Decision Making Attributes Statistical Significance Tests


New Member
Hi All

So I am looking for some advice regarding the best test to use for the scenario below (the data itself is not normally distributed and the groups are mismatched in number):

A teacher ranks a series of attributes that impacts their decision to use specific teaching practices with their student. These include:

The practice meets the needs of my student (It did not meet the needs of my student)
I had the adequate training (I did not have the adequate training required)
I had access to adequate resources (I did not have adequate resources)
It was cost effective (It was not cost effective)
It was age appropriate for the class (It was not age appropriate practice for the class)
It aligned with the schools teaching culture (It did not align with the schools culture)
It aligned with my own teaching culture (It did not align with my own teaching culture)
It aligned with the parents instructional culture (It did not align with the parents...)
It was beneficial to all students in the class (It was bot beneficial for all students)
It was easy to integrate into the classroom (It was difficult to integrate...)

These are ranked from 1 (not at all important) to 5 (extremely important). The same teacher then ranks a series of attributes that impacts their decision to cease using these same specific teaching practices with students (see bracketed statements) on the same scale of 1-5.

There are 151 respondents. The respondents are then divided into two groups. Group 1 is Early Childhood, Group 2 is Elementary teachers. Examination of the means and SDs demonstrated that the reasons for adoption of practice and the reasons for cessation of practice were closely related. Therefore, if a teacher ranked that a practice must "meet the needs of their student" as highly important in their decision to adopt the practice, they also rated that if a practice "did not meet the needs of their students" as highly important in their decision to cease using that practice.

For Early childhood teachers the means and SDs for the reasons for adoption and cessation, were closely related. However for the Elementary Group, three attributes for ceasing use of a practice were ranked higher than their reason for adopting the practice. This suggests that an attribute they had considered as being less important, ended up being more important in their decision to cease the practice.

For example, elementary teachers rated "I had the adequate training" as slightly important but in their reasons to cease the practice, they rated "I didn't have the adequate training" as moderately important in their decision to stop using it.

So is there a test that could be utilised to determine if the difference was statistically significant rather than just an observation of difference? I had thought of using Spearman but that may only report on the highly correlated attributes rather than telling me if the difference between the non-correlated attributes is actually significant.

Hope that makes sense!




TS Contributor
So you have 10x2 statements, and each statement received an assessement on an ordinal scale (means and sds don't apply here, by the way, and the question of normality is none at all) by each participant. You could therefore perform pairwise comparisons of responses using the sign test.

With kind regards