Hello, How do I code the data captured using Semantic Differential scale instrument so that I use it to find the mean and p-value in SPSS? I am comparing two products A and B. For example, the instrument has two bipolar scales such as: ugly o o o o o o beautiful and users are supposed to select any circle in between the two adjectives. Kindly advise

hi, the circles are usually numbered, so you have put 6 circles, it would be scored 1 2 3 4 5 6 and then your user chooses say, the nearest circle to ugly, that would be scored as 1. To find the mean, use descriptive statistics.

To find p-value, first determine what kind of test you want to use.

hope this helps
love satchi

It is pretty meaningless to report mean for this type of data.

Better approach would be to report median or frequency.

Both the mean and median have their flaws. Anyway in a table I would expect to see mean, 95% confidence interval, median, and range.

This is simply an ordinal scale, to which we are going to allocate arbitrary numbers. The difference between ugly point one and ugly point two is subjective.

Mean is hugely flawed in this context (reporting a 95% confidence interval would be even more ridiculous).

How is it useful in this situation to report a mean of, for example, 3.7 (95% CI 2.6-4.2)? It is meaningless.

Median (interquartile range) is still flawed, but gives you an overview of the data without making assumptions about the importance of the interval between data points.

Median and interquartile range I suppose I could live with, but ultimately I do want some idea of the spread of the data. Maybe even as far as histograms.

Whilst it might be subjective, if 60% choose option 1 but 40% choose option 6, this needs to be reflected in the reporting (granted the mean here would also give something 'off', since it would report an answer nobody would report).

Yup- so back to my original post.

Report as median (with interquartile range) and the frequency (percentage) in each category!