The two tests give me widely different results, probably because the sampled words aren't perfectly representative and so the results should have huge error bars to account for this sampling error.
The two tests give me widely different results, probably because the sampled words aren't perfectly representative and so the results should have huge error bars to account for this sampling error.