checkCIF procedure

It is expected that for centrosymmetric space groups, the r/p ratio should be at least 10:1. This also applies for non-centrosymmetric space groups when an element heavier than chlorine is present, because the heavy atom usually improves the diffracting power of the sample to an extent that makes it worthwhile to collect additional high angle data and thereby obtain higher r/p ratios.

For non-centrosymmetric space groups when only elements lighter than argon are present, it is expected that the r/p ratio should be at least 8:1.

If you are having difficulty obtaining a suitable r/p ratio, consider the following:
(a) Was theta(max) set high enough for the measurement? A theta(max) of 25 deg. for Mo radiation or 67 deg. for Cu radiation is the minimum requirement, but use higher theta(max) values if the crystal diffracts sufficiently.
(b) Consider using a low temperature measurement to improve the diffracting power and then collect to a higher theta(max).
(c) Check that the data are complete to your chosen theta(max) and that segments of the unique reflections are not missing. If data completeness is an issue leading to a low r/p ratio, you will receive a separate validation alert about that.
(d) Check that you are using all (or as many as feasible) of your reflection data in the refinement. If weak reflections are being omitted with a sigma threshold cutoff, such as the criterion defined under _reflns_threshold_expression (this may be particularly relevant for refinements based on F), then you should consider reducing the cutoff threshold as far as possible, or, preferably, use all reflections in the refinement.
(e) Reduce the number of refined parameters by, for example, not refining H-atoms, using restraints or constraints, or refining highly disordered or ill-behaved atoms isotropically.

Note that if you have measured data to at least sin(theta)/lambda of 0.6 (theta(max) = 25 deg. for Mo radiation; theta(max) = 67 deg. for Cu radiation) and you have used all available unique reflections in the refinement, the reflection/parameter test is not applied, because it is considered that you have measured an appropriate quantity of data and that you have made the best possible use of that data.