The second survey is more believable because there is randomness in the second rather than the first survey. Since the second company surveyed in all interest levels while the first company only surveyed their best customers. Random sampling like what the second company did helps produce representative samples by removing voluntary response bias and undercover age bias.
