I spent a small portion of the weekend focusing on evaluating student scores on the predictive assessment provided by Thinklink. The scores provided indicated that there was a large amount of students who were going to score in achievment level 3 and 2. A graph of individual strand breakdown is provided here:
Reducing the data by performing a principle components analysis for these 5 strands into 2 variables accounted for 80% of the variance in data. Performing a Quadratic Discriminant Analysis and Linear Discriminant Analysis in my earlier research has shown to be much more effective in correct predictions of students future standardized achievment. Here is an image of predictions using the QDA and LDA on the reduced data.
Using this type of analysis, the error in prediction is halved. The QDA and LDA both report that two students are in danger of scoring a level 3 in achievement this upcoming year. These two students were the two lowest scaled scores from the class which shows that there just needs to be some modification in color schemeing by the predicitve testing agency. This analysis will really help me concentrate my efforts on students who are in the most danger of scoring in a lower proficiency level.