The refractive deviations in position are very similar for both radio and optical astronomy, until you consider very low frequency radio waves ($<200$ MHz) when the effect becomes rapidly larger.
For plane parallel refraction an approximation for the deviation you are talking about is
$$\Delta \theta \simeq (n-1) \cot \theta,$$
where $\theta$ is the observed elevation, $\Delta \theta$ is the change in elevation from its true value due to refraction and $n$ is the refractive index averaged over airmass.
According to this source from the Green bank radio telescope, they use something like this, with an added model for how $n$ varies with height, scaled by the atmospheric pressure. The largest value of $n$ quoted is 1.00031 at ground level. This is basically the same as the refractive index of air at visible wavelengths and amounts to about 60 arcseconds at elevations of 45 degrees.
So, to my surprise, the effects of refraction on radio telescope pointing are similar to those for optical telescopes. It simply turns out that the real part of the refractive index (that controls the phase velocity of light and hence refraction) is just as close to 1 for radiowaves as it is for visible light.
Here is another source that gives some algorithms to calculate the effective (small) real refractive index for radio waves, with similar results.
This source claims the calculations are valid for mm-wavelengths and upwards. Of course they cannot be valid as one approaches the ionospheric cut-off at around 40 MHz (wavelengths of metres), where the refractive index will depart sharply from unity and the deflections must get commensurately larger.
I did manage to find something about positional refraction at these low frequencies. The LOFAR radio array can apparently do work down to frequencies as low as 10MHz, but the practical limit may be a little higher. Anyway, they do need to account for refraction in the ionosphere and I found this presentation, which contains a section on refraction and in particular, the figure below.
Thus for low frequency radio astronomy (<200 MHz), refraction is certainly a bigger effect than in the optical. e.g. At elevations of 45 degrees, the refracted positions are shifted by about 0.1 and 0.4 degrees at 50MHz and 30MHz respectively.