I suggest what you need is the approach advocated by Diamantopoulos et al., formative instead of reflective indicators (distinction part of structured equation modeling). See attachment.

First of all, grounded theory approach is needed before you delve into the methodology section. This kind of advancement into the research topic may seem confusing for a while. However, how can you go ahead without generation of themes, before you are actually researching something very fundamental and innovative nature? A "never-studied-before" topic needs innovation at every step. Lest there are loopholes left in the research methodology you select, go ahead to establish a dependable regimen for generating themes. This will be a theory-based pre-task. After that, a grounded theory based inductive approach may also be adopted.

Never-studied-before phenomenon may require qualitative research centering around target respondents. You need to identify dimension(s), if any, including underlying observable variables related to the construct. This can be done, including but not limited to,  by focus group discussion, depth interview or through expert opinion, or a combination of all.

Your research looks very interesting. From what I know about new measurement development I would say one of the first steps could be considering your 'logic of inference' - do you assume the latent variable to be an 'a priori' or 'a posterior' construct. Since you have something in mind 'professionalism' (in this case) on the use of social media' I would imagine that you're using a deductive approach (a priori). This might require locating sub-themes/sub-constructs that might make 'professionalism' from existing literature and hypothesizing that these will significantly represent the 'professionalism' scale you're building and then make questionnaire items accordingly. This approach will require a confirmatory factor analysis (reliability scores as in Cronchbach's Alpha scores) or exploratory factor analysis i.e. PCA (if you are not sure which items would load on which sub-construct).

If you however wished to bring the categories/constructs from the research participants (inductive approach) you might want to look at grounded theory or Q methodology. This approach would be most suitable if a researcher assumed that he/she doesn't know what latent variable(s) could be important or give rise to the phenomenon under investigation. So the researcher's question would be 'what variables/constructs give rise to phenomenon X'.

A radical approach would be using 'Retroduction' or 'Abduction'. I have not come across a scale developed using these logic of inferences but I am sure this would be an interesting approach. Retroduction is usually promoted by Critical Realists, it is very unique and combines both induction and deduction, qualitative and quantitative without violating philosophical assumptions. The idea is that the research could start from categories/constructs in the existing literature and check if they make sense for participants (bearing in mind that you may find new constructs from participants) or vice versa. In other words engagement in literature, theories and data will be an on-going iterative process.

I would recommend the following 2 sources of reference.

Best luck!

Cohen, R.J., & Swerdlik, M.E. (2009). Psychological testing and assessment: An introduction to tests & measurement. Boston, MA: McGraw-Hill Higher Education.

Hinkin, T.R. (1998). A brief tutorial on the development of measures for use in survey questionnaire". Organizational Research Methods, 1, 104-121.

Research methodology adopted depending on your research problem, research objective(s) and research question(s).  If the abovementiond are steering your research towards quantitative research using TAM model & you have 1 latent construct which is new after rigorous literature review, perhaps you can develop your own instrument / survey questionnaire.  However, you need to test for its reliability and various validity (content, criterion, construct - convergent / discriminant etc) before using data collected for hypotheses testing.

Thank you for your answers.

I am now extending my reading on "Theory of Perceived Risk" whether it is applicable to my study. If it is applicable, "professionalism concern" will not be a new construct and I will modify it to "Perceived professional risk."

Please comment on this.