Celebrity endorsement in promoting pro-environmental behavior

There are three primary add-on interventions and a market tool that are being introduced through out this study. The primary interventions included environmental education (EE), celebrity endorsement (CE), and individual pledge by providing a commitment card for the treated participant with the celebrity's signature showing her commitment. The market tool intervention is to introduce a tax policy imposing on the use of single-use plastic items.

We have three treatment with an “add-on” design regarding environmental education and celebrity endorsement.
The first treatment is to only provide environmental education (EE). The environmental education is delivered in the form of marine environment protection workshops organized by the researchers. In the workshop, participants first watch an informative video that describes the plastic pollution at both global and local level, provide solutions and ask for behavior changes at individual level. The participants then are asked to reflect on their daily consumption and disposal of plastics and how they can improve on it. The environmental education workshop aims to raise awareness and improve knowledge of marine environment issues.
The second treatment is to provide environmental education that is endorsed by a celebrity (CE). The celebrity endorsement treatment is built on top of the environmental education treatment. Students receive the same environmental education via workshops. After the informative video (environmental education video), students watch another video clip featuring a celebrity, who describe her personal perspective towards plastics pollutions and the solutions to the problems. In the video clip consists of three main parts. First, the celebrity appears in a Vietnamese well-known river having serious plastic pollution issues, then talking about the negative environmental impacts of plastic wastes. Second, the celebrity, then, appears in a workshop setting discussing with a group of students on the issues and how one could and should do, i.e., changing consumption habits, to reduce plastic consumption. Third, the celebrity also shares her own daily pro-environmental habits, i.e., reducing plastic consumption activities and waste segregation at home. To attract attention the content of the video is formatted as a story in which the celebrity inspires others to practice pro-environmental behaviours. To minimize confounding factors, the informative video without celebrity endorsement is also voiced by a female with a similar accent to the celebrity’s accent.
In the third treatment, we further add a personal pledge option to the previous celebrity endorsement treatment and explore how a personal pledge affect the effectiveness of celebrity endorsement (PP). In addition to receiving the environmental information endorsed by the celebrity, the participants will receive a poster (or a hard paper card) with the same to-do list/tips in the name of the celebrity asking the participants to join her in the movement and make a commitment to reduce their plastics consumption [to meet a specific target (e.g. “Stopping using plastic straws” or “reuse a plastic bag as least three times” or “Do not use plastic bags on Monday”, etc.)]. The participants sign the card if they want to make the commitment. We ask the participants to keep the poster/card and encourage then to post it in their dormitory rooms.
Besides the three treatment groups, we have a control group (C). Participants in the control group do not received any of the education and commitment treatment. But they are invited to workshops that are not related to MPP, but have a similar length, in order to control for any effect from participating in the workshop itself.
We use a between design in the experiment such that each participant will only be assigned to one of the treatment or control groups.
Besides the direct effect on the participants of the workshops, we are also interested in whether the environmental education and the celebrity endorsement of the message are transmitted through the direct recipients and have a second or spillover effect on those who are in their social circle. Given the context of university dormitory system, we choose the roommates of the workshops participants as our targeted. Roommates not only share the same physical environment which involve waste disposal on a daily basis, but also mostly likely to form a strong social bond because of the intensity of daily interaction. Based on which treatment or control groups our workshop participants are assigned to, their roommates are also divided into three treatment and one control groups accordingly.
The last intervention is an introduction of a fee/ charge of using single-use plastic items. At the end-line survey, we announce that if the participant chooses to use a single-use plastic bag he/she must pay a small amount of cash, i.e., 9 cents. This amount is very small, but it may serve as a gentle reminder of being responsible for using an environmentally harmful item.

2019-11-05

2019-11-22

Outcomes of interest and measures of outcomes
We are interested in the changes of knowledge, attitude and practices (KAP) of students regarding plastics pollution and general environmental issues. These are the direct outcomes of our intervention on plastic pollution, and the changes of KAPs regarding general/other environmental issues as the indirect outcomes. Knowledge on marine plastic pollution (MPP) issues are measured by student surveys. EE is expected to increase knowledge. Change in knowledge is a validity test of our EE intervention. Attitudes toward MPP issues are also measured by student surveys. EE is expected to lead to more positive attitudes toward environmental issues. Practices that help mitigate MPP issues are measured through both self-reported behavior in surveys and behavior observed in the field using field experiments. Changes in behavior is the key outcome because our intervention is aimed to influence.

We introduce an shop experiment to observe the consumption behavior of the participant toward plastic issues. Each participant will be offered a gift voucher of 110,000 VND to trade a fixed set of good items. This includes some packed milk, yogurt boxes, instant noodle packages, some candies, pens and a soft drink. These are among the most commonly consumed by students and the bundle of goods may require some single-plastic items, i.e., bags, straws and spoons.

Choosing celebrity
The candidate celebrity in this study is Hoang Thuy, who is best known for winning cycle 2 of Vietnam's Next Top Model. She is the first runner-up at the Miss Vietnam Universe 2017 pageant. She will represent Vietnam at the Miss Universe 2019 pageant. Our first pre-test with around 245 Vietnamese university students in April 2019 showed that over 65% of respondents know her. In addition, the pretest results also indicated that using celebrity endorsement in promoting pro-environmental behaviors is crucial (see appendix 2). We conducted a second round pre-test with a special focus on students’ perceptions of Hoang Thuy’s suitability as an endorser for plastic pollution campaigns. The results support our choices of celebrity endorser for this study. Over 80% respondents heard about Hoang Thuy (heard_of_HT) and 44% were aware of her engagement in environmental protection movement (HT_cleanup). Sixty percent respondents expressed positive view of her being able to inspire young people, while only 4% holds negative views (HT_inspire). Finally, 71% of respondents were positive about her being a suitable endorser for combating plastic pollution and only one respondent held a negative view on this (HT_plastic). Furthermore, respondents consider celebrity’s knowledge and experience as the most important aspect when judging the suitability of celebrity endorsers. Miss Hoang Thuy’s personal engagement in the effort to reduce plastic pollution fits this criterion perfectly. Based on the two round of pre-tests, we believe that Miss Hoang Thuy is an appropriate choice of celebrity endorser for this study.

We expect to capture the behavior changes as major outcome variables in natural field settings in several ways. First, we observe participants’ consumption of single-use plastic in a natural shop setting. We pay all participants , from both control and treatment groups, in our study with vouchers and they exchange the vouchers with a set of goods at the designated shops in the campus/students village. The set of the goods are things that commonly consumed by students, including a pack of milk, a soda drink, a pen and a container of mint tablet candy. These items are selected after consulting with a group of students. We work with the shops so that the students can exchange this set of goods at their cashier counter. They can decide whether to take plastic straws and use plastic bags at the counter when they finish the exchange. We do not add anything in the exchange process that may remind the students of the content of our study.
This setting provides a natural shopping environment for the students, so we can observe unprompted decisions of the students. We have an assistant at each counter to check the student ids, deliver the goods and record the plastic use for each student. We train the assistants to record the plastics use by marking the received vouchers with obscure symbols so that the students will not notice we are recording their plastic consumption. This setting will be repeated after baseline survey, after the workshops and after the endline survey. The vouchers are framed as compensation for participation in each stage of the study. The vouchers are only valid for five days and they are not transferable. Shops are chosen among the shop systems right at the or in close vicinity of the dormitory buildings to facilitate the exchanges. We expected that majority of the sampled students will come to the shop to exchange the voucher right after the voucher is handled. The key outcome variable is how many single use plastic items (i.e. straws and bags) a student use in each exchange. As well, we can also measure the time difference between the time point when the voucher is handled, and the time point when it is redeemed.
Second, we observe students’ voluntary participation in marine litter clean-up operations. We organize a trash clean-up event with the direct the help of the EfD Vietnam policy outreach and communication team sometime after the endline survey. We contact all the participants and invite them to join in the clean event in the vicinity (beach, lake or river side, a park). We use their signing-up decisions as a measure of real-life willingness to mitigate MPP problem.
Changes in KAPs regarding general environmental issues and PEB are the outcomes for potential spillover effects to PEB in other domains. Similar to measuring KAPs regarding plastic pollution, we measure awareness, knowledge and attitudes regarding other environmental issues with self-reported surveys questions. Beside the survey questions, we also measure pro-environmental behavior with an experiment. Specifically, we use a variation of a dictator game to measure individual’s willingness to contribute to an environmental NGO that does not focus on plastic pollutions. Every participant who finishes all stages of this study is entitled to a certain amount of money (i.e., 120 thousand VND ), as a bonus payment for the completing our study. Before delivering the money, we offer students an option to donate money out of the bonus payment to an environmental NGO. They can choose to donate any amount between 0 to 110 thousand VND and keep the rest to themselves. We will transfer the donations to the environmental NGO under the names of the individual students. We use the amounts donated to the NGO as a measure of the spillover effects on pro-environmental behavioral beyond the plastics pollution.

In order to reduce contamination of treatment and study potential spillover effects of the treatment, our treatment assignment is randomized at dormitory room level. We use all the undergraduate dormitory rooms as our sampling frame. We randomly select rooms for the study and randomly assign selected rooms to the treatment or control groups. We randomly select one student in each room as participants of the workshops, as our research subjects for the study of direct effects and randomly one of their roommates in each room for the study of spillover effects. Given we only pick one student in each room, within our sample, randomization occurs at individual level.

First sampled number of rooms are randomly selected. Then, in each room there is a student is randomly invited to participate in this study.

No

About 1,560 rooms

1,560 students

390 students control, 390 students environmental education (EE), 390 students EE and celebrity endorsement (CE), 390 students EE, CE and individual pledge.

Given that we have a relatively large sample in each treatment arm and we randomize the treatment allocation at individual level, , our experiment is able to detect relatively small effect size. Note that despite we collect panel data for the estimating the direct effect on the workshop participants, we can only test some other hypotheses with post-intervention cross-sectional data. Therefore, we calculate the minimum detectable effect sizes conservatively assuming we only have post-intervention cross-sectional data. For a binary outcome variable, such as use single-use plastics or not, assuming that 90% of the participants in the control group takes use single-use plastics and there and the attrition rate is 10%, we can detect a reduction by 8 percentage points using a two-sided chi-squared test with a statistical power of 0.8 at the significant level of 0.05. For a continuous outcome variable like the amount fo donation to an environmental NGO, we can detect an effect that equals to 21% of the standard deviation of the donations in the control group using a two-sided t-test with a power of 0.8 at the significant level of 0.05. The minimum detectable effect size is much smaller that the effects found in studies that examines the effects of educational video on pro-environmental behavioral (e.g. Giglio et al. 2018).