...
update: We are up to 30 responses; still at 11 error reports. ...
how many before it is significant?
Also, you made reference to my "only 11 out of 25" figure from the SD card survey. My point there was that 25 responses is statistically insignificant to start with. Out of that, only 11 problem reports, all from people using cards not on the manufacturer recommended list is hardly enough evidence to establish this as a widespread issue. This is out of the many thousands of units sold, and no problems of this type have been reported anywhere else that I'm aware of. Should those 11 people have customer service work with them to help resolve their issues? Absolutely, but if that response is "use an approved card", then that shouldn't be viewed as an unsatisfactory response.
No survey here could ever represent all the units that have been sold in the industry, so the statement about statistical significance against the thousands of units sold is irrelevant to the survey, not to mention that nobody here can perform any real statistical analysis against all of the units sold anyway so trying to claim any specific statistical analysis conclusions from this survey as significant or insignificant is bogus.
For example, in my limited experience, a RANDOM sample of 25 is statistically significant to I think a 99 percent confidence factor to a total population of several thousand. I know this because this is the sample rate the NRC requires for sampling in inspection programs in nuclear plants where the sample size is say several thousand valves in the plant. Depending on the acceptance criteria being used, in my industry, a single fail rate in the random sample invokes a corrective action plan (usually increasing the inspection sample size and/or doing targeted inspections to get a better understanding of root cause). Whether this survey constitutes a RANDOM sample is debatable, but that's my point about reaching ANY conclusions about statistical significance or insignificance of the survey results.
When I suggested the survey, I thought there were several things the survey could tell ts.com members a) cards that don't work, so cards to avoid, b) cards that aren't on the official list that people have had success with, so that maybe some alternatives to the tiny 'official' list would be available, and c) some general feedback to give an idea whether the issue is an isolated incident experienced by only a few or a more generic problem. There's enough there to start to help people with a) and b) and IMHO there's enough information there to reach a reasonable conclusion on c).
Someone that wants to use the survey to help with future decision making can take the information for what it's worth in assessing a), b) or c) (or perhaps some other reasonable conclusion I haven't thought of). Someone that's intent on arguing against IMHO overwhelming evidence that suggests there's something amiss on the DR70D design, and that Tascam seems to have been putting more effort into CYA than into solving the root cause of the issue, will come up with irrelevant talking points to continue to try to steer people away from the fact that the DR70D has an, as yet, unrevealed design issue that makes it ultra picky about the cards it uses.