In 2.24, the MPVI Inputs were sampled around 60hz, and averaged together to produce 20hz results.
In 3.0, the MPVI Inputs are sampled around 100hz, and are NOT averaged together. Each sample ends up being it's own data point.
What you miss when you use A/D inputs though, is the samples between the samples. For example, if the MPVI input is read every 10 milliseconds, what was its state at 1-9 milliseconds? You don't know.
When you use a serial input, some of the serial devices (you have to ask the mfgr) average the value between samples, so you get a more accurate value.
So it's possible to use a serial input that samples slower, but gives you a better overall reading of the sensor over using the MPVI A/D inputs.