Hello Shanjida,

What's the intended purpose for having the categories? Will these be more useful than the actual scores (since you're abandoning information by collapsing scores into bins/categories)? That should be the first consideration

Here are some common methods for creating categories:

1. Use existing definitions, where relevant. For example, in the USA, there are accepted definitions for income such as "below the poverty line," "above the poverty line." Alternatively, tax rates increase by income level (0%, 15%, 18%, etc.).

2. Consult domain experts. For example, there may be score ranges that are considered clinically elevated, as in the case of scores on a depression scale.

3. Use norms. You might define "high" as scores that are at least one SD above the mean, for example. This works best when you have either the population or a sample that you are confident is comparable to the population.

4. Choose thresholds to yield equal proportions of cases (but not equal bin/interval score lengths).

5. Choose thresholds to yield equal bin/interval score lengths (but not equal proportions).

6. Choose thresholds that maximize score differences on some external variable (this is not unlike AUC/ROC analysis, where the goal is to maximize some function of sensitivity/specificity).

One concern in methods such as #3 and beyond, is that these can be unique to your sample (e.g., generalizability might be an issue).

Good luck with your work.

David Morse thanks a lot ...its so helpful

For a numerical data, it is relatively easy to decide the three categories namely low, medium and high. For example, based on the information on Income, first find the suitable frequency distribution. Then, decide the 3 categories namely low - below 33 percentiles; Medium: 34-66 percentiles; High: Above 66.

For a qualitative data or for data dealing with categories, decide the score for each attribute. Say, for land ownership (below 2 acres; 2-5 acres; above 5 acres), for, Cattle possession ( below 20; 20-50 and above 50). Now for each attribute, you can give the score as 1,2 and 3. We are presuming that these classifications are valid and good enough for the research. Based on the sum of the scores, we will get the scores from 2 to 9. Note for both lows, the sum of score will be 2, for both Mediums it will be 4 and for both Highs, it will be 9. Now, the final classification will be as follows: 2-3-Low; 4-7 - Medium and above 7 as High.

thanks Sir Ramnath Takiar

Sorry for this too late response. Yes you do may use it. That is the write Formula. My encouragement