I'm trying to get a distance matrix between my different datapoints in a bidimensional space (one variable in X axis, another in Y, simple). To get the distance forming the matrix, i will use euclidean distances, and, due to X and Y variable being in different scale/units, I want to standardize them (for example, using a rank between 0 and 1, or using the mean as 1, doesn't matter).
Until here, everything is okey. But when I'm about to request this normalization in SPSS, the programme asks me if I want to perform this standardization "by variable" or "by case". I don't understand the difference between this two procedures and the help in the programme doesn't look very helpful.
Kindest regards,
Quique
Hello Enrique,
You would elect to standardize by variable, not by case. Standardizing by variable (for example, via z-scores) would re-express each of your scores onto a common scale. There is good reason for wanting all variables on a common scale, and that is that your choice of Euclidean distance will otherwise weight more heavily variables with larger standard deviations than variables with smaller standard deviations in any computation.
Standardizing by case (sometimes referred to as ipsatizing the scores) re-expressed a case's score set to have a specific scale. That results in negative correlations among variables, for the simple reason that if a case has a relatively higher value on variable X, then the variable Y score will be relatively lower.
Good luck with your work!
Please refer to:
1. Landau, S. and Everitt, B. S.(2004). A Handbook of Statistical analyses using SPSS.
2.Howitt, D. and Cramer, D.(2008). Introduction to SPSS.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS