Th3ere must be much less in science than elsewhere for 2 reasons. Deception is easily detected and generally there is much less monetary reward for deception.

Now for the other extreme, consider tax returns.
Here deception is the rule because it is not easily detected and the monetary reward is immediate. Politics is probably next. Here deception is called spin and has become acceptable within certain limits, not always legal limits as Clinton found out. His testimony was strictly in accordance to the judges prior rulings, but it exceeded the political limits.
Next is advertising, where exageration is expected. Then lawyers who are expected to lis to defend their clients.

Somewhere down the list is medical practice where mistakes are covered up and operations are performed more often than required.

Even less fraudulent than all the above is manufacturing. Mistakes are easily detected. Things do not work. So the process is self-regulating via quality control. That extends to software as well. Except maybe for Microsoft who seems to be find buggy programs on any given version acceptable. Yet they too eventually fix all the known bugs.