You conduct a randomized study and find in your sample of patients admitted to the hospital with dementia, 31% will have a return emergency room visit within three months of discharge. National data becomes available at the end of the year, which then shows that within the full population of patients with dementia who are admitted to the hospital the actual return to ED rate within three months was 29%. You conclude:

Your estimate was wrong.

The difference reflects sampling error.

There is a statistically significant difference in the sample and population estimates.

We would expect the population value to be less than the sample value because of the larger sample.