Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Regressing ranked variables for causality testing

From   Nick Cox <>
To   "" <>
Subject   Re: st: Regressing ranked variables for causality testing
Date   Tue, 23 Apr 2013 14:13:07 +0100

You don't seem to have read the Statalist FAQ before posting, as you
were asked to do.

Holmes and Hutton (1990a, 1990b). It's possible that someone
recognises those references, but you are asked not to use minimal date
(year) references.

Equations. That doesn't qualify as Plain Text. The equations you
posted are readable at the StataCorp archive

but not in the corresponding Harvard archive entry. You are assuming
your notation is transparent.

. findit holmes hutton

yields nothing by way of Stata programs, but they may not be the best keywords.

That's as far as I can help. Other people on the list know immensely
more about this area.


On 23 April 2013 13:48, Happy Phiri <> wrote:

> I want to use Multiple Rank F Test as developed by Holmes and Hutton (1990a, 1990b).  To perform the multiple rank test, all the variables
> including lagged variables should be transformed to the corresponding
> rank representation. the ranked value of each observation is used to test for the causality relationship using the folowing
> R(.)=α+∑δR(.)+∑λR(.)+∑θR(.)+ε
> Thanks
> Happy
> ----- Original Message -----
>> From: Nick Cox <>
>> To: "" <>
>> Cc:
>> Sent: Tuesday, April 23, 2013 1:11 PM
>> Subject: Re: st: Regressing ranked variables for causality testing
>> Ranking is easy enough: see -help egen-.
>> What that has to do with causality tests I do not know.
>> Even more crucially, I think you need to spell out precisely which way
>> of assessing causality you have in mind, with precise names and
>> literature references.
>> Nick
>> On 23 April 2013 07:50, Happy Phiri <> wrote:
>>>  I have data for 3 variables (time series data) which I would like to rank
>> using stata, and then run a regression on rank-transformed variables in order to
>> test for causality. How is this done in stata?

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index