Science is the mutual reification of abstraction and practice. This is not mere abstraction + experiment: that alone is mere conjecture, no matter how logically consistent the model or no matter how "empirical" the assumptions. No, science is a seamless fusion of craft and logic.
Physics doesn't work because "oh here's a very fancy model plus it lines up with assumptions correctly", it works because the physical tinkering and the abstract models tightly co-evolved as a single gestalt. Social "sciences" by contrast do not have this quality.
Psychoanalysis is far from perfect but unlike behaviorism it started en media res, in the interaction between subject and object, where all science MUST begin. It may not have ever developed sufficient rigor, but behaviorism by contrast is truly hopeless, as behaviorism can only posit axioms and an ultimately vague grammar of observables that didn't invididuate from a relevant practice but instead tried to derive an abstracted idea of "mind" with a set of building blocks that encapsulate pretenses rather than hard-won insight.
Operative word here: INDIVIDUATION
Science is a refinement of struggle, it grows and sets appendages from an embryonic mass, shaped by its attempts to calibrate itself with the chaos of the outside. Absent this, propositions are not even wrong, they're completely vacuous.
The very idea of correspondence (or lack thereof) between proposition and observation is one that doesn't exist outside of a practice to give sense to what a proposition or an observation is. There needs to be a formal notion of difference, and that itself is propositional, which itself brings us back to the problem of where those propositions got their meaning, and the only sensible answer to this is that the whole web of reference got its meaning through a chicken-and-egg process.
In other words, you need to grow your semantics, and that takes time, lots of time. Science may be a particular kind of struggle with a specific kind of outcome, but it is always struggle.