I selected 6 different brands to check their authenticity. 6 brands will be coded into dummy variable. Now, I want to check the authenticity of those brands. As, my dependent variable is authenticity (Data collected for authenticity was ordinal using 7 Likert Scale). Can such regression be tested? Urgent Response is needed. Thank You.
Do you mean 6 brands will be coded into 5 dummy variables?
This would just test differences among brands, so different from you wanting "to check authenticity of those brands." For this, how do you define authentic? For example, is this 73% giving a 6 or 7 on the scale, and you want to have this as the null hypothesis to test? You might want to edit your question to make it clearer by saying more precisely what you want to check. (and of course you should do this before collecting data in case an alternative design is better suited).
How is your dependent variable actually measured? If it is a single, Likert scored variable, then you should use a non-parametric test, such as Kruskal-Wallis. If you have multiple Likert-scored items that you can combine into a scale, then you can use Analysis of Variance (which is statistically identical to dummy variable regression).
David L Morgan and Wolfgang Schoppek , he is not saying that we wants to compare the authenticity of the brands. What David is describing might (or might not) be appropriate if this were the case, but I assume Wali Muhammad Khoso would have asked about that if that were his goal.
Daniel Wright My goal is to know if my chosen brands are authenic. I have two categories of dummy variable. One is, which brands are authentic and in second step, those brands are in category of local and global. Brand authenticiy which is my dependent variable is constructed with 15 items. Actually i am new to this field thats why i dont know how to ask question in more technical way.
David L Morgan Its multiple likert scored items. There are 15 items for brand authenticity.
With 15 variables, you should use Exploratory Factor Analysis to find out if they all measure the same underlying construct. I recommend that you use an extraction with Principal Factors and an oblique rotation.
Confirmatory Factor Analysis would also be an option for testing the homogeneity of the authenticity scale. Or at least a plain Reliability Analysis.
I think regression does not look possible. Instead of regression, you can compare the brands according to authenticity. It is more rational.
Wali Muhammad Khoso , you are saying you want to say which (if any) are authentic. This sounds like things either or are not authentic. Is this correct? Do you have this as a variable (in methods terms a "gold standard")? Or if you are trying to use some scale, whether composed of several variables or just one, do you have a cut-off to say what you consider "seems authentic"? Or are you just trying to look at variability in authenticity, which is a different question (and if it is this I'd recommend removing this question and asking a new one)?
Daniel Wright I think this was a less than crystal clear question about a relatively straightforward design: are there variations in authenticity (as a dependent variable) across six different brands (as a set of independent variables). That is way I recommended ANOVA -- although dummy variable regression would of why yield the same result.
David L Morgan , don't you think Wali Muhammad Khoso would have said that in his question? It seems a clear question, are these authentic or not. He doesn't say in the question about comparing the brands. Whether the design is appropriate for that question or not (and it likely a not) may mean the questioner will design a new study to address this or at least be able to talk about cut-off values. I worry making up the research questions of people to fit a particular statistic. The research questions that people have are important, so I worry about people guessing these (or in this case changing them) since the person may come away expecting answers to their question. Do you think ANOVA will answer the question actually asked?
The question is obviuosly not clear as we usually see in RG. Providing a sample of data and/or some more info is inevitable.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression