These questions were designed to be the Reading related questions, combining questions that were previously used, ones that appeared to have face validity and ones that fellow practitioners also noted. The question I want to address here are:
- how do we compare and contract the questions
- should some of the questions be deleted
- what are the most predictive questions
The simple answer is that no matter which analysis you attempt, there will be a mixture of uncertainty, qualitative interpretation and questionable statistical validity. Or to put it another way, whatever I do, somebody will tell me I am wrong. So starting from that point, where do we go?
Firstly, I decided to review the questions and give the seven reading-related questions a ranking of up to five stars which is nothing more than my subjective statement of what looked good and what I would include in future questionnaires. And since it is a personal subjective view, it cannot be wrong, only (maybe) misguided. Above is my brief personal reflection, which ends up with two questions, namely
- How often do you confuse visually similar words, such as cat and cot?
- How difficult do you find reading aloud?
One of the problems within any analysis of this kind is that a Likert scale is being used. This means most statistical analyses are not valid, since, for example, we cannot know if the difference between each of these answers progresses in a linear way, or even if everybody was consistent in the “linearisation” of the answers. So comparing means or even looking at correlations is arguably suspect.
However, there are two analyses which I do intend to do in future.
- Do a table that plots the two 4-star questions against each other, such that it is not a correlation but contingency table, which may help understand the relationship (can we call it a pseudo-correlation?) between the two questions.
- Using the self-reporting, look at the level of Type 1 and Type 2 errors that would occur is these two questions were used. Do I mean only use two questions instead of the seven reading questions for that section, or do I mean use only two questions to identify dyslexic individuals? You will need to read the next entry to answer that one.