Does framing influence sign up for smoking cessation resources among smokers in Moldova?

The survey aims to generate findings to support digital health solution to help smokers overcome barriers to quit. We will leverage techniques like precommitment, planning for obstacles, and feedback on short-term health improvements when not smoking. To
understand take up of such a service, the survey experiment aims to study what the most effective messaging strategy is. We change the framing of the sign up message one a general one and a reserved spot message for treatment. The survey will be conducted on
Facebook and the treatment will be randomized at the individual level. In the future, we will design and test a chatbot-based solution to support current smokers in their efforts to quit smoking.

2024-04-11

2024-04-30

Our main outcome of interest is if the participants sign up for the chatbot, variable T24 where we vary treatment across groups. We want to understand the effect of framing sign up message that highlights reserving a spot for the chatbot as mentioned in our design. We are also interested in looking at who is likely to sign up for smoking cessation. Q1-Q8 contain demographic information about the participants. T2-T5 focus on current/ past smoking behavior including the frequency of using tobacco products which feeds into likelihood of signing up for cessation. In T7 and T8 we elicit secondary beliefs among individuals regarding the norms about smoking at home and workplace. We also elicit individuals belief on smoking in T20-T22. Understanding existing beliefs will help inform their decision to quit smoking along with intentions in T13 and T14.

We have outcomes on smoking beliefs for past smokers and non smokers. Questions
T9-T12 focus on quitting experience for smokers. We will use this to make inference
about effectiveness of methods that are currently available

For the smokers, we want to create a chatbox that helps then quit smoking. They will
be exposed to two different types of framing in addition to the standard message

• Control: Standard messaging-> With the Ministry of Health, we are creating
an evidence-based chatbot to help people stop using tobacco/smoking products.
Would you like to be notified when we launch the chatbot?

• T1 : seat reserved framing -> With the Ministry of Health, we are creating an
evidence-based chatbot to help people stop using tobacco/smoking products. We
have reserved a spot for you! Don’t miss this chance. Would you like to be notified
when we launch the chatbot?

Not available

Each participant will be allocated a unique and randomly generated user identification
number and subsequently divided into two groups, corresponding to the total number
of experimental arms. The assignment of respondents to experimental groups will be
randomly assigned by a random number generator algorithm. This method ensures a
systematic and unbiased distribution of participants across the experimental conditions,
facilitating the integrity and validity of the study’s comparative analysis. This approach
will ensure that respondents are balanced across treatment arm.

Individual

No

N/A

We divide the sample evenly to include 870 respondents per experimental arm, would be requisite to achieve the desired power level. Total of 1740 participants

870 Control 870 Treatment 1

We gathered empirical evidence from 28 distinct message framing experiments conducted across 23 countries over the last three years, facilitated through social media platforms. The determination of power was executed utilizing a one-way analysis of variance (ANOVA) power analysis, employing the ’power oneway’ command within the STATA software suite. Assuming an observed average within-group variance of 0.25, the calculation included the average outcomes of the control group and the two treatment groups that exhibited the closest means to the control. It was as 0.40 for the control group and 0.46 for the treatment group. An anticipated power level of 0.80 was established for the experiment. Consequently, we determine that a total sample size of 2,192 participants - equally split between the treatment and control arms. Assuming a survey completion rate of 90 percent, we projected that the initial sample should consist of approximately 2,436 smokers commencing the survey. Given the publicly available data indicating a tobacco usage rate of approximately 29 percent in Moldova, the initial survey outreach would encompass around 8,398 respondents to ensure the targeted sample size is met.