That's stupid, because they should really weigh their results by population. It's not like people in Alaska are going to vote at the same numbers as people in California. And what's their phone bill after 50,000 phone calls?
I have questions.
1. Did they use a random-digit dialer? Live interviewer or IVR, and were their known effects controlled?
2. Did they call cell phones or provide Spanish-language options? Blended-frame?
3. What were the timeframe and calling hours? How many attempts per number?
4. How did they randomize within households?
5. What demographic data was collected?
6. Were questions and answers randomized in any way?
7. Was the questionnaire tested for wording and order effects?
8. Was a likely voter screen applied? How hard were leaners pushed?
9. Was the sample weighted, and how? ACS, voter reg, both? Was aging out/in a factor? Was the process simultaneous, iterative, or both?
10. Do they know how to calculate the margin of error for each proportion? What confidence level was used?