Hello, I have made a test for student to scale their scientific attitude. I used 12 question or items. Maximum point for every question is 4 and minimum score is 1. But, I still hesitate about the range score to make a judgement for the high and low level. Any suggestion?
How to make a judgement or differentiate high and low students scientific attitude?
Whether to use range score 1-4, 1-5 or 1-10 etc. depending on your objective. Want to hi-light the following:
1. Higher the range score will yield the following benefits:
- E.g. range score 1-7 offers more variance than a lower range score e.g. 1-5
- Offers higher degree of measurement precision
- Provide better opportunity to detect changes and more power to explain a point of view
2. Range score 1-4 is even rating scale which offers no neutral option for respondent to select i.e. offer no fence sitter which is good in certain situations. But sometimes, certain respondents are really neutral on certain subjects whereby forcing them to answer a question with even rating scale will cause bias which does not truly represent their opinion.
There is no definitive way to divide scores into high and low categories.
One method is to use reasonable cut-offs. For example, in your case, you might call “high” a score that averaged 3 or greater on every question. So that would be 36–48. And then perhaps averaging lower than 2 on every question is “low”, so 12–23. And then 24–35 is “medium”.
A more statistical way is to use percentiles. The value for the 90th percentile is the score below which 90% of participants scored. Common break points for percentiles are: below 25th; 25th to 50th, 50th to 75th, 75th and greater. But you might want to use other percentiles.
Examples of the methods I mentioned are here:
http://rcompanion.org/handbook/E_05.html
Thank you. But, which better one, differentiate in 2 cases (high and low) or 3 cases (high, medium, low)? any suggestion?
For a statistical answer, in theory, you could use the clustering analysis described in the link I posted, but with one variable instead of two.
But I suspect, practically, you just need to make a decision whether you want 2 cases or 3 cases, and then decide how it makes the most sense to divide the data into these cases.
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude