Physically, a non-inertial frame of reference should modify what is seen but not shut-off length contraction, or time dilation, compared to what is calculated solely for inertial frames. Changes of reference frame are arbitrary and CANNOT change physical results, but observers can change their reference frames and physically see different things (it is not an illusion or perception, but becomes real to them).
Depending on the phenomena, a non-inertial reference frame may even lead to simpler results, important not to the phenomena but only in some human Platonic model, so there is no favoritism for or against using any non-inertial, inertial, the CMB, or other frame of reference. This is the basis of using special relativity (SR) with any kind of motion, accelerated or not.
The expectation is that SR does apply to ARBITRARY motion, not just to inertial frames. This expectation also requires general relativity (GR) and quantum mechanics to obey SR. There is no known exception to SR, according to all experimental data, not even in quantum mechanics. Even a body in free fall in a gravitational field [GR], locally obeys SR in spite of the gravitational field not being somehow anulled, as well-known.
This expectation has been realized by placing a 4D spacetime description, e.g. Minkowski, not the Lorentz transformation, as the basis of SR.
Time dilation and length contraction, real for two sets of non-comoving observers, where one set shares a common value for t, are not the same as those shared by the other set as a common value for t′ — i.e., the relativity of simultaneity — and that makes it much more efficient to think in terms of 4D spacetime, rather than space and time separately [3].
The discussion also suggests that the Lorentz transformation alone should have a less central role in higher education of SR, preventing current misconceptions [1] and providing for better reasoning, also of GR in undergraduate studies. This should prevent any surprise on the title of this discussion.
NOTES:
1. Preprint 24 Common Misconceptions of Mass and Energy in Special Relativity
End of rational discussions on SR, the sooner these answers are accepted, the better. You should convince yourself that these answers are correct. Why?
Yes, new entries could be added, personal style choices could be different -- but these are just fruitless, pretend facade choices. This is a meta-study of thousands of published results, with many contributors, with 19 informative references, as cited. So, not just the author agrees 100% with them, this is not just a personal opinion.
The list has also benefited from almost 10 years of public discussions, including WP and RG, and it has 24 entries collated in that experience.
All new questions here will be read and answered when possible (we have four readers in the team) -- but, "new" questions, even if repeated, may fit in the list of already given 24 misconceptions, so look there first!
2. We should follow what is correct, historically and in physics, not the false controversy that is oftentimes still presented. Mathematically, let us note the words of Minkowski [2], and not the difficulty that Einstein [2] momentarily had with them. The difficulty was that the Lorentz transformation became second-fiddle, introduced to keep the Minkowski metric valid, not the other way around. The original reference by Minkowski on spacetime, began with, "Henceforth, space for itself, and time for itself shall completely reduce to a mere shadow, and only some sort of union of the two shall preserve independence." [2], 120 years ago -- and this has been supported by all experimental data, and denied by none.
3. Geometry in spacetime is non-Euclidean for rotating observers. There are no paradoxes. That is another reason to abandon Lorentz transformations as the basis of special relativity. The basis of special relativity was changed to spacetime, more than 100 years ago, and Einstein was part of this change. This is the realization that we live in at least 4D, that space can transform into time and vice-versa. This discovery was not an invention, it was there billions of years ago, as we can measure with today's time machines, also called telescopes. On Earth's academic culture this dates back to more than 100 years, undergraduate college education needs to change, popular accounts need to change. There is no Ehrenfest paradox, Sagnac problem, or actual faster than light paths measured on star trajectories (even though we must experimentally raw-measure v>c in far enough paths, before proper correction in cosmology terms).
4. The term "relativistic electrodynamics" should be avoided as it points to the misconception that there might be an electrodynamics that is not relativistic. This does not exist in nature, although it exists still in some books.
5. The Lorentz transformation is a second fiddle in the spacetime treatment, not the basis. Quite a different position than initially proposed by Einstein a few years earlier, but the change to spacetime as the basis was accepted by Einstein, and so used in his general relativity to model gravity, as creating a curvature of the same spacetime.
6. The expectation has been realized by placing a spacetime description, e.g. Minkowski, not the Lorentz transformation, as the basis of SR. The discussion suggests that the Lorentz transformation alone should have a less central role in higher education of SR. As Mikhail G. Ivanov said in the first answer, "The Lorentz transformation [is] selected because L.transformation conserves the standard form of Minkowski metric."
7. So, we do need to abandon the Lorentz transformations as the basis of SR. It is finally time everyone does that. For more than one century, the Lorentz transformation is NOT an axiom, NOR the basis of special relativity, NOR "abandoned", though, but used as a CONSEQUENT of the spacetime approach, and so validated. This was said and done long ago, it is not a matter of research anymore, nor questions in RG, not even in history.
8. Without 4D spacetime, there is no GR, and no SR as we know it. With it, one can calculate ARBITRARY accelerated motion, motion in curves, and do it without the Ehrenfest "paradox", for example.
9. In summary, this is how the Ehrenfest "paradox" is solved: the coordinate system is not global, each part is considered in its own local non-inertial frame, and global quantities can be derived by integrating local contributions. The global geometry is non-Euclidean, so the usual Euclidean value of pi=3.141592... may not apply for all the global circumferences calculated at all points. The geodesic takes the role of a straight line, extending the concept of a "straight line" that is followed under no external force. The geometry in the neighborhood of each point is approximately given by the Langevin-Landau-Lifschitz metric.
[1] https://www.researchgate.net/post/What_are_the_worse_yet_enduring_misconceptions_about_mass_and_energy_in_special_relativity
[2] http://rgs.vniims.ru/books/spacetime.pdf
[3] Preprint Special Relativity Applied to Arbitrary Motion