My thesis is about relationship between voluntary disclosure and firm value. to measure voluntary disclosure, I use a weighted index that takes into account importance or usefulness of disclosure items which can expressed as follows:
WDIj=total(di*wi)/w
voluntary disclosure index for company j is equal to total of disclosed items multiplied by their own weights that is divided to the sum of weights.
To do so, I made a survey questionnaire to ask about importance of each voluntary disclosure items on a 5-pont scale. Now, I want to know, should I use simple average of responses about each item in the formula, or standardize them to a zero mean and unit variance?
Hello Mahdieh,
First, I presume that, from your survey instrument, you will average all judgments of importance for a given item/disclosure type across respondents. Thus, some items will have higher means and others will have lower means (in all likelihood), and higher-mean items would be characterized as more important.
If that is the case, then it won't matter whether you use the sample mean of all item ratings or linearly transform them to have an arbitrary mean of zero as regards the resultant rankings of company disclosure index. Now, it can make a difference in rankings if the items have different variance/SD. That is, items will have influence in the final rankings proportional to their SD. So, if you want all items to have the same actual weight/influence, then rescale them all to have the same SD.
All of which is to say that standardizing won't hurt, and may help.
Good luck with your work.
- Badges
- Science topic
- Similar topics
- Indexes