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, statalist.org is already up and running.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Zhi Su <firstname.lastname@example.org>:
This is an ill-posed problem. You don't say whether each coef b1,
b2, b3, b4 is supposed to be just insignificant, at what level, or
whether a joint test of significance is the criterion. You don't say
anything about the distribution of the variables, or how the VCE is
calculated, all of which could influence your solution. Further, in a
multi-dimensional setting like this, there could be many solutions,
and the direction you approach from could easily determine which
solution you find.
E.g. suppose v's are all Y plus negligible error so that any r
approaching 1 drives variation in Y to zero; then any r approaching 1
is a solution, but so is any sum of r's approaching one. There will
be various linear combinations of r that " just make b1, b2,
> b3, b4 go from signicant to insignificant."
On Sun, Jul 3, 2011 at 11:36 PM, Zhi Su <email@example.com> wrote:
> Dear Statalists,
> I need to do some sensitivity tests that test how the coefficients
> of a set of independent variables change according to the changes of
> some ratios. The dependent is
> Y-r1*v1-r2*v2-r3*v3-r4*v4, where Y, v1, v2, v3 and v4 are known and
> r1, r2, r3, r4 are random values on the interval [0,1). The
> independent varaibles are x1, x2, x3, x4.
> The regression is
> When r1, r2, r3, r4 all equal to 0, b1, b2, b3, b4 are all significant.
> Then I need to change the values of the ratios r1, r2, r3, r4. The
> magnititudes and significance of b1, b2, b3, b4 will change.
> I need to record the values of r1, r2, r3, r4 that just make b1, b2,
> b3, b4 go from signicant to insignificant.
> Do anyone have some ideas about the codes for such test?
348 Holmes Hall
360 Huntington Avenue
Boston, MA 02115
* For searches and help try: