Dear colleagues,
The VAS is a commonly used tool in health outcome studies, when using it to assess importance of certain action or intervention, how we can interpret the results, for example on 1 to 10 line (where 1; the least and the 10; the highest), the average of importance was 4 out of 10, what does that mean? and if we want to identify the level of importance (very important, important, somewhat important, not important), where are the cut off point for each level? to say for example, one third of student view this intervention is very important.
Can we interpret the result as 1-5 is not important and 6 to 10 is important.
Thank you.
Dear Dr. Izzeddin please find the following study
http://www.ncbi.nlm.nih.gov/pubmed/14622683
I use a 1- 100 cm line; using the same ruler and one appraiser for visual consistency. If it is a pre-post test or survey, we can then run either paired (same subject) or individual t-test. 1- 100 provides for better statistical analyses.
You ask "Can we interpret the result as 1-5 is not important and 6 to 10 is important"
Short answer is emphatically NO
Your question refers to a VAS scale bounded by 1 and 10 but subdivided into difference response categories. This is more like a categorical rating scale than a VAS scale which generates a quasi-continuous "score". In its original form VAS scales required the researcher to measure the point where a responded physically marks the line. Hence the "length" of the response might be measured in units (cm). Some early VAS scales included text descriptions for the endpoints - others did/do not.
The classical 5-point rating scale EVGGFP provides a categorical rating of self-assessed health status. The "conversion" of the categories into a numeric value (say 1- 5) is completely lacking in legitimacy since it imposes an equal weight to the difference between categories and implies that "4" is twice the value of "2". This is a widespread practice but in the specific context of self-reported health status it is frankly unforgivable.
Your choice of analysis really should be determined by the measurement properties of the rating scale - in general therefore non-parametric methods. There are well documented methods available too for the processing of such ratings to derive a cardinal value based on aggregated responses in study populations. Such an index could then be used to rescale that original responses so as to generate a variable that could be used in parametric analysis
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Graphs