There is a contention regarding the scale of measurement for rating scales. Is data obtained through them amenable to multiple regression analysis? What assumptions must be tested before performing MR on such data?
We have three kinds of measures or scales- nominal, ordinal and interval/ratio. Generally, we use ratio/interval values in simple linear regression and in multiple regressions. However, even qualitative measures, can be applied with multiple regression provided they are measured with dummy numerical measures. Data from a rating scale (especially continuous rating scale) are ratio/interval values.
Hi,
Your question: Can multiple regression be performed on data obtained using rating scales?
If you data related to rating (ratio) scale, you can use the multiple regression analysis procedure and find out the influencing factors from that. We can use in multiple regression analysis when the data are in the form of nominal.
Suppose, Logistic regression can be divided into to two types(9a). binomial (b). multinomial.
1. Binomial or binary logistic regression deals with situations in which the observed outcome for a dependent variable can have only two possible types.
For example: 1. Yes 2. No. That is, in binary logistic regression, the outcome is usually coded as yes = 1 and no = 0
2. Multinomial logistic regression deals with situations where the outcome can have three or more possible types
For example: we would compare between three as follows:
1. disease A 2. disease B 3. disease C.
If your data, related to ordinal then you can go to Multiple ordinal logistic regression analysis.
Analysts typically use linear regression for rating scales (0-10) and ordinal logistic regression for Likert scales. However, neither of these are satisfactory and principled, although practically speaking the results may be more or less useful. If you want to take a principled approach, you may want to take a look at this paper: "Inflated Discrete Beta Regression Models for Likert and Discrete Rating Scale Outcomes", Cedric Taverne & Philippe Lambert. It is available for download on arxiv.org.
Thank you Senthilvel Vasudevan for your detailed and insightful explanation.
Thank you Ravi for the practical view and sharing the link to the publication.
What possible test i may perform when my study is parametric and both my variables have Likert scales responses that are categorized into 3. My study is about employability indices relationship with profitability performance indices in universal, commercial and thrift banks. I did multiple regression but I do not know how to find x and y coming the results of Minitab. Please help anyone
Hi
1. Yes.
2. Regarding the theory of multiple regression analysis, please refer to:
Richard Johnson and Dearn Wichern (2007). Applied Multivariate Statistical Analysis. Chapter 4, pages: 149-208.
.Joseph Hair, Wuilliam Black, Barry Babin, Rolph Anderson and Ronald Tatham (2006).Multivariate Data Analysis. Chapter 4, pages: 237-243.
Norman R. Draper, Harry Smith (1998). Applied Regression Analysis, 3rd Edition, Wiley.
3.. The applications/calculations of multiple regression analysis can be done using SAS. Please see, Willam Mendenhall and Terry Sincich (2003).Chapter 4, pages:182-185.
4. In SPSS, the calculations can be also developed . See,
Landau, S. and Everitt, B. S.(2004). A Handbook of Statistical analyses using SPSS. Chapter 4, pages:92-126.
Regards,
Zuhair
- Badges
- Science topic
- Similar topics
- Medicine
- Physiology
- Biosignals
- Biomedical Signal Processing
- Statistical Analysis