Dummy Regression Workfile Structure

[neptunhiker]

Hi everyone,

I have the following hypothetical dataset of avg. sunshine hours on weekdays in 10 different countries, i.e.:

• USA - Monday - 5 hours
  USA - Tuesday -6 hours
  USA - Wednesday - 4 hours
  USA - Thursday - 5.7 hours
  USA -Friday - 7 hours
  Canada - Monday - 4 hours
  Canada - Tuesday -4.5 hours
  ...
  ...
  ...
  Germany - Friday - 2 hours

Now, I want to find out whether there are signifikant differences between avg. sunshine hours on weekdays across countries. My approach is to do a dummy regression like the following:

sunshine.JPG (11.85 KiB) Viewed 10459 times

In order to so, I have set up my workfile as a dated/regular frequency workfile with the frequency option "daily 5 day week". For starting date I chose any arbitrary date starting on a monday, and as end date I chose a date which is 10 countries * 5 days/country = 100 weekday days after the start date. The reason I did this is that I can now easily run a dummy regression with Wednesday as the constant by the following command:

Code: Select all

ls sunshine c @expand(@weekday, @drop(3))

My question is, is that the correct way to do it? Or should I rather specify the workfile as unstructured/undated? But then I wouldn't know how to run the dummy regression.

Thanks for hints and suggestions.

[EViews Gareth]

Yes, that is one way to run that particular regression.

[neptunhiker]

Hi Gareth,

Thanks for the support. How could I run the regression, if the data were structured as unstructured/undated? I don't like the way the data is structured in my example, because the data are in reality not structured by a certain date range.

[EViews Gareth]

Have dummy variables for whether the observation is on a Monday, a Tuesday, etc....