I too the liberty of examining the original survey. The headline finding is that
Leaving aside the fact that a one percent increase is hardly a 'significant' improvement, the survey methodology makes no mention of error margins, which is highly suspect. Error margins in surveys are often in the range of 2-5%, which could completely (and 'significantly') reverse this claim. Hmm. I wonder what else is wrong with this survey..."Nationally the percentage of passengers satisfied with theirjourney overall was 85%. This is significantly improvedcompared to Autumn 2011 (when 84% of passengerswere satisfied). 83% of passengers were satisfied overallwith their journey in Spring 2012."
Oh, here's one (on the same page, no less):
So a one percent increase is significant, but a one percent decrease is insignificant. Yeah, that makes sense. I'll stop short of calling the watchdog far from as independent as claimed, but those are some pretty major flaws in interpretation."For regional operators 86% of passengers were very or fairlysatisfied with their journey overall, not significantly differentto Autumn 2011, when 87% were satisfied."
Diving further into the report, one can find numerous instances of a 2% change (up or down) being described as insignificant, but a 1% change as significant. And still, though sample sizes are provided, no margin of error. And a quick glance shows me at least one instance of a single percentage point rise in a small sample being treated as more significant than a larger percentage point rise in a big sample.
You can download the report yourself (free) from here to check out your individual train operators, but frankly I've seen enough to ignore all the findings as fundamentally flawed.
In other news, if you're the conspiracy theory type, you can find 'evidence' here that TfL is experimenting with nerve gas to kill its customers directly (rather than as is the current favoured method: indirectly by frustration).