"Actually, if you are talking about the net migration figures, it's not the case that if you operate the same statistical process each year, then trends up or down should be accurate. The margin of error on the net migration estimates (remember, they are just estimates, not actual factual data as to net migration), is 35,000 either way, which is pretty massive given the kind of numbers we are talking about. So if the estimate comes out as 100,000, the true figure could in fact be 135,000, or 65,000. And one year, the estimate might be too high, and the next it might be too low. So it's actually very difficult to get a true idea of levels of net migration."
Well it's not quite like that.
The published figure was 212,000 for y/e September 2013. The actual figure is equally likely to above as below 212,000. We can say that the figure is 212,000 plus or minus 35,000 (at 95% certainty).
So that means that there's a 95% chance that the actual migration was between 177,000 and 247,000.
But it's a bell curve.
Here's a shitty graph I just made in Excel:
i.imgur.com/ubpE6Sa.png
2012 is in orange, 2013 is in blue.
The green shading is the 95% range for 2013.
There is an overlap between the two curves, but the chance of it being close to 212,000 (the mean) is much higher than it being 194,000 (one standard deviation below the mean), and basically, as noted in the report, the change from 2012 to 2013 is statistically significant (because if you are talking about 2013 being much lower than the sample mean and 2012 being much higher than that's like the square of the respective probabilities, so statistically significant).