In theory, a descriptive norm influences an individual by changing the individual's judgement of what is right. It is based on the idea that individuals use information from groups to make judgements about reality itself. (Deutsch and Gerard, 1955)
One way of conceptualising this in the train situation is social proof. You follow the people around you (i.e. are informationally influenced by them) because their actions are evidence supporting movement in a specific direction. You are assuming that members of the group know something you don't and are therefore following them. (Cialdini 1983)
Informational Influence on Smoking
If there was informational influence based on smoking, littering, vandalism, etc, then the action would stop being "obviously wrong". The informational influence would change the beliefs of the person who is being influenced, they would have to genuinely stop believing that smoking was "wrong".
So, for example, if I believed that smoking was bad because it caused cancer. Then, I hung out with people who smoked a lot, and I changed by beliefs to something like, "Oh, smoking can't be causing cancer. These people wouldn't smoke if it caused cancer", then that would be informational influence.
Spreading Disorder and Broken Windows
Now, what you might be getting at here is the somewhat debated "spreading disorder theory". This is a theory about the normative influence of littering and vandalism that combines descriptive and injunctive norms. (Keizer et al., 2008)
This model suggests the following mechanism:
- People don't engage in things like vandalism and littering because they are "wrong".
- People think vandalism and littering are "wrong" because there are injunctive norms against them. E.g. littering is convenient, but people are afraid of the social punishment that they might face if they littered, so they refrain from littering.
- When people see an environment with littering, they are informationally influenced and their judgement of the existing social pressures changes. Essentially, where they previously believed "Littering leads to social punishment", now they believe "Littering does not lead to social punishment".
- This leads to a weakening or disappearance of injunctive norms in general. So, "Littering does not lead to social punishment" is accompanied by the belief that, "There are fewer social punishments in this environment" (I.e. I can violate injunctive norms here OR there are fewer injunctive norms here).
- So, individuals are more likely to violate injunctive norms against littering and vandalism because they judge injunctive norms to be weaker/less present in their environment.
This is similar to Broken Windows Theory (Kelling and Wilson, 1982), which is also worth looking into and will have similar implications for your question.
So, in this model, there is an informational influence from littering, but it's indirect. It influences judgements of descriptive norms which then, in turn, influences behaviour.
An Empirical Note
Now, the studies I have cited will have some experimental evidence, but it's important to note that this is a somewhat debated area of social psychology. There are criticisms of these models, both on specific points of design and more broadly about conclusions. O'Brien et al. (2019) published an interesting meta-analysis that discusses some of this and might be a good starting point for empirical considerations.
So, the answer to your question is:
- If the group is "obviously wrong", there can't be direct informational influence unless it changes the individuals' belief about whether the group is wrong.
- There are models that suggest an indirect informational influence with things like littering. Here, the informational influence erodes the normative influence that would otherwise prevent things like vandalism and littering.
- The empirical evidence for this model is still somewhat debated.