Estimation of domain discontinuities using Hierarchical Bayesian Fay-Herriot models
Section 4. Results
4.1 Model selection
In Subsection 3.2, two different versions for the
fixed effects (FE_uq and FE_eq) and three
different covariance structures of the random effects (RE_f, RE_d and RE_s) are considered for the bivariate FH model. The
step-forward selection procedure from Subsection 3.6 is applied to each of
these six combinations separately to select covariates. Recall from Subsection 3.1
that for the bivariate FH model and the univariate FH model for the
discontinuities, potential covariates are available from the Municipal Base
Administration and the Police Register of Reported Offences. Names of these
covariates start with
and
respectively.
For the univariate FH model for the alternative survey, direct estimates from
the regular CVS are also considered as covariates (van den Brakel et al., 2016). Names of these covariates start with
See the
appendix for an overview of the covariates.
The finally selected models for the bivariate FH model
are summarized in Table 4.1. The models presented in Table 4.1 are
selected with the step-WAIC-se procedure. The step-WAIC procedure selects models with a substantially larger
amount of covariates which improve the WAIC only marginally. For
and
the step-WAIC result in a model with 4 covariates with unequal
regression coefficients and diagonal covariance matrices for the random effects
(RE_d and RE_s). For
and
the step-WAIC result in a model with respectively 3 and 2
covariates with unequal regression coefficients, also with diagonal covariance
matrices with equal variances for the random effects (RE_s). With only 25 domains, there is a substantial risk
that these models overfit the data. An exception is
where both
selection procedures result in the same model.
With the step-WAIC-se procedure
more parsimonious models are obtained as follows from Table 4.1. For total
offences,
and
a model with
only one covariate with equal regression coefficients for both surveys (FE_eq) is obtained in combination with a full covariance
matrix (RE_f) with large random domain effects with a strong
positive correlation of 0.98 for
and 0.81 for
Also for
a more
parsimonious model, with one covariate and equal regression coefficients (FE_eq) is obtained with the step-WAIC-se procedure. In this case a diagonal covariance matrix with equal variances
(RE_s) is selected. For
and
the step-WAIC-se procedure avoids the selection of large amounts of
covariates, found with the step-WAIC approach. The
selected model for
has a full
covariance matrix with a weak positive correlation of 0.1 (RE_f), and one covariate with unequal regression
coefficients (FE_uq). The model for
has a diagonal
covariance matrix with equal variances (RE_s) and one covariate
with unequal regression coefficients (FE_uq). Since
parsimonious models are preferred in this application, the models obtained with
the step-WAIC-se approach are finally selected. See van den Brakel and Boonstra (2018) for a more detailed discussion of the model selection
resulting in the finally selected models.
The models selected with the univeriate FH model for
the direct estimates of the discontinuities, developed in Subsection 3.3,
are summarized in Table 4.2. The models are selected with the step-WAIC-se procedure. For
the step-WAIC results in a model with four covariates. For the
other variables the same models are selected as with the step-WAIC-se procedure. The univariate FH models developed in van den Brakel et al. (2016) for the alternative survey approach are summarized in
Table 4.3.
Standard model diagnostics test the underlying
assumptions that the random domain effects and the residuals are normally and
independently distributed. Since the number of domains in this application is
small, the power of the tests for normality are weak and do not indicate
deviations from normality. Therefore the posterior predictive tests as
summarised in Subsection 3.5 are used to evaluate the model adequacy. In
addition, the domain predictions aggregated to the national level are compared
with the direct estimates at the national level to evaluate the bias introduced
with the small area estimation procedures in Subsection 4.2. The posterior
predictive
-values for the
domain estimates of the target variables and the discontinuities are summarized
in Table 4.4 for the bivariate FH model and Table 4.5 for the
univariate FH model for the discontinuities. The general measure for goodness-of-fit
indicates that
the fit for the discontinuities of
is of reduced
quality (other models considered had similar high values). The values for the
bivariate FH model are slightly better compared to the univariate FH model for
the discontinuities. The posterior predictive
-values for
maximum
and minimum
values do not
indicate problems with the tails of the distributions. For these posterior
predictive
-values there
are no systematic differences between bivariate and univariate FH model. The
values for
and
for the
discontinuities of the bivariate model are comparable with the values for the
univariate model. The posterior predictive values for the mean
and asymmetry
of the distribution
indicate that
the distributions are symmetrically concentrated around their mean. The
posterior predictive
-values for the
variance
indicate some
undershrinkage for the discontinuities of
and
under both the
bivariate and univariate FH model.
Table 4.1
Final models bivariate FH model selected with step-WAIC-se. All models contain an intercept.
: correlation between the random effects
Table summary
This table displays the results of Final models bivariate FH model selected with step-WAIC-se. All models contain an intercept.
: correlation between the random effects. The information is grouped by Variable (appearing as row headers), Model and Covariance structure random effects (appearing as column headers).
Variable |
Model |
Covariance structure random effects |
type |
|
|
|
|
FE_eq:
+ PR_weapon
|
RE_f |
8.77 |
5.32 |
0.98 |
|
FE_eq:
+ PR_propcrim
|
RE_s |
1.17 |
1.17 |
- |
|
FE_eq:
+ MBA_immigrnw
|
RE_s |
0.20 |
0.14 |
0.81 |
|
FE_uq: MBA_immigr
|
RE_s |
0.78 |
0.78 |
- |
|
FE_uq: PR_propcrim
|
RE_s |
0.79 |
0.39 |
0.2 |
Table 4.2
Final models univariate FH model for direct estimates of the discontinuities. All models contain an intercept
Table summary
This table displays the results of Final models univariate FH model for direct estimates of the discontinuities. All models contain an intercept. The information is grouped by Variable (appearing as row headers), Model and Variance random effects
(appearing as column headers).
Variable |
Model |
Variance random effects
|
|
PR_propcrim
|
0.80 |
|
MBA_benefit
|
1.03 |
|
PR_threat
|
0.038 |
|
MBA_benefit
|
0.928 |
|
PR_assault
|
0.485 |
Table 4.3
Final models univariate FH model for alternative CVS from van den Brakel et al. (2016). All models contain an intercept
Table summary
This table displays the results of Final models univariate FH model for alternative CVS from van den Brakel et al. (2016). All models contain an intercept. The information is grouped by Variable (appearing as row headers), Model and Variance random effects
(appearing as column headers).
Variable |
Model |
Variance random effects
|
|
CVSR_victim
|
0.003 |
|
CVSR_nuisance
+ MBA_benefit
+ PR_propcrim
+ PR_drugs
|
2.997 |
|
CVSR_nuisance
+ MBA_old
|
0.805 |
|
CVSR_funcpol
|
4.995 |
|
PR_propcrim
+ MBA_old
|
7.725 |
Table 4.4
Posterior predictive p-values for the final multivariate FH models from Table 4.1
Table summary
This table displays the results of Posterior predictive p-values for the final multivariate FH models from Table 4.1. The information is grouped by Variable (appearing as row headers),
,
,
,
,
and
calculated using Discontinuities and Target variables units of measure (appearing as column headers).
Variable |
|
|
|
|
|
|
Discontinuities |
|
0.980 |
0.797 |
0.069 |
0.337 |
0.968 |
0.416 |
|
0.343 |
0.841 |
0.833 |
0.454 |
0.437 |
0.912 |
|
0.927 |
0.940 |
0.034 |
0.345 |
0.988 |
0.465 |
|
0.772 |
0.595 |
0.392 |
0.610 |
0.762 |
0.484 |
|
0.925 |
0.261 |
0.029 |
0.258 |
0.970 |
0.070 |
|
Target variables |
|
0.859 |
0.249 |
0.024 |
0.317 |
0.524 |
0.089 |
|
0.308 |
0.779 |
0.492 |
0.420 |
0.474 |
0.708 |
|
0.766 |
0.317 |
0.108 |
0.433 |
0.504 |
0.156 |
|
0.742 |
0.929 |
0.584 |
0.457 |
0.875 |
0.797 |
|
0.695 |
0.339 |
0.168 |
0.379 |
0.655 |
0.194 |
Table 4.5
Posterior predictive p-values for the final univeriate FH models for direct estimates of the discontinuities from Table 4.2
Table summary
This table displays the results of Posterior predictive p-values for the final univeriate FH models for direct estimates of the discontinuities from Table 4.2. The information is grouped by Variable (appearing as row headers),
,
,
,
,
and
(appearing as column headers).
Variable |
|
|
|
|
|
|
Discontinuities |
|
0.985 |
0.828 |
0.071 |
0.390 |
0.972 |
0.438 |
|
0.382 |
0.885 |
0.816 |
0.464 |
0.523 |
0.920 |
|
0.970 |
0.941 |
0.072 |
0.434 |
0.978 |
0.554 |
|
0.814 |
0.607 |
0.378 |
0.607 |
0.783 |
0.488 |
|
0.946 |
0.272 |
0.052 |
0.383 |
0.963 |
0.092 |
4.2 Estimation results
In this Subsection
estimation results for the three different modelling approaches are discussed.
In Subsection 4.2.1 the HB predictions for the target variables under the
regular and alternative survey obtained with the bivariate FH model are
compared with the direct estimates and with the domain predictions obtained
with the univariate FH model where the direct estimates of the regular approach
are potential auxiliary variables in the model selection. Subsequently results
for the domain discontinuities are discussed in Subsection 4.2.2. Here the
results obtained with the univariate FH model for the discontinuities are also
discussed.
With model-based small
area estimation, the design variance of the direct estimators is reduced at the
cost of accepting some amount of design bias. To evaluate differences in the
direct point estimates and the small domain predictions, the following two
measures are defined. The first one is the Mean Relative Difference (MRD),
which summarizes the differences between the direct estimates and the domain
predictions:
and
is the domain prediction based on the
bivariate FH model or the univariate FH model. The second measure is the
Absolute Mean Relative Difference (AMRD) between the direct estimate and the
domain prediction, which is defined as:
the increased precision of the small domain predictions is measured with
Mean Relative Difference of the Standard Errors (MRDSE) between the direct
estimates and the domain predictions and is defined as
these measures are defined in a similar way for the estimates and
predictions of the domain discontinuities
and
4.2.1 Results for variables under the regular and
alternative survey
In Table 4.6 the domain
predictions and their standard errors averaged over the domains as well as the
MRD, AMRD and MRDSE are given for the alternative survey under the univariate
FH model with the models presented in Table 4.3. Results under the
bivariate FH model, based on the final models of Table 4.1, are presented
in Table 4.7 for the variables under the alternative survey and in Table 4.8
for the variables under the regular survey. Comparing the standard errors (SE)
and the MRDSE in Table 4.6 and Table 4.7 shows that the bivariate FH
model results in stronger reductions of the standard errors for all variables
with the exception of
This comes at
the cost of an increased bias. Comparing MRD and AMRD in both tables shows that
the deviations between the direct estimates and the small area predictions are
larger under the bivariate FH model. Comparing the SE and MRDSE in Tables 4.7
and 4.8 shows that the improvement in precision with the bivariate FH model for
the regular survey is smaller, as expected since the sample size of the regular
survey is larger. The bias in the bivariate FH model predictions for the
regular survey are also smaller, which follows from a comparison of MRD and
AMRD in Tables 4.7 and 4.8.
The domain predictions under
the univariate and bivariate FH model are plotted against the GREG estimates in
Figures 4.1 through 4.5. The graphs also contain the GREG estimate at the
national level versus the domain predictions aggregated to the national level
according to (4.4). Figures 4.1 and 4.3 show that there is only a small
amount of shrinkage for
and
Figure 4.2
shows for
that the
bivariate FH model shrinks the domain predictions for the alternative survey
while the amount of shrinkage for the univariate FH model for the alternative
CVS and the bivariate FH model for the regular survey is smaller. For
see Figure 4.4,
there is a small difference between the amount of shrinkage of the alternative
CVS under the bivariate and univariate model. From Figure 4.5 it follows
that the bivariate FH model for
cannot
adequately model the observations under the alternative survey with the
auxiliary information from the two registers (MBA and PRRO). In this case the
domain predictions of
under the
alternative approach display extreme overshrinkage. The univariate FH model
indeed selects the same auxiliary variable from the regular survey only, see
Table 4.3 and results in more realistic domain predictions.
For variables related to
opinions and views such as
and
the reduction
in the standard errors is accompanied by a relatively strong increase in the
bias. This is especially the case with the small area prediction of the
bivariate FH model for the alternative survey. For these variables, there are
no strongly correlated covariates in the MBA
and PRRO. In these cases the
univariate FH model performs better since related covariates from the regular
survey are selected (see Table 4.3), while the bivariate model doesn’t
detect correlation between the random effects (see Table 4.1).
Table 4.6
Average of domain predictions alternative survey with univariate FH model from van den Brakel et al. (2016)
Table summary
This table displays the results of Average of domain predictions alternative survey with univariate FH model from van den Brakel et al. (2016). The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), AMRD (%) and MRDSE (%) (appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
AMRD (%) |
MRDSE (%) |
|
33.21 |
2.90 |
-0.44 |
7.03 |
47.74 |
|
19.83 |
1.64 |
-0.96 |
7.58 |
41.16 |
|
1.29 |
0.08 |
-0.74 |
5.02 |
37.96 |
|
55.09 |
2.54 |
-0.11 |
6.43 |
61.98 |
|
9.85 |
0.84 |
-3.17 |
11.86 |
60.69 |
Table 4.7
Average of domain predictions alternative survey with bivariate FH model
Table summary
This table displays the results of Average of domain predictions alternative survey with bivariate FH model. The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), AMRD (%) and MRDSE (%) (appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
AMRD (%) |
MRDSE (%) |
|
33.26 |
2.82 |
-0.99 |
6.93 |
49.36 |
|
19.82 |
1.21 |
-2.54 |
11.97 |
56.47 |
|
1.28 |
0.08 |
-0.98 |
4.28 |
35.72 |
|
55.08 |
1.97 |
-0.49 |
8.97 |
70.06 |
|
9.91 |
0.73 |
-4.81 |
14.70 |
65.35 |
Table 4.8
Average of domain predictions regular survey with bivariate FH model
Table summary
This table displays the results of Average of domain predictions regular survey with bivariate FH model. The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), AMRD (%) and MRDSE (%) (appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
AMRD (%) |
MRDSE (%) |
|
41.34 |
3.76 |
0.96 |
4.56 |
17.95 |
|
24.22 |
1.07 |
-0.01 |
6.02 |
46.78 |
|
1.60 |
0.09 |
0.38 |
2.62 |
15.93 |
|
60.82 |
1.47 |
-0.77 |
5.38 |
64.83 |
|
12.18 |
0.88 |
1.58 |
7.84 |
43.70 |
The direct estimates at
the national level are accurate estimates since they are based on sufficiently
large sample sizes. Therefore the bias in model-based domain predictions is
often assessed by comparing the direct estimates at the national level with the
domain predictions aggregated to the national level. The target variables in
this application are all defined as population means. Therefore the aggregated
domain predictions are obtained as the average over the domains weighted with
the relative domain sizes,
with
the population size of domain
and
the size of the total population.
Table 4.9 compares
the weighted average of the domain predictions according to (4.4) with the
national GREG estimates. For the univariate FH model for the alternative CVS,
the aggregated domain predictions are almost exactly equal to the GREG estimates
at the national level. For the bivariate FH model the differences are slightly
larger but the aggregated domain predictions are still very close to the GREG
estimates at the national level. The largest relative difference amounts to 3%
and is observed for
under the
regular survey.
Table 4.9
GREG estimates national level and aggregated HB predictions regular and alternative survey approach (4.4)
Table summary
This table displays the results of GREG estimates national level and aggregated HB predictions regular and alternative survey approach (4.4). The information is grouped by Variable (appearing as row headers), Regular, Alternative and Discontinuity (appearing as column headers).
Variable |
Regular |
Alternative |
Discontinuity |
GREG |
biv. FH |
GREG |
biv. FH |
uni. FH |
GREG |
biv. FH |
uni. FH |
FH
|
|
43.79 |
42.47 |
34.09 |
34.02 |
34.09 |
9.7 |
8.45 |
9.7 |
9.04 |
|
25.07 |
24.89 |
20.48 |
20.49 |
20.48 |
4.59 |
4.40 |
4.59 |
4.69 |
|
1.67 |
1.66 |
1.34 |
1.34 |
1.34 |
0.33 |
0.32 |
0.34 |
0.33 |
|
59.88 |
60.36 |
55.10 |
55.06 |
55.12 |
4.78 |
5.29 |
5.04 |
5.07 |
|
13.02 |
12.76 |
10.32 |
10.33 |
10.32 |
2.70 |
2.43 |
2.70 |
2.63 |
Description for Figure 4.1
Figure representing the domain predictions of offtot under the univariate and bivariate FH model plotted against the GREG estimates. Also contain the GREG estimate at the national level versus the domain predictions aggregated to the national level according to equation 4.4. The upper panel represents the regular survey using bivariate FH model. The middle panel represents the alternative survey using bivariate FH model. The lower panel represents thealternative survey using univariate FH model. There is only a small amount of shrinkage for offtot.
Description for Figure 4.2
Figure representing the domain predictions of unsafe under the univariate and bivariate FH model plotted against the GREG estimates. Also contain the GREG estimate at the national level versus the domain predictions aggregated to the national level according to equation 4.4. The upper panel represents the regular survey using bivariate FH model. The middle panel represents the alternative survey using bivariate FH model. The lower panel represents the alternative survey using univariate FH model. The figure shows for unsafe that the bivariate FH model shrinks the domain predictions for the alternative survey while the amount of shrinkage for the univariate FH model for the alternative CVS and the bivariate FH model for the regular survey is smaller.
Description for Figure 4.3
Figure representing the domain predictions of nuisance under the univariate and bivariate FH model plotted against the GREG estimates. Also contain the GREG estimate at the national level versus the domain predictions aggregated to the national level according to equation 4.4. The upper panel represents the regular survey using bivariate FH model. The middle panel represents the alternative survey using bivariate FH model. The lower panel represents the alternative survey using univariate FH model. There is only a small amount of shrinkage for nuisance.
Description for Figure 4.4
Figure representing the domain predictions of propvict under the univariate and bivariate FH model plotted against the GREG estimates. Also contain the GREG estimate at the national level versus the domain predictions aggregated to the national level according to equation 4.4. The upper panel represents the regular survey using bivariate FH model. The middle panel represents the alternative survey using bivariate FH model. The lower panel represents the alternative survey using univariate FH model. For propvict there is a small difference between the amount of shrinkage of the alternative CVS under the bivariate and univariate model.
Description for Figure 4.5
Figure representing the domain predictions of satispol under the univariate and bivariate FH model plotted against the GREG estimates. Also contain the GREG estimate at the national level versus the domain predictions aggregated to the national level according to equation 4.4. The upper panel represents the regular survey using bivariate FH model. The middle panel represents the alternative survey using bivariate FH model. The lower panel represents the alternative survey using univariate FH model. It follows that the bivariate FH model for satispol cannot adequately model the observations under the alternative survey with the auxiliary information from the two registers (MBA and PRRO). In this case the domain predictions of satispol under the alternative approach display extreme over shrinkage.
4.2.2 Results for discontinuity estimates
In the last four columns of Table 4.9, the GREG
estimates for the discontinuities at the national level are compared with the
domain predictions obtained with the univariate FH model for the alternative
CVS, the bivariate FH model and the univariate FH model for the
discontinuities, aggregated to the national level using (4.4). The differences
between the GREG estimates for the discontinuities at the national level and
the aggregated domain predictions are the largest for the bivariate model and
the smallest for the univariate FH model for the alternative CVS. This can be
expected since the bivariate FH model shrinks both the domain estimates for the
regular and alternative survey. With the univariate FH model for the
alternative CVS, only the estimates for the alternative survey are replaced by
domain predictions, while the estimates for the regular survey are not
adjusted. In addition the domain predictions for the alternative survey have
larger MRD’s and AMRD’s under the bivariate FH model compared to the univariate
FH model (compare Table 4.6 and 4.7). The differences for the univariate
FH model for the discontinuities are smaller compared to the bivariate FH model
but larger compared to the univariate FH model for the alternative CVS.
In Tables 4.10, 4.11, and 4.12 the domain
predictions and their standard errors for the discontinuities averaged over the
domains as well as the MRD and MRDSE are summarized for the univariate FH model
for the alternative CVS, bivariate FH model and the univariate FH model for the
discontinuities respectively. The MRD’s are large because the GREG estimates
for the discontinuities in the denominator of (4.11) frequently take values
close to zero, which make these indicators unstable. Therefor the AMRD is
replaced by the median of the absolute relative differences,
and is
abbreviated as MARD. The latter are indeed more stable indicators for bias. The
MARD is the smallest for the univariate FH model for the alternative CVS, since
this approach only adjusts the domain predictions of the alternative CVS. The
MARD values for the bivariate FH model on their turn are smaller than those for
the univariate FH model for the discontinuities.
With the exception of
the standard
errors for the domain predictions under the bivariate FH model are smaller
compared to the univariate FH model for the alternative CVS. In the case of
and
this is the
result of slightly more precise domain predictions for the alternative survey
with respect to the univariate FH model (compare Table 4.6 with 4.7), a
clear improvement in precision of the domain predictions of the regular survey
compared to the GREG estimators (Table 4.8 and 2.2) and the positive
correlation between the random effects. In the case of
and
this is mainly
the result of a clear improvement of precision of the domain predictions with
the bivariate FH model for the regular compared to the GREG estimators (Table 4.8
and 2.2) and also a clear improvement of the precision of the domain
predictions with the bivariate FH model for the alternative survey compared to
the univariate model (compare Table 4.6 with Table 4.7).
For all five variables, the smallest standard errors
are obtained with the univariate FH model for the discontinuities. This comes
at the cost of a larger bias, as illustrated with the MARD values. An exception
is
for which the
bias in terms of MARD for the bivariate FH model is clearly larger than the
univariate FH model for the discontinuities. For this variable the bias is the
lowest with the univariate FH model for the alternative CVS, but the reduction
of the standard errors is also smaller.
The last columns of Tables 4.11 and 4.12 contain
the shrinkage factors for the domain discontinuities averaged over the domains.
For the univariate FH model for the discontinuities the shrinkage factors for
the predictions of the domain discontinuities, i.e., the weights attached to
the direct estimator for the discontinuities, are defined as
For the
bivariate model the shrinkage factors for the predictions of the domain
discontinuities are defined as
with
The average
shrinkage factor is defined as
Note that this
statistic is not available for the discontinuities obtained with the univariate
FH model for the alternative CVS, since under this approach domain
discontinuities are obtained as the contrast between the GREG estimate for the
regular survey and the domain prediction for the alternative approach. With the
exception of
the shrinkage
factors under the univariate FH model for discontinuities are a factor 10
smaller compared to those of the bivariate FH model. The question rises whether
the extremely small shrinkage factors of the univariate FH model for the
discontinuities overshrink the direct estimates of the discontinuities.
Table 4.10
Domain predictions for discontinuities univariate FH model for the alternative CVS
Table summary
This table displays the results of Domain predictions for discontinuities univariate FH model for the alternative CVS. The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), MARD (%) and MRDSE (%) (appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
MARD (%) |
MRDSE (%) |
|
9.08 |
3.92 |
-5.67 |
22.14 |
48.47 |
|
4.55 |
2.46 |
42.98 |
23.44 |
29.45 |
|
0.33 |
0.07 |
-5.80 |
11.49 |
57.49 |
|
5.52 |
4.72 |
99.26 |
47.72 |
40.86 |
|
2.70 |
1.83 |
-142.60 |
30.49 |
32.75 |
Table 4.11
Domain predictions for discontinuities bivariate FH model
Table summary
This table displays the results of Domain predictions for discontinuities bivariate FH model. The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), MARD (%), MRDSE (%) and
(appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
MARD (%) |
MRDSE (%) |
|
|
8.07 |
2.68 |
1.94 |
23.34 |
63.60 |
0.208 |
|
4.40 |
1.56 |
55.49 |
41.24 |
55.09 |
0.317 |
|
0.31 |
0.09 |
-4.01 |
18.43 |
44.47 |
0.327 |
|
5.74 |
2.46 |
228.50 |
73.43 |
68.75 |
0.019 |
|
2.27 |
1.10 |
-113.00 |
27.11 |
59.02 |
0.376 |
Table 4.12
Domain predictions for discontinuities univariate FH model for the direct estimates of the discontinuities
Table summary
This table displays the results of Domain predictions for discontinuities univariate FH model for the direct estimates of the discontinuities. The information is grouped by Variable (appearing as row headers), HB est., SE, MRD (%), MARD (%), MRDSE (%) and
(appearing as column headers).
Variable |
HB est. |
SE |
MRD (%) |
MARD (%) |
MRDSE (%) |
|
|
8.35 |
2.25 |
-33.39 |
26.63 |
69.35 |
0.012 |
|
4.43 |
1.48 |
45.45 |
42.14 |
59.99 |
0.082 |
|
0.32 |
0.06 |
-13.85 |
20.97 |
63.04 |
0.049 |
|
5.67 |
2.39 |
186.07 |
65.33 |
69.61 |
0.014 |
|
2.54 |
0.91 |
-55.59 |
45.73 |
65.84 |
0.032 |
Plots of discontinuities estimated with the GREG
estimator, the univariate FH model for the alternative CVS, the bivariate FH
model and the univariate FH model for the discontinuities are provided in
Figures 4.6 through 4.10. The predictions for the domain discontinuities
obtained with the three models are more stable compared to the GREG estimates.
This is e.g., clearly illustrated with
(Figure 4.7),
where the GREG estimates for the discontinuity are sometimes positive and
sometimes negative. The predictions for the domain discontinuities under the
bivariate FH model and the univariate FH model for the discontinuities are
consistently positive, which appears more plausible since it is unlikely that
the domain discontinuities have opposite signs. The predictions for the domain
discontinuities under the univariate FH model for the alternative CVS are
closer to the GREG estimates and consequently less stable. A similar pattern
can be observed for the other variables.
These plots illustrate that for
and
the bivariate
FH model results in a clear improvement of the predictions for the domain
discontinuities compared to the univariate FH model for the alternative CVS.
For
the standard
errors for the discontinuities increase with the bivariate FH model compared to
the univariate FH model for the alternative CVS. The bivariate FH model for
cannot
adequately model the observations under the alternative survey with the
auxiliary information from the two registers (MBA
and PRRO). In this case the
domain predictions of
under the
alternative approach display overshrinkage. The univariate FH model indeed
selects an auxiliary variable from the regular survey, see Table 4.3, and
clearly performs better.
It was anticipated that it would be difficult to
produce reasonable predictions for the domain discontinuities with the
univariate FH model for the direct estimates of the discontinuities since it is
hard to imagine that the available auxiliary variables from registers like the
MBA
and PRRO contain good predictors for systematic differences in survey
errors. Nevertheless, reasonable results are obtained with this more pragmatic
approach. A possible interpretation is that the discontinuities are to some
extent proportional to the values of the target variable and therefore show
some systematic pattern that can be explained partially with the selected
covariates. A point of concern are the very small shrinkage factors under this
model, which might be an indication that the model gives too much weight to the
synthetic estimator.
Description for Figure 4.6
Figure representing the discontinuities of offtot based on the GREG estimator (upper panel), univariate FH model (second panel), bivariate FH model (third panel) and univariate FH model for direct estimates discontinuities (lower panel) with a 95% confidence interval. The predictions for the domain discontinuities obtained with the three models are more stable compared to the GREG estimates. The bivariate FH model results in a clear improvement of the predictions for the domain discontinuities compared to the univariate FH model for the alternative CVS.
Description for Figure 4.7
Figure representing the discontinuities of unsafe based on the GREG estimator (upper panel), univariate FH model (second panel), bivariate FH model (third panel) and univariate FH model for direct estimates discontinuities (lower panel) with a 95% confidence interval. The predictions for the domain discontinuities obtained with the three models are more stable compared to the GREG estimates.The bivariate FH model results in a clear improvement of the predictions for the domain discontinuities compared to the univariate FH model for the alternative CVS.
Description for Figure 4.8
Figure representing the discontinuities of nuisance based on the GREG estimator (upper panel), univariate FH model (second panel), bivariate FH model (third panel) and univariate FH model for direct estimates discontinuities (lower panel) with a 95% confidence interval. The predictions for the domain discontinuities obtained with the three models are more stable compared to the GREG estimates. The standard errors for the discontinuities increase with the bivariate FH model compared to the univariate FH model for the alternative CVS.
Description for Figure 4.9
Figure representing the discontinuities of propvict based on the GREG estimator (upper panel), univariate FH model (second panel), bivariate FH model (third panel) and univariate FH model for direct estimates discontinuities (lower panel) with a 95% confidence interval. The predictions for the domain discontinuities obtained with the three models are more stable compared to the GREG estimates. The bivariate FH model results in a clear improvement of the predictions for the domain discontinuities compared to the univariate FH model for the alternative CVS.
Description for Figure 4.10
Figure representing the discontinuities of satispol based on the GREG estimator (upper panel), univariate FH model (second panel), bivariate FH model (third panel) and univariate FH model for direct estimates discontinuities (lower panel) with a 95% confidence interval. The predictions for the domain discontinuities obtained with the three models are more stable compared to the GREG estimates. The bivariate FH model for satispol cannot adequately model the observations under the alternative survey with the auxiliary information from the two registers (MBA and PRRO). In this case the domain predictions of satispol under the alternative approach display over shrinkage.