发帖
雷达卡
intro 12 — Convergence problems and how to solve them
Description Remarks and examples Also see
Description
It can be devilishly diff i cult for software to obtain results for SEM s. Here is what can happen:
. sem . . .
Variables in structural equation model
(output omitted)
Fitting target model:
initial values not feasible
r(1400);
or,
. gsem . . .
Fitting fixed-effects model:
Iteration 0: log likelihood = -914. 65237
Iteration 1: log likelihood = -661. 32533
Iteration 2: log likelihood = -657. 18568
(output omitted)
Refining starting values:
Grid node 0: log likelihood = .
Grid node 1: log likelihood = .
Grid node 2: log likelihood = .
Grid node 3: log likelihood = .
Fitting full model:
initial values not feasible
r(1400);
or,
. sem . . .
Endogenous variables
(output omitted)
Fitting target model:
Iteration 1: log likelihood = . . .
.
.
.
Iteration 50: log likelihood = -337504. 44 (not concave)
Iteration 51: log likelihood = -337503. 52 (not concave)
Iteration 52: log likelihood = -337502. 13 (not concave)
.
.
.
Iteration 101: log likelihood = -337400. 69 (not concave)
Break
r(1);
In the f i rst two cases, sem and gsem gave up. The error message is perhaps informative if not
helpful. In the last case, sem (it could just as well have been gsem) iterated and iterated while
producing little improvement in the log-likelihood value. We eventually tired of watching a process
that was going nowhere slowly and pressed Break.
Now what?
1
Description - Stata.pdf
(3.45 MB, 需要: 30 个论坛币)
提升卡
置顶卡
沉默卡
变色卡
抢沙发
千斤顶
显身卡
京公网安备 11010802022788号
