Saturday, May 20, 2017

Week 23 - Interpreting and Using Evidence

Gray (2012) offers the following pieces of advice about data:
Data is not a force unto itself. It is what individuals (or groups) do with data that brings meaning and power.
Data is not a perfect reflection of the world. The choices we make about data, including what we choose to collect, how we collect it, how we analyse it and how we interpret our analysis all influence the findings and conclusions we can make.
Data does not speak for itself. It requires interpretation and analysis (hopefully by knowledgeable individuals). In education especially, it needs to be understood within its particular context. This might be in relation to a particular class, school, community etc.
Interpreting data is not easy. Really understanding what the data is telling us can be very tricky.

Despite these cautions about data, the evidence you collect as part of your teaching inquiry can be very useful.
Tools for statistical analysis of the data
A calculator on this web page (http://www.surveysystem.com/sscalc.htm) can help you to determine the appropriate sample size for a survey (e.g. how many responses you should collect) to have the result within a certain margin of error (which is also called the Confidence Interval) and for a given population (e.g the total number of students in your school). The web page also provides a calculator to find the margin of error of a particular survey result from a given population. For both calculators, you can choose how sure you would like to be about the result from a statistical perspective (which is called the Confidence Level).
For example, assume your school has 1,000 students. If you ask 200 students a question and 80 of them give the same answer (40%) then a 10% confidence interval means that when you ask the whole school the same question, between 300 (30% of 1000) and 500 students (50% of 1000) would pick that answer.
Confidence Level of 95% means there is a 95% chance that the real number of students picking that answer is between 300 and 500.
Confidence Level of 99% means there is a 99% chance that the real number of students picking that answer is between 300 and 500, i.e. to get more accurate results you need to ask more people
When you use the tool “Determine Sample Size” with the population of 1,000 and Confidence Level of 95%, it will tell you to ask least 88 students the question to obtain the required Confidence Interval of 10%.
The sample (e.g 200 students to be asked the question) should also be randomly chosen.
If the actual number of students giving the same answer is 120, using the second calculator with 95% confidence level, you can say that there is a 95% chance that the margin of error of your result is 8.4%.
Overall, the lower your margin of error, and the higher your confidence level, the more reliable your results will be and the more useful they will be for helping you to make decisions.

No comments:

Post a Comment