Hello
I used proc gee to estimate initial values for parameters.
proc gee data=new;
class var1 id ;
Model res= var1 var2 var3 var4 var5 var6 var7 / dist= mult;
repeated subject=id / logor=exch;
run;
Also I used these codes to find s1,s2 and s12.
proc means data=new STD;
var PP LL;
run;
&
proc corr data=new ;
var PP LL;
run;
But I have problem with estimate of the s1, s2 and s12 in PROC NLMIXED and receive different errors. The result for my parameters is meaningless.
Parameter EstimatesParameter Estimate StandardError DF t Value Pr > |t| 95% Confidence Limits Gradienta1a2var1_avar2_avar3_avar4_avar5_avar6_avar7_ab1b2var1_bvar2_bvar3_bvar4_bvar5_bvar6_bvar7_bs1s2s12
| -3.4120 | 1.4995 | 62 | -2.28 | 0.0264 | -6.4095 | -0.4145 | -54.9813 |
| 0.2214 | 1.1533 | 62 | 0.19 | 0.8484 | -2.0840 | 2.5267 | 71.2026 |
| -0.03426 | 0.09628 | 62 | -0.36 | 0.7232 | -0.2267 | 0.1582 | 49.4322 |
| 0.09149 | 0.3459 | 62 | 0.26 | 0.7923 | -0.5999 | 0.7828 | 0.32528 |
| -0.02989 | 0.06602 | 62 | -0.45 | 0.6524 | -0.1619 | 0.1021 | -91.7194 |
| 0.08576 | 0.04168 | 62 | 2.06 | 0.0438 | 0.002453 | 0.1691 | 50.9831 |
| -0.06421 | 0.06531 | 62 | -0.98 | 0.3294 | -0.1948 | 0.06635 | 72.6902 |
| -0.3454 | 0.07183 | 62 | -4.81 | <.0001 | -0.4890 | -0.2018 | -58.7358 |
| -0.09685 | 0.08108 | 62 | -1.19 | 0.2368 | -0.2589 | 0.06523 | 179.168 |
| -4.5876 | 118.41 | 62 | -0.04 | 0.9692 | -241.28 | 232.10 | -0.00321 |
| -2.1619 | 141.76 | 62 | -0.02 | 0.9879 | -285.53 | 281.21 | 0.000434 |
| -0.02860 | 9.8942 | 62 | -0.00 | 0.9977 | -19.8067 | 19.7495 | -0.01680 |
| 0.4285 | 27.2215 | 62 | 0.02 | 0.9875 | -53.9864 | 54.8434 | -0.00174 |
| 0.1637 | 11.2263 | 62 | 0.01 | 0.9884 | -22.2773 | 22.6047 | -0.01512 |
| 0.03240 | 2.5030 | 62 | 0.01 | 0.9897 | -4.9711 | 5.0359 | -0.03055 |
| 0.1275 | 6.2623 | 62 | 0.02 | 0.9838 | -12.3907 | 12.6457 | -0.02998 |
| -0.1555 | 4.8749 | 62 | -0.03 | 0.9747 | -9.9004 | 9.5894 | -0.00437 |
| -0.00080 | 5.9949 | 62 | -0.00 | 0.9999 | -11.9843 | 11.9827 | -0.03039 |
| 0.06539 | 0.01152 | 62 | 5.68 | <.0001 | 0.04237 | 0.08842 | 124.931 |
| 1.9650 | 6.7415 | 62 | 0.29 | 0.7717 | -11.5111 | 15.4411 | -0.00558 |
| -0.03700 | 0.01338 | 62 | -2.76 | 0.0075 | -0.06375 | -0.01024 | -0.00096 |
As you can see, the p-values for second joint variables are meaningless. So the results do not seem correct.
How can I find the best s1, s2 and s12 for my joint model?
Thanks for your answer.
Mahdia
