**2. Materials and Methods**

The study was conducted using a mixed method iterative design involving the target group in various stages of the project. By doing this, the results of the first two research phases were enriched in a third step by a quantitative online survey. The study was approved by the ethical commission of the faculty for social sciences of the Ludwig-Maximilian University of Munich (GZ 15-01). Each participant provided informed consent prior to participation. Participants were offered shopping vouchers to remunerate them for their time (phase 1 and 2). In phase 3, to increase motivation, respondents could participate in a lottery containing five 100-Euro vouchers.

### *2.1. Phase 1: Semi-Structured Interviews*

In the first qualitative phase of the study, we conducted 26 semi-structured one-to-one interviews lasting between 30 and 60 min exploring existing knowledge and norms of smoking around children as well as perceived barriers to avoiding it [28]. Furthermore, participants were asked to sugges<sup>t</sup> potentially successful key messages and access paths to reach the target group. Given the sensitivity of the topic, one-to-one interviews seemed to be the most appropriate method; providing an empathetic and supportive environment [10]. Male and female adult caregivers of children below the age of six years, who were either smokers themselves or lived in a smokers' household, were eligible. They were also either German, Turkish, Russian or Spanish speakers being unemployed or holding a job with a low job prestige [29]. Recruitment was done at five social institutions offering vocational training, language, integrative or health promotion courses for individuals with low SES, one paediatric practice, the pulmonary service of a children's hospital, and at shopping malls and playgrounds in socially disadvantaged parts of Munich, Germany. Recruitment and interview staff were fluent in German and had either Turkish, Russian or Spanish as their mother tongue so that all interviews could be performed in the participants' preferred language.

### *2.2. Phase 2: Focus Group Discussions*

Based on the results of phase 1, we developed eight key messages, which could be distributed via online social networks. In all messages, a child addressed one health or social consequence of SHS and then offered one potential measure to address the problem. These messages were then visualised by both a comic-style illustrator and a 3D artist. Thereafter, 20 of the 26 participants of the first project phase took part in focus group discussions, of four to six persons, which lasted about 60 min. During the discussions, participants had the chance to give feedback on the design and content of the messages. Based on the focus group discussions, the most promising messages were selected and revised including, among others, the addition of a voiceover.

The interviews, as well as the focus group discussions, were recorded using two audio recorders and transcribed literally, with Turkish, Russian or Spanish transcripts being translated into German. After transcription and translation, we inductively analysed the materials by a stepwise formation of categories [30]. At the same time, analyses were guided by deductive categories that were based on established health behaviour models including evaluation of the perceived threat of SHS for children [31], social norms related to SHS [32] and perceived barriers to avoid SHS [33].

### *2.3. Phase 3: Quantitative Assessment*

Using an online survey (SocSciSurvey; https://www.soscisurvey.de/), key messages and communication channels were evaluated with respect to the fit to the target group, acceptance and general impression. The campaign was offered without audio to ensure that all participants evaluated it in the same way of presentation.

The study population was recruited via snowball sampling. For this, participants in the first two project phases were asked to invite five friends via online social networks to answer the online survey. The number of participants was increased via the social network of the study team and face-to-face recruitment at local playgrounds and in a paediatric practice. Participants could thus answer the questionnaire at home using the link they were given or, in the case of face-to-face recruitment, by direct data entry in a laptop provided by the study team.

The online survey included items on socio-demographics (age, sex, country of birth and educational status) and smoking behaviour. Using five-point Likert scales, participants were asked to assess each of the illustrations and to give an overall evaluation of the campaign. The following aspects were evaluated:

	- ◦ First impression (1 item)
	- ◦ Appeal (1 item)
	- ◦ Quality (5 items)
	- ◦ Suitability for social media (3 items)
	- ◦ Content (3 items)
	- ◦ Effectiveness (4 items)

The exact wording of all items of the online survey is provided in the online Supplement S1.
