Sampling Explorer
Chapter 6. Variable error and coverage bias.
The simulated population has 10,000 households. Of these, 3,500 rent and 6,500 own. Forty percent of renters and 15 percent of owners report a housing cost problem, so the true share is 23.8 percent. The tool draws 300 random samples and plots the 300 estimates. The data are simulated.
0average of the estimates
0bias, in points
0middle 95 percent of estimates
The dark line marks the true share. Try two things. First, raise the sample size with no renters missing, and watch the estimates move closer together. Then set 40 percent of renters missing and raise the sample size again. The estimates move closer together around the wrong value.