topic Re: Cross-country analysis in Statistical Procedures

Cross-country analysis

Miracle — Wed, 18 May 2016 06:18:02 GMT

Dear Sir or Madam,

How are you?

I am writing to seek your advice on the appropriate statistical analysis to my research question about the differences between 5 countries on consumption measures(in millilitres) which are believed to be followed a Gamma distribution in relation to a risk assessment profile score(consisted of 4 questions(Yes/No) and services use(Yes/No). There are a few thousands of data from each country.

Any insight would be greatly appreciated.

Re: Cross-country analysis

Norman21 — Wed, 18 May 2016 06:21:40 GMT

Dear wong,

This is a good question, but you should first answer another question: What is your hypothesis?

You can then determine which statistical test is suitable to test this hypothesis.

Re: Cross-country analysis

Miracle — Wed, 18 May 2016 11:22:49 GMT

Hi Norman.

Thank you for your reply.

Is this a correct null hypothesis?
There is no differences between countries on the consumption measure in relation to the score.

Thank you very much.

Re: Cross-country analysis

Norman21 — Thu, 19 May 2016 02:58:26 GMT

Hi wong,

That is a good start, but is it complete? Don't you want to add something about other factors? And when you say "differences between countries", do you mean on average over the past ten years, or in the past year? Do you want to calculate consumption per capita? Or restrict analysis to 18-24 year old males?

It is best practice to be as specific as possible before you run a statistical analysis.

Re: Cross-country analysis

Miracle — Thu, 19 May 2016 03:10:27 GMT

Hi Norman.

Thank you for your reply again.

The data I am going to analyse is the first wave of the study where we only focus on the 16-65 years old.

Ultimately, I will have to adjust for other (categorical/binary/continuos) explanatory variables i.e. gender, age, race and etc.

In regards to the consumption per capita, it is beyond my understanding capability and I don't think I need to worry about it at this stage.

Thank you very much.

Re: Cross-country analysis

Norman21 — Thu, 19 May 2016 03:15:28 GMT

Hi wong,

So, are you expecting an answer along the lines of "country A had a higher consumptions than countries B, C or D"?

But what if country A has a larger population than the other countries? Surely you have to take that into account?

Re: Cross-country analysis

Miracle — Thu, 19 May 2016 03:34:13 GMT

Hi Norman.

Thank you for your quick reply..

Yes, at this stage I am just interested to know if there are any differences between the countries in consumption in relation to the score based on our sample only. If there are, then how different and in which way etc.

Sorry but I don't understand how the population number across countries going to affect my current analysis if we just look at the sample collected. Perhaps you could shed some light on this?

Thank you very much.

Re: Cross-country analysis

Norman21 — Thu, 19 May 2016 03:42:37 GMT

Hi wong,

Are you saying that your data have come from a survey of people from different countries?

Or, are you obtaining data from an international database, such as this: http://faostat.fao.org/site/567/DesktopDefault.aspx#ancor

Or perhaps a mixture of the two?

Re: Cross-country analysis

Miracle — Thu, 19 May 2016 03:45:49 GMT

Hi Norman.

Thank you for your quick reply again.

It was a survey-based study from each country.

Thank you very much.

Re: Cross-country analysis

Norman21 — Thu, 19 May 2016 03:48:54 GMT

Hi wong,

OK. It would have been helpful to say that in the first post!

Analysis of survey data depends on how the data were collected. Here is a link to a site that has much information, together with examples of SAS code and output:http://www.ats.ucla.edu/stat/sas/topics/survey.htm

Re: Cross-country analysis

Miracle — Thu, 19 May 2016 03:59:14 GMT

Hi Norman.

My apology for not making my problem clearer in the first place.

Thank you for the link.