4 ms·
I would upvote if it wasn't for the stance of disagreeing with the implications of data that you don't have any reason not to trust. What you describe clearly
by t_mann 4y ago
I would upvote if it wasn't for the stance of disagreeing with the implications of data that you don't have any reason not to trust.
What you describe clearly is a problem with data collection, where crucial unobservable variables had to be replaced with questionable assumptions. We should be wary because we can't fully trust those assumptions, and hence not the data, not because of what the implications would be.
- carbocation 4y agoObservational data is so profoundly affected by the decisions and assumptions of those who chose to select them and analyze them in a specific way that I cannot agree with this premise that the implications should be “trusted”. Those implications are a function of the choices made by the people making those decisions. The data have no opportunity to tell an independent story. There are tools that can make this better (eg, finding an instrumental variable), but this wasn’t done here.