Yesterday, from a suggestion of @Dimitriy V. Masterov here, I saw from the given link about one of the reason we can avoid clustering is
You want to say something about the association between schooling and wages in a particular population, and are using a random sample of workers from this population. Then there is no need to adjust the standard errors for clustering at all, even if clustering would change the standard errors.
I am not fully understanding the words "particular population" and "random sample" here. For example, in one paper, Dasgupta, 2019 examines the impact of anticollusion laws on firms' asset growth in a global context where firms are not utility and financial firms. So, whether in their case, this research satisfies two conditions "random sample" and "particular population"?
