If you bring two identical ADXL202 with their sensitive X-axis into a line, pointing to the center of the Earth, then each will measure 1g. Taking the average by summing up their signals and dividing the result by two will yield 1g.

Now imagine you put the two ADXL202 into a rocket which is leaving the Earth's surface with an acceleration of 1g, then both ADXL202 will measure 2g. Taking the average again by summing and dividing will also yield 2g. So, the series circuit of two will not help to cancel acceleration affects.

Of course, you can reverse one of them and subtract the signals from each other. But this will not change anything.