As the manual points out, this may sometimes happen. What the device really understands the RDS deviation is a level of spectrum components around 57 kHz after the digital FM demodulation, multiplied by a constant that represents relationship between average and peak value for normalized RDS signal shape. There is some additional criteria like RDS spectrum mask pass method but it is not and cannot be so strict due to noise. Additionally, if phase relation has been found between pilot and spectrum components around 57 kHz, the device switches to higher sensitivity, i.e. expects that it is RDS and skips some criteria.
Usually false RDS detection may happen if the signal reception is not perfect, the pilot has 3rd harmonic or if specific sine tones are transmitted (there are several other reasons, not so important). The FM demodulation has triangular-shaped noise spectrum, so RDS is the most critical part of the received baseband signal if affected by noise or interference; determination of the RDS deviation becomes impossible although nothing else is affected. This phenomena is unfortunately stronger in the P175 due to aliasing effect given by relative lack of DSP power. The effect of noise can be reliably suppressed for the overall peak deviation, this is however inapplicable for the RDS deviation.
In general, RDS can still be decoded although one cannot recognize it in the spectrum due to noise. That does not apply for indication of the RDS deviation which is reliable only during perfect reception. In other situations a decision must be made: it is a noise or RDS? In your version, it has been shifted in favor of the RDS.
