Unless the majority of your readers are
Microsoft employees, a RANDOM sampling of
your readers would not have any intristic
bias towards software Microsoft produces.
The key to having a successful survey are
1. selecting a truly random sample of
appropriate size and
2. wording the questions of the survey so
that they do not convey any bias
Disqualifying any employess of hardware and
software vendors does not guarantee the
randomness of the sample. It may, however,
introduce bias (whether or not it does
introduce bias could only be settled by
comparing distributions among samples taken
from only vendors only not-vendors).
The 300 number also, can not be compared to
your 300,000 subscriber number, for
determining statistical relevance. You need
to compare it to the number of subscribers
who comprise your population, Infoworld print
subscribers who do not work for a hardware or
software vendor.
And the results are only relevant to that
population. The only problem I have with
this survey is that Sandy Reed tried to paint
the results as being representative of both
readers of Infoworld print and readers of
Infoworld Electric.
The population the sample was taken from does
not necessarily reflect either.
IMHO, the best thing to do for future readers
choice awards would be to return to the
original self-selecting poll within the pages
of your print magazine. This would make
ballot stuffing extremely difficult (but not
impossible). It would make the readers
choice award fun again, instead of getting
all bogged down with statistics and what not.
regards,
Lee Malatesta
XLConnect Solutions, Inc.
opinions expressed herein are my own
and do not represent my employer
For questions or comments, email the Forums Editor