I need to analyze different factors affecting the working capital of a company. I created a questionnaire with 14 different factors and it is a Likert scale question.
For example:
Q1) What is the impact of size of the company on working capital on the scale of 1 to 10?
Q2) What is the impact of inaccurate scoring of customers on working capital on the scale of 1 to 10? and so on..
I want to know which factor has the most and the least affect.
Can anyone guide me on how to conduct an analysis for such question? I have done an analysis using median, where I have calculated medians of responses for each factor and gave ranks to them. But, this is too simple and mainstream for a master's thesis. Do let me know which analysis should I use?
Do you want to see how each is associated with working capital. If so, for each variables, make a scatter plot with each each of the 14 scales on the x-axis and working capital on the y-axis. You may need to jitter scores. The choice of any statistical procedure to measure each of these associations would be influences by these plots.
If your goal is to see how each of these affects workings capital, that would be difficult. See Pearl ( http://bayes.cs.ucla.edu/BOOK-2K/ ) and Imbens and Rubin ( Book Causal Inference for Statistics, Social and Biomedical Scien...
) and others on trying reach causal conclusions in these circumstances .
Try a model of structural equations where these indicators are predictive of working capital
Aliq Singh, I think your question needs to be clearer in some respects before people can be really helpful. When you say you have 14 factors, do you really mean "factor" in the conventional sense of a collection of items that are all interrelated, or do you mean 14 items, each of which is responded to somewhere on a 10-point range of options?
Greetings!
You could analyse the Likert scale in interval level (summatted Likert scale) rather than at ordinal level.
Irrespective of numeric rating scales (NRS), Verbal descriptive scales (VDS) or VAS you must use rank-invariant methods for description and analysis in order to obtaining valid and interpreatable results. It is possible to calculate sums and differences but it is wrong to treat ordered categorical data as if they were laboratory data with the properties of standard magnitude and equi-distance- such mis-use is very common.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Applied Statistics
- Survey Design
- Questionnaire