50 CHAPTER 4 Did We Meet or Exceed Our Goal? Example 2 If 233 out of 250 users were able to complete a task in an unmoderated usability test, is there enough evidence to conclude at least 90% of all users can complete the task? There are at least 15 successes and 15 failures, so using the large-sample method is appropriate: ^ p - p 0:932 - 0:9 z = r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi = r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi = 0:032 = 1:687 0:019 pð1 - pÞ 0:9ð1 - 0:9Þ n 250 We look up the obtained z-score of 1.687 using a normal table of values or use the Excel function = NORMS- DIST(1.687) = 0.9541 and subtract that from 1 to get the one-tailed p-value of 0.0459, which indicates a statistically significant result. We can be 95.4% sure at least 90% of users can complete the task given that 233 out of 250 did. The 90% adjusted-Wald confidence interval is between 90.1% and 95.4%. COMPARING A SATISFACTION SCORE TO A BENCHMARK