I have recently attended a presentation by a doctoral researcher who presented a structural model without the null hypothesis. Although I am a complete novice in structural models, and although I have never seen this before, it seemed a little strange. What are your expert views? Should there be a null hypothesis?
Structural Equation Modeling supprots or does not support that the data fit the model or not. You can easily use null and alternative hypotheses. Thus, fitting of data to the model is different from using hypotheses, but you can utilize both.
Why should that be strange. I your results show that you have not influence of different parameters to the result, you are working with the H zero. I think this could be a scientific approach.
Dear @Mohamed, don't believe in the null hypothesis, link is attached! "...For instance, in Structural Equation Modeling (SEM), when the resulting equations fail to specify a unique solution, the model is said to be untestable or unfalsifiable, because it is capable of perfectly fitting any data i.e. if a model is "always right" and there is no way to disprove it, this model is useless. A good hypothesis or a good model needs a high degree of specification...."
http://www.creative-wisdom.com/computer/sas/hypothesis.html
Dear Ljubomir,
you are right, but a model which is always working is the wrong model for the special case. And what you can prove is not to use such meaningless models for your task. That´s a scientific method.
Dear @Mohamad, the following article discusses this issue: "Introduction to Structural
Equation Modeling: Issues and Practical Considerations" by Pui-Wa Lei and Qiong Wu, The Pennsylvania State University. See the link:
http://ncme.org/linkservid/47EFEB5A-1320-5CAE-6EC90BFDF09AA39E/showMeta/0/
For fun I want to show a fairy tale story. Children are brought by storks. So you should prove to find a disconnection between number of storks and babies. Thats a H zero between these two facts. You know you won´t find any coherence. ;-))
Hypothesis testing and equation fitting are separate concepts that can lead to meaningless results, but have the same error. The error is improper specification. We can show that H1 is different from H0. The difference is meaningless if we have not specified what difference is important and at what error rate. SEM can use the H0 that the model does not fit the data and accept H1 that it does fit. The model fit is meaningless if there is lack of specification as to what the model must do at which error rate or how to use the parameters at which error rate.
The question is not "Can we reject the null?' it is "Can we trust or use H1?"
In its simplest form a null hypothesis is the lack of change from the equipoise position. The belief system that prevails before your study is initiated. So I have to ask, if you are not changing paradigm, what is the point of the research.
- Badges
- Science topic