Ask us for advice. We love survey analysis! (and need to put that SPSS license to good use)
AnswerPanel respondents are not randomly selected. Textbooks call this nonprobability sampling. There are biases, such as respondent self-selection bias. Large commercial panels are also nonprobability samples. That’s not to say AnswerPanel isn’t incredibly useful, but keep in mind that it is not a true random sample.
Weighting your sample could make it more representative. The 2010 U.S. Census provides demographic data that can be used to calculate weights (assuming your sample is from the U.S.).
There are a bewildering number of online survey applications. SurveyMonkey is the most popular and not a bad place to start.
Statistics.com. Over 100 online courses.
We’re fans of the Dummies series:
These guys literally wrote the book:
More useful references: