Pearson's chi-square test. The expected frequencies are the frequencies of the large population, while the observed frequencies are the ones from the created sample. Stratified sampling ensures that this test has a very large p-value, unless the created sample is too small.