It is different because, for starters, I'm talking about taking into consideration a continuum between more and less direct observation or description, not about dismissing or ignoring studies.
It is different because, secondarily, I'm questioning different types of presuppositions and their effect upon methodology and how such things limit what is thought to be possible.
You are arguing that our observations of a supernova 150,000 light years away NEVER HAPPENED!!! You are saying that we should ignore the observations.
No. I'm saying it's more accurate, and in many ways helpful for people to understand what's really going on -- if we are more descriptive about what we ACTUALLY did and what was recorded through what sort of instrument, and what we CONCLUDED about it after we filtered and processed all this through such and such presuppositions, theories, formula, etc. -- this is more informative, more descriptive, than jumping from the data that one obtained from instruments right to the conclusion.
I'm talking about being more transparent about the processes, methods, inferences, assumptions, etc.
And again, I am not using speculation as an entirely negative term -- science IS about speculation -- speculation that is based upon various observations using various instruments and methods and tested in an attempt to see if the speculation may be refuted or not. Karl Popper talks about this in Conjectures and Refutations. Go ahead, speculate, and try to disprove your speculations, and share your findings with others who may also try to refute your speculations, share and discuss.
In a given research project some of it may involve pretty direct observation and description. Other parts involve a small degree of abstraction and speculation based upon evidence that has withstood a considerable degree of scrutiny. Other parts involve a GREATER degree of abstraction and speculation, or a GREATER reliance upon assumptions which may not be testable.
So there are different aspects to a given project that may or may not be 'hard science' or 'soft science', may or may not be more or less speculative or tied up in some philosophical presupposition.