Database Reference
In-Depth Information
▪ We can be 95% sure that Pawnford fifth and sixth graders participate at higher rates than
at Kingston and Bishop Village, and we can be 95% sure that Castleborough fifth and
sixth graders participate at higher rates than at Bishop Village.
Lastly, we can't really combine these results to compare
all
students in each city, because the
sampling plan was somewhat haphazard. If President Fischer wanted to compare the overall
picture, he would need his statistician to create a stratified sampling plan that took into ac-
count the relative differences of students in each of the grade groupings.
Summary
Variation and uncertainty really matter when communicating data. It's easy to take the simple
route and just show averages or percentages, glossing over the complexity of the situation at
hand. But how much the data points vary, and how confident we can be in the proportions
can have a huge impact on the conclusions and decisions that follow.
The methods we've used in this chapter—the control chart and the binomial confidence in-
terval plot—don't come close to covering every situation, and the wise communicator of data
will research principles and techniques that may apply to the specific circumstances.
In the next chapter, we'll consider ways to communicate multiple quantities in the same
views.
