Hello Prafula,

Unfortunately, I don't fully understand your data set and the intended research question/s you're trying to address given the query as posted.

So, if you could kindly elaborate and make more clear what you're doing, I suspect you would get more focused recommendations.

The term, "normalized," is used in several ways (e.g., recasting an observed distribution to match the normal distribution; rescaling scores such that the sum of their squared values equals some arbitrary vector length; and others), so the intention (and rationale for) invoking normalization isn't apparent.

Second, there are numerous ways to combine scores from multiple variables into a single index. The way that makes the most sense depends not only on your intended use(s) for the index, but the nature of the variables, their quantification and, for metric scores, the variance-covariance matrix of the target variables.

Good luck with your work.

Hi,

You can probably form an index using normalized Likert scores and divide it using the mean or median. In Stata, you can utilize the 'c.' command to treat this index as a continuous variable for interactions. To examine individual effects, you may create a new interaction term using only the fourth dummy from each of your four Likert variables, allowing you to explore how the highest ratings jointly affect the dependent variable.

Hope this helps.