Analyze Data
T-test Example
Research reports often include mean values for particular questions. For example, a customer satisfaction rating for a hotel may be reported as 4.1 on a 1–5 scale and the hotel might be concerned about this if its rating for the previous year was 4.3.
However, in order to know whether they really need to be concerned, it is important to determine whether the difference in means is statistically significant, and that can be done using a t-test.
T-tests can also be used to investigate which possible explanations for the change are statistically significant. For example, these might show that satisfaction hasn’t changed at all for people who booked their room over the phone, whereas that satisfaction in both years is lower for those who booked via the Internet. Therefore, the overall drop in satisfaction can be explained by an increase in the proportion of people booking online. This explanation, provided through the additional analysis, gives a lot more direction for trying to improve customer satisfaction than just looking at the overall mean would.
| Last Year | This Year | |
|---|---|---|
| Phone | Mean = 4.75 (n = 500) |
Mean = 4.75 (n = 301) |
| Internet | Mean = 3.85 (n = 500) |
Mean = 3.82 (n = 699) |
| Total | Mean = 4.3 (n = 1000) |
Mean = 4.1 (n = 1000) |