I got the impression that the scientifically-relevant (e.g. non-ethical) questions you list would be asked. Having said that, the community should be more concerned with 'finding the right tests' - our "safe" method of testing drugs we suspect are okay is animal testing (whether that is ethically right or wrong), and animals are probably easier to test objectively in this situation, as well as possibly immune to the placebo effect.
I'm under the impression that questions like: "Is it possible that lab conditions affect the psychology of the animal?" aren't asked because:
* No one wants to know the answer if it means that we no longer have a test framework to experiment within. If tomorrow we found out that we couldn't use animals as test subjects because the validity of any such experiments was suspect, then how would 'we' generate grant money?
* Hand-wavy answers like, "It's impossible to know, so we'll just assume that the answer is no."
Wrong! Those are exactly the questions that are asked.
The tests use a group of control subjects vs test subjects which is standard throughout science. This means it takes out a factor such as what environment the animal is stored in by simply storing them all in the same type of environment. This is done every time a study is run to remove minor variations.
Science is all about controlling the variables, it is actually what scientists spend most of their time thinking about.
What really went wrong in this case is when they got to the point of testing humans they stopped doing "science" and started doing "business".
If they hadn't sweeped the bad data under the rug they could have published a lovely paper about how the standard tests for depression in animal models failed (in reboxetine's case, one goat a herd does not make, the tests could just be bad for reboxetine for some reason).
Then you have the wrong impression. That question is asked, and the answer is "yes." The followup questions are "by how much?" and "what can we do to minimize the effect on what we're studying?".
Validity tests are important because if you're going to blow $50,000,000 on research you want _something_ to help guide the way, and you don't want it messed up by having non-conducive lab conditions.
Searching the free access medical literature at NCBI I quickly found:
"Refining animal models in fracture research: seeking consensus in optimising both animal welfare and scientific validity for appropriate biomedical use"
"From Bedside to Bench and Back Again: Research Issues in Animal Models of Human Disease" with the abstract starting "To improve outcomes for patients with many serious clinical problems, multifactorial research approaches by nurse scientists, including the use of animal models, are necessary. Animal models serve as analogies for clinical problems seen in humans and must meet certain criteria, including validity and reliability, to be useful in moving research efforts forward. This article describes research considerations in the development of rodent models."
I'm certain that you'll easily be able to find others.
