You should add more context. Index for what? And created using which variables?

Jokin Ezenarro , Let's take MCA for example. I am creating an asset index based on 6 binary variables. For each of the six variables, we assign 1 if a person owns that particular asset.

So using STATA MCA, we have

MCA var1 var2 var3 var4 var5 var6. (equal weighting)

//generate the index

predict asset

br asset

//check the mean

mean of asset

some of the values are negative

It is a bit more complex in MCA, as you're creating a continuous variable from categorical variables, and the interpretability is more complex. But if the variables can either be -1 or 1, there is no problem in your index being negative.

Of course, you need to take into account how that index is generated and what are you using it for, in order to properly interpret the negative meaning.

Thank you Jokin Ezenarro . I am creating an asset index based on owning five basic assets. This is to help have an index representing asset levels.

For a basic example, if I sum the assets owned by an individual one then divide by 5, I will have an index.

Example for respondent A

Asset ownership assignment

car: Yes 1

bike: No 0

tv: Yes 1

stove: Yes 1

Fridge No 0

A basic asset level will be 1+0+1+1+0 = 3/5

= 0.6

In this case, the index is 0.6, showing the asset levels.

But if I use MCA, I may have a negative value like -0.05

This is my issue. after mca variables the predict command is that all there is in generating an index using MCA?

Are you using mean centering or some other sort of data centering?