Null Hypothesis

[statnewbie76]

I am doing a research study and I am collecting data from Hotels. I am confident that the business models of the hotels ARE NOT suited for today's networked economy (beginning premise). What would my Null hypothesis be?
Ho1: The business models in the hotel industry are suited for the current economy OR
Ho2: The business models in the hotel industry are NOT suited for the current economy.

I understand that NULL hypothesis is the devils advocate so technically the Ho2 should be the NULL hypothesis.....but as I mentioned - my expectation to begin with was that Business Models ARE NOT suited for the networked economy,....so the devil's advocate in this case would be that the hotel's Business Models ARE SUITED for the networked economy.

Here is the rub - my data proves me correct. From a sample of 212 hotels majority of the hotels business models ARE NOT suited for the networked economy. The P value of the majority of the tests are BELOW .05 i.e. P<.05 hence we REJECT the NULL Hypothesis. But if the NULL Hypothesis is Ho2- than I am rejecting What the Data is saying. Hence, can Ho1 be the NULL hypothesis. Please review what I have stated carefully. The obvious choice may seem Ho2 - but that would throw everything into a quandary. Any help here is appreciated very much.

Thank you.

 

[Junes]

Hi there,

Before we can answer your question, we would need to know more about your study. What do you mean by "the business models in the hotel industry are suited for the current economy"? That's a bit too vague for statistical testing. What did you measure? What was your research design? What were the independent and dependent variables? What kind of test did you use?

Generally speaking, null hypothesis testing starts with the null hypothesis of no effect or something being equal to 0. Then, if there's enough evidence against the null hypothesis, we reject it. But there are alternatives, depending on what you want to do.

 

[statnewbie76]

Hi Junes thank you for the post. There are three things I am evaluating about hotels to assess whether they are suited for the networked economy or not. First, how intensely are they using information, particularly internet technologies. So I have asked them a series of questions on services and processes supported by information technology or not. Secondly, I am assessing partnerships of hotels. How many partnerships do they have with other hotels. Hotels that are most well suited for the networked economy are those that have several partnerships and jointly they offer value proposition to potential customers. Finally, I am asking whether they have a strategy to use their customer networks. Based on how they have answered these questions, I can than assess whether they are well suited for the networked economy or not.
The first test I am using is a simple Binomial test using .7 as the test proportion. Example, I am asking a question on services supported by online internet technologies. As answers multiple service options are given i.e. hotel reservations, payments, etc. If options are selected it indicates the hotels are using internet technologies to support their business activities. If the services are not selected by a hotel it means those services are not supported by the internet technologies. Binomial Test measures whether a distribution of values is binomially distributed. We set our test proportions to .70. Hence, if at least 70% or more of the hotels have checked ‘selected’ than it is an indication that the hotels are using internet technologies to support their services. Hope this helps in answering your question.

 

[Junes]

I'm not sure hypothesis testing is the right approach in your situation. The .7 that you set seems kind of arbitrary. What is your rationale for .7? Why not .65 or .75?

Maybe you could just report your results descriptively, with confidence intervals around the survey answers. You can then comment on the level of 'suitedness' irrespective of some seemingly arbitrary threshold.

 

[Buckeye]

So, what you're saying is you have some p values and you want to tailor your hypothesis to fit them? At least that's what it sounds like.

 

[statnewbie76]

So, what you're saying is you have some p values and you want to tailor your hypothesis to fit them? At least that's what it sounds like.

No I do not want to tailor my hypothesis. i simply want to know what will be the NULL hypothesis? kindly review my posts in detail.

 

[statnewbie76]

Hi junes, I appreciate your reply very much. But I am sticking with Binomial test - and why .7? because 70% is a good benchmark. yes it is arbitary and there is no real reason why. But again I want to know can my NULL hypothesis be Ho1? If that is the case - i am in safe grounds. but If Ho2 is the null hypothesis - than I am not sure where I stand?