I appreciate the time and effort that davisite has put into his...

I appreciate the time and effort that davisite has put into his chart.

It's funny how two people can look at the same chart and draw very different conclusions.When I look at his chart, two things stick out:

1. The standard error US 5% BID does not overlap the US combined vehicle. If the null hypothesis were true, these would overlap. Therefore something else is going on here.

2. All other things being equal, the size of the standard error for the US data should be about 30% less than the size of the standard error for the Aussie data, due to much higher US sample size (inverse square root relationship). This is not what we see. This means that the US results must have been all over the shop.

Both these points support the company's assertion that there were QA issues with the US batches. On this basis I am prepared to reject my previous speculation about swapped batches. Perhaps the proponents of the null hypothesis might do the same.