Look I’m not defending Ariely, just saying that random sampling can be more complex than each record exactly the same sampling chance. And if the population has

Look I’m not defending Ariely, just saying that random sampling can be more complex than each record exactly the same sampling chance. And if the population has a few overpopulated groups but you’d like results for all groups, you don’t throw extra samples at it but use smarter sampling.