Saturday, November 19, 2011

A school principle wants to survey students to determine whether having internet access at home improves grade

1.) what type of sample would you suggest


2.) describe a technique for choosing the sample





i need help..... thank you

A school principle wants to survey students to determine whether having internet access at home improves grade
Does the principal have a purpose in mind the answer to the question, "Do students having internet access at home get better (or worse) grades than students without internet access at home get?"





Suppose that the principal believes that grades reflect learning, and he wants an indication of whether he can encourage learning by encouraging students' guardians to provide or deny home internet access to their wards.





Why survey students? The obvious reason is that students are in school and therefore accessible to be surveyed. But they still have to asked, and their answers have then to be tabulated correlated with their grades. But internet access is not a yes or no question. How available is the internet? How convenient is the connection and what is its speed? Dial-up, DSL, or cable? Is there competition from siblings and/or parents for computer time? What about chores, homework, sports, etc.? If a student lives with other students, then the student might be asked to provide information about the internet's home accessibility for those others, as well as for himself.





If the principal wants an indication of how he should guide guardians with whom he speaks about their wards, perhaps he should be surveying those guardians for the accessibility data. A "random" sample of the student body may not be what would be most useful. He could also ask those students and any other of his students sharing the same guardian(s) and compare their answers to check reliability.





But suppose the principal decides to use a representative sample of his entire student body to estimate the correlation of home internet access to grade (improvement) for the whole student body. The problem of sample selection is then how to select a sample that will be representative. Most anyone will tell you that the sample should be "random," but random surveying is more difficult than some people suppose it to be.





The principal could be tempted to go to three of his teachers and ask them to survey all of their students about their internet access, or all of their students of some randomly chosen periods... But it's doubtful that such an approach would yield a random sample. The principal's choice of teachers will be biased in some way. Perhaps he'll exclude coaches, because students don't carry pens to P.E. Perhaps he'll exclude art teachers, because students aren't "serious" in art class.





I suggest that the survey be given to every student. Practically, this means that the whole student body be surveyed at the same time. School absences are not random phenomena, so for rigorously valid results, absent students should be surveyed on their return to school.





If the sample size is to be reduced and remain representative, then some surveys should be selected/eliminated "randomly." For example, if IDs were assigned haphazardly, then the students can be sorted by ID and the survey of every seventh student by this sequence could be considered for the statistical study. Those implementing such a procedure ought to ask whether the effort required to eliminate part of the sample is justified by the reduction in effort required to process the remaining surveys. The answer will depend on details of the process which I cannot evaluate from this distance.





Imagine if only every seventh student by ID were selected to be surveyed. How would the principal collect his data? Would these students be called to a special assembly and interrogated there? If so, their grades might fall because of the classroom time they missed. Surveying the entire population is costly but eliminates sampling error.


No comments:

Post a Comment