Improving Instructional Data & Feedback: Sampling

TOOLS FOR IMPROVING INSTRUCTIONAL DATA & FEEDBACK

Sampling

Sampling to minimize data volume and maximizing feedback and information.

The teachers in the case study implemented a sampling strategy. Used appropriately, sampling is an effective way to manage process data while providing meaningful information. Sampling works in situations where you need to continually monitor a performance or process without always looking at the entire data set.

How Sampling Works

Sampling allows you to take a large amount of data and make it manageable. The basic premise is this: By taking the square root of your total and "sampling" that group of data or persons (at least once a week or once every two weeks) you can begin to see patterns and trends which can be generalized to the body of data.

As an example, lets say your curriculum at 4th grade requires that students know the 50 states, the major cities and major landforms in the United States. Together, it totals 100 locations.

How to Sample

Random.org

You begin by creating a baseline by testing the 4th grade on all 100 locations and recording the results for each student in a spreadsheet or database. Since the square root of 100 is 10, you set up a process for randomly "sampling" 10 of the 100 locations once a week.
At the prescribed day or time, the instructor randomly draws 10 locations out of a fishbowl or logs onto Random.org and generates a random set of 10 numbers which correspond to each location.
Students then identify the randomly drawn locations on an 8½ by 11 map of the United States. The maps are then corrected and the score for each student recorded on a spreadsheet.
Create a line graph displaying the data. After a few weeks the data can begin to be analyzed, looking for patterns and trends. Is the number of correct responses going up or down? Why?
Students should be encouraged to graph their own results. Many teachers post a class run chart on the wall to encourage the class to work together to improve.
If a student scores seven perfect scores in a row there is a 99.9% probability that the student knows all 100 locations and can be "validated" as knowing them all. If they get a perfect score once it's 50/50 that they just got lucky- an easy set of locations to identify, etc. Twice in a row and it’s now 75% probable that they know them, a 3rd time it's 87% and so on. Seven times is a statistically significant benchmark.
Teachers can do an item analysis to identify the recurring errors which helps them guide their instruction.

Tools for Improving Instructional Data & Feedback