From an investor point of view, correlation can be good enough. If startups with property X are measurably more likely to succeed, it doesn't matter whether X causes success or if X is caused by some other property Y of the startup that causes success but isn't easy to determine in an interview.

In your example, X = "has multiple founders", Y = "first founder is not impossible to work with". Clearly Y causes success, but X is easier for investors to determine.

From a founder point of view, you want to focus mainly on the real causes of success rather than appearances, but give some thought to appearances. Customers are far more influenced by appearances than investors (because they have less time to dig in), so when you devote time to appearances you should think mainly about customers. Customers tend to be skeptical of one-man shows.