Hi Mike,
When looking at the time domain data you need to take an average of the signal that is within the noise. So if you see a bunch of noise, then the mean value would be in the center of the noise. If there is a signal within the noise, then you would measure between the mean of the high value and the mean of the low value, which is the peak to peak value.
As far as determining the PSRR of a 10kHz signal, yes it is a little difficult to measure as you are outside the bandwidth of the part. If you see anything, it would be the aliased signal folding back into the passband. When you do not see any value, then the rejection is within the level of noise. The analog supply is the most sensitive, and Figure 13 of the ADS1220 datasheet shows the rejection out to 1kHz.
If you are using a switched mode power supply, it is always advisable to connect an LDO to the output of the switcher to reduce the high frequency switching noise prior to the supply input to the ADC.
Best regards,
Bob B