Physician burnout is a serious problem and you have the impression that morale among the physicians at your facility is at a low point. You administer a survey to get some quantitative data, as well as some suggestions for improvement. One of the most common suggestions is that hiring physician extenders (PAs, scribes, or other extenders) would improve things. Your supervisor gives you permission to hire some physician extenders, but wants to know if the investment achieves significant results. You decide to administer a second survey six months following the hiring of physician extenders and will do a t-test to see if there’s a significant change in the survey results compared to the results from the survey prior to the hiring.
Here is the data, from the two surveys:
Physician number |
Survey #1 average score (out of 10) |
Survey #2 average score (out of 10) |
01 |
7.2 |
7.5 |
02 |
6.5 |
6.7 |
03 |
5.5 |
4.8 |
04 |
5.5 |
6.4 |
05 |
6.8 |
7.4 |
06 |
8.0 |
9.0 |
07 |
8.3 |
7.8 |
08 |
4.3 |
5.5 |
09 |
7.1 |
8.0 |
10 |
6.6 |
7.1 |
11 |
4.5 |
5.0 |
12 |
7.9 |
7.4 |
13 |
4.6 |
5.3 |
14 |
6.5 |
6.0 |
15 |
5.8 |
6.5 |
This is an example of using a paired samples (repeated samples) t-test. In Excel, the “type” parameter for the paired samples t-test is ‘1’. Since we don’t know if the new hiring will improve or worsen the survey results (you never know … the new hires could make things more difficult), this will be a two-tailed t-test. The “tails” parameter in Excel is ‘2’.
Note: The result that Excel displays when you run the t-test is the p-value. You don’t have to do anything else. It may just be a single value that Excel displays, or it may appear in a table of values and labeled ‘P (T<=t) two tail’. If Excel displays a table of values it also will include the t-statistic and the t-critical value. You don’t have to go to a table of critical values to find it.
t-Test: Paired Two Sample for Means | ||
Survey#1 | Survey#2 | |
Mean | 6.34 | 6.693333333 |
Variance | 1.639714 | 1.46352381 |
Observations | 15 | 15 |
Pearson Correlation | 0.877182 | |
Hypothesized Mean Difference | 0 | |
df | 14 | |
t Stat | -2.20396 | |
P(T<=t) one-tail | 0.022383 | |
t Critical one-tail | 1.76131 | |
P(T<=t) two-tail | 0.044766 | |
t Critical two-tail | 2.144787 |
b) Yes, there is a significant change
c) we have enough evidence to conclude that there is a significant change in survey
Get Answers For Free
Most questions answered within 1 hours.