyers, and consumer behavior and evaluation are influenced by selection bias when actual purchases are made.

From the usage logs of the dedicated smartphone application, the following outcomes will be specifically created for exploratory analysis:
- the types of missions accomplished,
- the total number of mission attempts,
- the total number of missions completed, and
- the total points earned by users (points obtained by completing missions, aggregated per user).

Survey responses are mandatory to obtain a commemorative item. From these survey responses, the following outcomes will be specifically created for exploratory analysis:
- the awareness rate of donations to environmental conservation activities
- the extent to which donations influenced purchase motivation
- the percentage of the service fee to be used for environmental conservation activities
- environmental awareness
- satisfaction with the service
- valuation of the service

Settings
In this study, a field experiment will be conducted at Fuefukigawa Fruit Park in Yamanashi Prefecture. The focus of this experiment is a service supporting environmental conservation, which will be offered as part of the attractions at a Halloween event hosted by the park. The development of vending machines and systems used to deliver this service is managed by NEC Solution Innovators, Ltd., and our field experiment will be conducted concurrently with their business validation test, which aims to assess the purchase behavior for these conservation-oriented services.
During the experimental period, visitors will perceive that the service is provided by the park's management (Yamanashi Fruit Park Partners), and thus will remain unaware of NEC Solution Innovators' involvement. As a result, this field experiment falls within the framework of promoting sales methodologies for services supporting environmental conservation sold by non-profit organizations.
The quantity of units available for sale will be capped at 200 over the two-day intervention period due to the limited stock of commemorative items awarded to those who complete the mission. Additionally, contingent on the sales performance on the first day, where sales are 40 units or fewer, the price for the second day may be adjusted as part of the business validation analysis. The determination of any price adjustment will be made solely by the business validation team, thus precluding the involvement of the researchers in this study.


Intervention
For this study, the intervention involves the utilization of a touch-screen vending machine, which potential consumers will operate to purchase the service. This machine will be strategically positioned at the entrance of the "Fruit Plaza" within the park, a high-traffic area frequently visited by many patrons. The vending machine will support QR code payments exclusively, with only one specific QR code payment service available.
The vending machine includes six different screens: the initial screen, consent screen, service explanation screen, payment interface, usage instructions, and a photo & card printing feature. Our intervention will focus on the service explanation screen.
There are two versions of the service explanation screen: a Product-focused screen and a Cause-focused screen. The Product-focused screen highlights the service details at the top and the major portion of the display using images and text, while the use of the environmental conservation funds is described only with text at the bottom and a minor portion. Conversely, the Cause-focused screen emphasizes the use of the environmental conservation funds prominently at the top with images and text, relegating the service information to a smaller text section at the bottom.
Although the textual content is identical on both screens, they differ in the following three aspects: 1) the visual content displayed (service details or the use of conservation funds), making one more visually compelling; 2) which content occupies the top, prominent position; and 3) the proportion of the screen each content type occupies (dominance).
Based on the type of service explanation screen displayed, potential consumers will be divided into two groups:
Product-focused group (control): those who view the Product-focused screen
Cause-focused group (treatment): those who view the Cause-focused screen


Hypothesis
Attribution theory posits that individuals tend to infer intentions and motivations behind human actions and social phenomena. Our study's focus on services supporting environmental conservation is no exception; consumers are likely to speculate whether the primary purpose of selling these services is to boost sales by leveraging environmental conservation or if it genuinely aims to support environmental preservation.
In the context of Cause-Related Marketing, cause-focused and product-focused messages are believed to emphasize different underlying motives of the service. Specifically, cause-focused messages are perceived to enhance the altruistic impression that services are sold for the sake of environmental conservation, while product-focused messages tend to reinforce the self-serving impression that environmental efforts are merely a means to increase sales (Samu and Wymer, 2009; Samu and Wymer, 2014; Fu et al., 2023).
In our study, park administrators are selling services supporting environmental conservation, making it less likely for potential consumers to perceive them as being motivated by profit. Consequently, the influence of product-focused messages is expected to be limited, as consumers are less inclined to believe that environmental conservation is being exploited for sales.
Based on these premises and in alignment with the findings of Fu et al. (2023), we posit the following hypothesis:

Hypothesis: Cause-focused messages will increase the purchase rate of services supporting environmental conservation more than product-focused messages and will reduce the dropout rate at the intervention screen, leading to higher rates of service purchases.


Analysis Plan for Testing Hypothesis
We will use hypothesis testing of population proportions to analyze purchasing behavior. To control for weather conditions and the presence of personnel near the sales area, we will also conduct an analysis using a logit/probit model with these covariates.Additionally, this experiment may not satisfy the Stable Unit Treatment Value Assumption (SUTVA). An example of this is when a group of visitors receives different messages—some seeing the cause-focused message and others seeing the product-focused message—leading to a group discussion that influences their purchasing decision. This situation would violate the no interference assumption.To check for robustness, we plan to conduct an additional analysis excluding payment logs that occurred in rapid succession within a certain interval.


Supplementary Analysis
Based on previous research, cause-focused messages enhance the intention to contribute to the cause. In this study, the cause refers to environmental conservation within the park. The dedicated app, which is part of the service being sold, offers quizzes that provide learning opportunities about the park's environment and its fruits.Therefore, if potential consumers who receive the cause-focused message exhibit an increased intention to contribute to environmental conservation, their usage behavior of the service—such as the number of quiz attempts and responses—may differ from those who receive the product-focused message. This difference may arise since a heightened intention leads to a stronger motivation to learn about the environment.Moreover, differences in service usage may lead to variations in responses to post-usage surveys, influencing environmental awareness and service evaluation between the groups.

To explore whether these possibilities actually occur, we will conduct supplementary analyses from the following perspectives:
1) To compare environmental awareness and service satisfaction between the product-focused and cause-focused groups, we will perform group comparisons. Since this analysis aims to generate hypotheses for future studies rather than test them, we will concentrate on identifying trends rather than conducting formal hypothesis tests.
2) To capture relationships between variables, we will conduct cross-tabulations and correlation analyses.

by comuputer (Specifically, using a computer to generate random numbers and determining the displayed screen based on those results.)

Individual

No

N/A

We are uncertain about the sample size that can be secured for this experiment. The reason for this uncertainty lies in the fact that the field experiment will be conducted alongside a business validation experiment that aims to determine the sales performance of the charity-linked service. At the preparation stage, it is challenging to project the potential sales volume. Based on the findings of previous research, we anticipate that the effect size will not be substantial. Therefore, we consider a sample size of at least 800, which can detect a small effect (Cohen’s h = 0.2), to be ideal. Supplementary Information: The unit of analysis for this experiment is the transaction log that proceeds to the service description screen. In other words, the total number of potential purchasers who view the service description screen (including both actual buyers and non-buyers) constitutes the sample size. Due to the limited quantity of commemorative items, the maximum number of charity-linked services available for sale over two days is 200 units. Consequently, if sales performance is strong, we expect that the sample size will also be influenced by this upper limit.

The sample size for each group is currently undetermined for the same reasons mentioned in the overall sample size section.
Since two types of screens will be displayed randomly, it is anticipated that each group will have approximately half of the total sample size.

Minimum Detectable Effect Size (MDES): Based on our estimated sample size range, the minimum detectable effect sizes (MDES) for α = 0.05 and power = 0.80 are as follows: Please note that all MDES values are expressed in terms of Cohen's h. - Sample size 50 per group (total 100): MDES = 0.56 - Sample size 100 per group (total 200): MDES = 0.40 - Sample size 150 per group (total 300): MDES = 0.32 - Sample size 200 per group (total 400): MDES = 0.28 - Sample size 250 per group (total 500): MDES = 0.25 - Sample size 300 per group (total 600): MDES = 0.23 - Sample size 350 per group (total 700): MDES = 0.21 - Sample size 400 per group (total 800): MDES = 0.20 - Sample size 450 per group (total 900): MDES = 0.19 - Sample size 500 per group (total 1000): MDES = 0.18 These values indicate the smallest effect sizes (in terms of h, or the arcsine transformation of the difference in proportions) that can be detected with a power of 0.80 and a significance level of 0.05 for the given sample sizes. We anticipate that the actual sample size will fall within this range, and therefore the detectable effect size should be within the aforementioned values. According to Cohen's h, an h of 0.2 is considered a small effect and 0.5 a medium effect. Thus, a total sample size of 100 to 200 can detect medium effects, while a total sample size of about 800 can detect small effects. The calculations for these MDES values were performed using R.