Subset of summary results

[Saqlain]

I am using MNP package for using multinomial porbit model.

model<-mnp(choice ~ 1 + AUS + Sub + AFR + BEG + CAM, base = 4, invcdf = FALSE, trace = FALSE, n.draws = 15000, p.var = 0.25, p.scale = 1, coef.start = 0, cov.start = 1, burnin = 0, thin = 0, verbose = TRUE)

The part of my data is:
         AUS         Sub       AFR           BEG    CAM  choice
1      1      3.178054      14.74568     1         4.00     2
2      0      3.401197      14.56130     1         5.00     4
3      0      3.178054      14.70488     1        7.00      2
4      0      5.940171      11.42076     0        0.00      4
5      0      2.484907      14.14389     1        9.00      3
6      0      2.772589      14.34799     1        4.00      3
7      1      4.204693      14.64967     1        10.70    1
8      1      3.988984      14.57800     1         4.50     2
9      1      4.317488      14.46320     1         5.00     1
10    1     4.204693      13.55739     1         3.52      1

And the summary of the results is

Call:
mnp(formula = choice ~ 1 + AUS + Sub + AFR + BEG + CAM, base = 4,
    invcdf = FALSE, trace = FALSE, n.draws = 1400, p.var = 0.25,
    p.scale = 1, coef.start = 0, cov.start = 1, burnin = 0, thin = 0,
    verbose = TRUE)


Coefficients:
                           mean          std.dev.        2.5%           97.5%
(Intercept):1 -0.82756      0.13703      -1.09865     -0.572
(Intercept):2 -0.18204      0.12870      -0.47696      0.017
(Intercept):3 -1.09682      0.19403      -1.47622     -0.752
AUS:1                0.02495     0.11110      -0.19510      0.241
AUS:2                0.41382     0.16693       0.15241      0.808
AUS:3                0.18614     0.12702      -0.05246      0.459
Sub:1                 0.45978     0.13664       0.20319      0.736
Sub:2                 0.12482     0.13152      -0.15545     0.377
Sub:3                 0.11627     0.14044      -0.16041     0.381
AFR:1                 -0.08944   0.15293      -0.39566      0.202
AFR:2                 0.72140   0.27844       0.26318       1.350
AFR:3                 0.18651   0.18408      -0.14078       0.573
BEG:1                0.77832    0.14485       0.51547       1.083
BEG:2                -0.48614    0.25192      -1.07826    -0.088
BEG:3                0.21806    0.18540      -0.12307       0.613
CAM:1               -0.51257  0.21925      -0.94448      -0.069
CAM:2               -0.86292  0.29425      -1.48575      -0.393
CAM:3               -0.434    0.23812        -0.92518       0.036

Covariances:
        mean std.dev.     2.5% 97.5%
1:1  1.00000  0.00000  1.00000 1.000
1:2  0.31050  0.20152 -0.11133 0.688
1:3  0.05604  0.31220 -0.52210 0.563
2:2  1.28810  1.03709  0.17646 4.266
2:3  0.23577  0.39763 -0.26753 1.319
3:3  0.63592  0.25132  0.33837 1.283

I want to retain the summary of results with (Intercept):1, (Intercept):2, (Intercept):3, AUS:1, AUS:2, Sub:1, Sub:2, AFR:1, AFR:2, BEG:1, BEG:2, CAM:1, CAM:2 and the covariance matrix as well.  I am anxiously seeking some help regarding this problem.
Thanks
Saqlain

[Özgür Asar]

Did you try

summary(model)$coefficients

summary(model)$covariances

names(model) would help you as well.

ozgur

[Saqlain]

If I use summary(model)$coef.table, it gives me all the coefficients but as I said I want to retain a part of coefficients not all in the summary.
I may further go ahead a step. This is the data of trading companies in which they use two quality signals for their products. The choice variable has been created as: (1 if company hold first quality signal), (2 if company hold second quality signal), (3 if company hold both quality signals) and (4 if company hold none of the quality signals).  I need only the coefficients regarding choice==1 and choice==2 but intercepts for all three choices.

[Özgür Asar]

What about

summary(model)$coefficients[,1] ?

ozgur

[Saqlain]

yes, it gives me the required summary statistics with the following command

summary(Mnp)$coef.table[c(1:3,5,6,8,9,11,12,14,15,17,18),]

Normally if we compare the two summary results before and after skipping some variables, we note bit change in the estimates of the rest of the variables. But in this case, it shows the same. I want to ask are the results in this case valid and publishable?

Thanks

[Özgür Asar]

>yes, it gives me the required summary statistics with the following command

>summary(Mnp)$coef.table[c(1:3,5,6,8,9,11,12,14,15,17,18),]

>Normally if we compare the two summary results before and after skipping some variables, we note bit change in >the estimates of the rest of the variables. But in this case, it shows the same. I want to ask are the results in this case >valid and publishable?

I expect difference as well. You can try assigning two models to different objects such as Mnp1 and Mnp2, then try to extract interested statistics from them and compare the results.

>Thanks