In his experiment, Duncan Watts created a website where respondents could rate songs. His sample size was 14,341 respondents. If Watts selected a new sample of 14,341 different respondents and their average ratings of each song were significantly different than the first sample, what problem would Watts’s research suffer from?