Answers which are wholly based on a theoretical model are generally downvoted and may be deleted.
What are the characteristics that (dis-)qualify an answer as theoretical?
The general rule for answers is that:
Every answer should be suitable for review and voting by our community of experts in evaluating evidence. Answers which require non-trivial specialist expertise or are anecdotal in nature are not acceptable due to the nature of this community.
One of the premises of skepticism is the application of the scientific method: empirical proof validates (or disqualifies) theoretical models. All questions we allow here are empirical in nature, thus answering via a purely theoretical model is inappropriate: experiments are not "validated" by theory, but vice-versa.
Here is a list of common examples of types of unacceptable theoretical answers.
Answers based on simplified calculations instead of measurements are theoretical. By nature, they implicitly assume a mathematical model, but they generally fail to show that the model is adequate to the circumstances of the question. They also don't investigate their own inaccuracy. They are a form of Original Research.
It isn't enough to add two numbers together, and use that to draw a conclusion. They must also show that these are the right numbers to add, and that adding only those two numbers is the right operation to match reality, and no other factors or complications have been omitted. The popularity of the Martingale Betting System and pyramid schemes are examples where people can be taken in by back-of-the-envelope calculations that don't consider all the issues.
Regarding uncertainty, it's easy to add two numbers, but knowing how to calculate the resulting uncertainty (see propagation of uncertainty) is typically university level knowledge and needs to be done if one is matching a calculation to a claim.
We don't normally accept such calculations. In some rare cases, we can accept such calculations if they are directly quoted from a reputable expert in the field. Even then, they are arguments from authority and might be downvoted or removed if the calculations are used out of their realm of applicability, or abused to draw unwarranted conclusions.
Even when applicable, this is very weak evidence and needs to be described as such (it shows that a claim is likely or unlikely, maybe, but not definitively true or false). Some questioners may specifically choose to disqualify this type of evidence.
An answer is theoretical when it is based upon a model, and the correct application of such a model can only be evaluated by experts in the field.
We don't accept research-level material, because we our community generally is not qualified to vote on it.
The best way is to find a peer-reviewed paper that replicates the material. For example, in some cases researchers answered with their own research, but published it first on a peer-reviewed journal.
As a last resort, post your research level material on the appropriate SE site (e.g. Physics, Chemistry, History...) and then use that as evidence, once voting has taken place and shows consensus that it is indeed valid. Clearly, this is weaker than a proper peer-reviewed publication and you may need to weaken your conclusions accordingly.
Alternatively, if the question can only be answered with such research-level answers, it may be closed. If we can't hope to get good answers that can be fairly judged on their merits, the question would be better resolved in conferences and journals.
Answers that rely only on logic and maths are theoretical. All our questions are inherently referring to experimental evidence. If your answer does not contain any, it's almost certainly not answering the right question.
Of course, all answers will rely on some level of logic. However logic by its own nature can only useful to connect the dots. At its heart every answer must be based on facts, which you are required to identify and support with references.
Also, please be aware of the distinction between formal logic, the kind we are speaking about here, and informal logic which is never allowed as the main support for an answer.
It cannot be that axioms established by argumentation should avail for the discovery of new works, since the subtlety of nature is greater many times over than the subtlety of argument.
If there are any assumptions in your answers, you need to provide evidence for them. Make sure that your answer does not become a "Back of the envelope" or "Research level" answer though.
If you think that there aren't any assumptions in your question, then your answer is probably a "common sense" answer and needs to be rewritten.
It is a fallacy to make appeals to common sense (a.k.a. argument from incredulity). Science is replete with stories of careful research that overturned basic tenets of common sense (e.g. that the Sun orbits the Earth). Answers that rely on common sense are not acceptable.
Common sense is the collection of prejudices acquired by age eighteen.
-- Albert Einstein (maybe)
If there are any passages in your answer that rely on common sense, you will need to reference them.
If the whole answer relies on common sense, you will need to rewrite it.
This site is fairly unique in this approach in the Stack Exchange network. This leads to many users, especially new users, to expect to be able to answer with theoretical answers. In fact, this answer is an attempt to guide new users to answering questions correctly, and for experienced users on how to moderate the site.
These answers are low quality for this site and should be down voted.
New users should be routed here for guidance on how to fix via comments.
Moderators, and high rep users, might vote to hide such answers (through soft-deletion) in some specific cases:
The purpose of hiding is to prevent voting until glaring problems are fixed. Original posters and high rep users should try to fix the answer via editing (if possible) and flag for moderator attention, or comment if allowed.