Summary
A Fox News host, Jessica Tarlov, highlighted Trump’s declining approval ratings, citing a new Quinnipiac poll showing 53% of voters disapprove of his second-term performance.
Trump is “underwater on everything” including economy, federal workforce management, Ukraine-Russia policy, and trade relations with Mexico and Canada.
The polling slide comes amid stock market tumbles, federal worker layoffs, and trade tensions.
Tarlov noted that Republican members of Congress are facing hostile town halls where constituents are challenging them on federal workforce cuts, including at the VA. Some Republicans have dismissed these angry constituents as “paid actors.”
I do think they need to have a good sample size to get results that make sense, yes.
Doing some research on it, seems like Quinnipiac’s standard polling is conducted “in more than 20 states and cities as well as nationally” (particular locations not specified) and has typically 1,000+ respondents using random digit dialing. Meaning they just slap random numbers into the phone and see who picks up.
I can’t say I can immediately come up with a better solution right off the top of my head right now, but I can identify several problems with their approach, namely being actual size of sample (1000 respondents is 0.000003% of the population) and the way they gather their responses (when was the last time you answered the phone for an unknown caller? Now when was the last time your grandmother did?).
From what I remember from university stats, 1000 is the standard sample size for these types of things and accurate to a reasonable margin of error (±3% iirc).
This is assuming that it’s truly a representative sample though, and frankly I’m with you that I do have my doubts that over-the-phone surveys sufficiently represent the youth.
You are still showing your lack of understanding of statistics. It requires a large sample size, yes, and the larger it is the more accurate it will be, but the proportion of the total population doesn’t have to be large at all. Margin of error and confidence are both controlled by the absolute sample size (assuming a random sampling). Here’s a link that covers the basics. Most serious polls use a sample size of about 5000. Now, 1000 isn’t terrible, but it isn’t good, but not because it’s such a tiny fraction of the population.
As for the methodology, which is also critical, I don’t know what method for phone surveys is considered adequate, and it is well known that phone surveys are becoming less relevant even if they are more accurate than polls where the respondents are self-selected.