Press releases must be concise to be read and quoted. I struggled with how to CONCISELY describe the categories of people who answered question 1 about voting frequency. Since SurveyUSA polled people in the state who may or not be registered to vote, if someone answered question 1 as “don’t vote regularly” I consider them a non-voter, since they don’t usually vote or never vote or can’t vote, and the people who vote in presidential elections only or in all elections as “voters.” This means that if I want to report the results for “voters,” I have to look at the cross tabs and average together two categories: the people who vote in presidential elections only or in all elections. = (63*0.24+66*0.45) / (1-.31) = 65% That gives me 65% of Texas voters wanting secession in question 2, and I just figured 65 is close enough to 66 because the rest of the paragraph is about all voters. The 59% number comes from a similar calc, averaging the two categories of Democrat voters together (58% and 60%).
For the paragraph below the table, the phrase “who expressed an opinion” means I stripped out the people who answered “not sure” because I find it annoying when a journalist quotes a positive percentage from my press release without including the fact that some of the people were undecided. If the reader is not told about the undecideds, then he will assume the remainder were all opposed. So for question 17, looking at the cross tab and the percentage of respondents in each category,
I calculate the overall support among decided voters as = (50*0.24+61*0.45) / (100*0.69-(24*0.24+12*0.45)) = 68%
and I calculate the support among decided regular decided voters as = 61 / (100-12)*100 = 69%
Similarly, the 81% number comes from 67*100 / (100-17).
PS. A journalist asked me how the numbers in my press release fit with the raw data reported by SurveyUSA. Here is my response:
Yes, the 66% number is from question 2, from the “Voting by Party” cross tab for voters who said they vote in every election in question 1. https://www.surveyusa.com/client/PollReport.aspx?g=93235179-7a22-454b-ac4d-91b2e5119238
Press releases must be concise to be read and quoted. I struggled with how to CONCISELY describe the categories of people who answered question 1 about voting frequency. Since SurveyUSA polled people in the state who may or not be registered to vote, if someone answered question 1 as “don’t vote regularly” I consider them a non-voter, since they don’t usually vote or never vote or can’t vote, and the people who vote in presidential elections only or in all elections as “voters.” This means that if I want to report the results for “voters,” I have to look at the cross tabs and average together two categories: the people who vote in presidential elections only or in all elections. = (63*0.24+66*0.45) / (1-.31) = 65% That gives me 65% of Texas voters wanting secession in question 2, and I just figured 65 is close enough to 66 because the rest of the paragraph is about all voters. The 59% number comes from a similar calc, averaging the two categories of Democrat voters together (58% and 60%).
For the paragraph below the table, the phrase “who expressed an opinion” means I stripped out the people who answered “not sure” because I find it annoying when a journalist quotes a positive percentage from my press release without including the fact that some of the people were undecided. If the reader is not told about the undecideds, then he will assume the remainder were all opposed. So for question 17, looking at the cross tab and the percentage of respondents in each category,
I calculate the overall support among decided voters as = (50*0.24+61*0.45) / (100*0.69-(24*0.24+12*0.45)) = 68%
and I calculate the support among decided regular decided voters as = 61 / (100-12)*100 = 69%
Similarly, the 81% number comes from 67*100 / (100-17).