During the GPS testing lab each team recorded the UTM coordinates 25 times at each of five points.We finished with 200 measurements for each of the test points. We combined all the measurements for each of the points and calculated descriptive statistics.
The difference of each point from the average coordinate was calculated using the Pythagorean theorem. Histograms of the differences show a big difference in the results. In point 1 there was a great deal of difference among the results of each team. The combined data shows the wide variation. It would probably be a good idea to remeasure this point. Since the teams all used the same model of GPS receiver and measured during the same time period, these results point to a likely user error.
For point 5 the differences are much smaller overall. They are also much more concentrated, indicating that all the readings were concentrated around the average, a consistent fix on the coordinates.
Note that these charts used all the data points.
During the GPS testing lab each team recorded the UTM coordinates 25 times at each of five points. After combining the readings I wanted to see if the results were statistically compatible or if they shouldn’t be compared. Another way to put it: are each team’s results form the same population or are there other factors that make the results not comparable.
There are some statistical methods to do this analysis. However, a simple way is to take the average of the eastings and northings, calculate the difference of each coordinate pair from the average, and plot the differences. The results should look like a scatter shot from shooting a target.
At point 5, the results seem to be randomly scattered around the average, or bullseye. The size of the spread is not very great.
Click on the thumbnail to see a larger version.
Point 1, on the other hand, shows the readings divided into distinct groups. Also, the magnitude of the differences is much greater.
All of the teams worked in the same time period, which means that GPS reception conditions must have been similar. The differences could be due to the differences between the GPS receiver units. More likely there was some variation in the operators or human error. I would recommend remeasuring this point to rule out human error.