Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: st: Creating an index
Ozgur Ozdemir <email@example.com>
RE: st: Creating an index
Tue, 15 May 2012 20:53:29 +0000
thanks for your response. it seems, I need to use
sheaf coefficient like technique. "Combining information from multiple variables using models for casual indicators" explains how several variables can be grouped and used as an independent variable but not sure how I can use the new variable as a dependent variable in a test ?
in your example, if I want to use the family socio economic status as dependent variable in a regression, first I need to create the index and calculate the actual socio economic values. there might be also a direction problem such as some variables might increase the index while they are decreasing but others might increase the index while they are increasing ?
Skype : ozgurozdemir2005
> Date: Mon, 14 May 2012 09:30:42 +0200
> Subject: Re: st: Creating an index
> From: firstname.lastname@example.org
> To: email@example.com
> --- On Sun, May 13, 2012 at 5:30 PM, Ozgur Ozdemir wrote:
> > thanks for your quick response. I have already written a chapter on the impact of board structure on firm performance which include board specific data such as board size, number of independent directors etc. In the next chapter, I would like to introduce a variable called "Governance Index" then research association between the governance index and some other firm level characteristics. Therefore, I need a single measure to proxy governance as it will be a dependent variable in chapter 2. therefore, I cannot in chapter 2 argue that I reduced the number of variables and used factor analysis as it will bring suspects about a multicollineratiy problem for Chapter 1 however was not a case at all. So, I need to get this index in a different way. i found sheafcoef but not sure if it might help.
> You seem to treat your audience as the enemy that is trying to bring
> you down, so you need to "trick" them by not using factor analysis
> even though you want to. It usually works much better to think of your
> work as part of a larger literature. Each contribution is there to
> make a reasonable argument, which is not perfect but reasonable. The
> combined work of all the people in that literature is what will
> determine the current conclusion, not a single contribution. That way
> your audience consist of your colleagues rather than your enemys and
> purpose of your contribution is to communicate your argument as openly
> as possible.
> The choice between a factor-analysis like solution and a sheaf
> coefficient like solution depends on how you think the observed
> variable relates to your latent variable (which you call index). If
> the latent variable is influencing the observed variables than you
> want to use a factor analysis type technique, for example it is likey
> that the latent intelligence is influencing the observed answers in an
> intelligence test. If the observed variables influence the latent
> variable you want to use a sheaf coefficient like technique, for
> example we can think of the parents' occupation and education as a set
> of resources going into a family pot of resources which we can call
> family socioeconomic status.
> Hope this helps,
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
* For searches and help try: