This is a stat question, not an SPSS question, but I know
this group has the right collective minds to help me.
I have been reviewing some web-based reports for a data
collection and reporting company that works with a number of After-School
programs around the U.S.
Built into their collection system is a web-based survey that students at
participating sites complete on a quarterly basis designed to measure ‘attitudes
toward school’. I should state up front that I have no knowledge as
of yet whether reliability or validity analysis have been done.
They have generate a site-level report that shows the mean ‘scores’
on this survey across the course of a year. In addition, they have the
same type of graph for individual students, showing their ‘scores on the
survey. I would like to be able to suggest to them to flag students with
scores falling outside the expected fluctuation we normally see in surveys or
assessments so that service providers and site managers can quickly identify
students that may need intervention, etc.
My initial idea is to calculate the change scores (Q2-Q1)
for all students and use the distribution of these changes scores to create a
band or interval around the mean. Subsequently, any individual score
outside that band could be flagged for attention. I was thinking of using
the SD of the Change Score distribution to do this, but I’m unsure.
Any help or insights would be great.
Hope this makes sense,
Jason
Locked
