More misplaced statistics on the Welsh language

The Welsh Language Board (which has made previous appearances here) is keen to promote the greater knowledge of Welsh. But it might try to promote the greater knowledge of statistics, too.
A flyer advertising the services of a local Welsh language group supported by the WLB has contrived to quote three statistics about the language, and get all of them wrong.
Cyfieithu Cymunedol Conwy is one of a network of local groups, called Mentrau Iaith, or Language Initiatives, which promote the language locally and provide translation services. There are seven of them in Wales.
The Conwy branch made three claims in its flyer, picked up and challenged by a Straight Statistics reader. These were: 92 per cent of Welsh speakers think it is important to offer bilingual signs/leaflets; 87 per cent of people in Wales believe it is important that the private sector uses bilingual signs; and 76.1 per cent of the population of Gwynedd can speak Welsh.
In response, Cyfieithu Cymunedol Conwy cited a survey done for the WLB by Beaufort Research in 2005 for the first two claims, and the 2001 Census for the third. But my correspondent was not satisfied. He questioned the WLB itself, which acknowledged that all three figures were wrong.
The 76.1 per cent of Gwynedd people who can speak Welsh turned out to be the answer to a 2001 Census question about the number of people who claim to have at least one skill in Welsh, rather than the number who claim they can speak it, which is just under 70 per cent.
The 92 per cent who believe it is important to offer bilingual signs/leaflets should have been 82 per cent. It derives from a poll carried out by NOP as long ago as 1995. WLB says the Conwy group “misquoted us here”. The poll, available on the WLB website, does not show an actual table of results.
But the most intruiging of the WLB’s responses was to the question he asked about the claim that 87 per cent of the people of Wales believe it is important that the private sector uses bilingual signs. WLB admits: “The statistic quoted is not correct. This was a genuine mistake on our part and we apologise for the error.”
The error appears to have come about in this way. In the 2005 poll, in answer to the question “how important do you think it is that privately-owned companies provide bilingual signs?” the respondents replied: 

  • Very important 26 per cent
  • Quite important 39 per cent
  • Not very important 19 per cent
  • Not at all important 13 per cent
  • Don’t know/not answered 4 per cent

The 87 per cent was reached by assuming that all bar those who didn’t know or answered “Not at all important” could be described as believing they were important. That included the 19 per cent who said they were “Not very important”.
The same poll asked how people respond to bilingual signs and packaging, to which only 14 per cent said yes to the statement “I like it and I’m more inclined to use their businesses or their products” while 38 per cent said yes to “I don’t care one way or the other and it doesn’t influence me at all”.
To another question, 43 per cent said that they agreed that banks, mobile phone companies made enough use of the Welsh language, against 34 per cent who disagreed.
So it is clear that the 87 per cent claim, as well as being wrong mathematically, is an answer selected to give an impression that the poll, taken as a whole, does not justify.