A New York Times focus group with Trump-supporting voters between the ages of 20 and 25 mostly found glum outlooks on the economy and world affairs — and an overall sense that the system, even under the man they voted for, just isn't working.The focus group included 10 people, including seven Republ...
Focus group of 10 people feels kinda meaningless.
What about a focus group of 10 large language models?
Assuming they are trained on datasets that do not control for this classification, and are derived from the product of more than ten individuals I think it’s safe to expect, unironically, that yes the results would be more predictive. (ETA: but not grounded, and pointless as a survey.)
Always been curious how effective focus groups are and what sample sizes would be considered representive of the general population.
This is regularly answered by statisticians
“We picked 10 unicorns and asked their opinions to try and make people believe all horses feel this way”