I find it weird when they use percentages for groups that are smaller than 100. 9% of a group of 32 people makes it feel like 2 people and a torso handed over their phone without the rain apology. (I'm aware this is my poor math skill, and it's something I'm working on.)

I don't know if the sub has been edited, but it mentions lack of controls:

> How trustworthy are these results? The accumulated findings from several experiments help build a convincing case, but unfortunately the field study - which had the potential to provide the most persuasive evidence - is seriously flawed. The actor apologised for the rain then asked to borrow a phone, or in the comparison condition he just asked to borrow the phone. There was no proper control condition. This means we don't know if the impact of the apology was specific to making an apology or merely an effect of uttering any kind of ice-breaker.

It doesn't seem to mention randomization either.

EDIT:

> This is significant because past research shows how mindlessly we often act in social situations. For example, back in the late 70s, Ellen Langer and her colleagues found that people were just as likely to give way at a photo-copier if a queue-jumper uttered the nonsensical excuse "because I need to make copies" as when he uttered "because I'm in a rush."

Compare this to situations in the UK where if a shopkeeper returns nearly all your change, but keeps a penny, the customer will often wail until they get it, and make a fuss about it.