7.2 Estimation of domain quantiles

If we may want to estimate the population quantiles and medians of the self-reported weight and height among people with or without a university degree, we may consider the following codes.

# For people who did not attend an university 
Quant.NUni<- svyquantile( ~ HWT_DHT_M_TRM+HWT_WGHT_KG_TRM, 
  quantile = c(0.025, 0.05, 0.1, 0.5, 0.9, 0.95, 0.975),alpha = 0.05,
  design = subset(CLSA.design, ED_HIGH_TRM == "Non_university"), 
  interval.type = "Wald", ties = c("rounded"), ci = TRUE, se = TRUE)
Quant.NUni; SE(Quant.NUni); 
# For people who attended an university 
Quant.Uni<-svyquantile( ~ HWT_DHT_M_TRM+HWT_WGHT_KG_TRM, 
  quantil =c(0.025, 0.05, 0.1, 0.5, 0.9, 0.95, 0.975),alpha = 0.05, 
  design = subset(CLSA.design, ED_HIGH_TRM == "University"), 
  interval.type = "Wald",ties = c("rounded"), ci = TRUE, se = TRUE)
Quant.Uni; SE(Quant.Uni);

SAS

PROC SURVEYMEANS data = CLSAData 
QUANTILE = (0.025 0.05 0.1 0.5 0.9 0.95 0.975) NONSYMCL;    
VAR    HWT_DHT_M_TRM HWT_WGHT_KG_TRM ;
DOMAIN ED_HIGH_TRM; 
STRATA GEOSTRAT_TRM ;  
WEIGHT WGHTS_INFLATION_TRM;                                       
RUN;

SPSS and Stata In \(\texttt{SPSS}\) and \(\texttt{Stata}\) packages, there is no formal procedure available to produce quantile estimates for domain analysis.

Result comparison

	HWT_DHT_M_TRM				HWT_WGHT_KG_TRM
	Non university		University		Non university		University
Quantiles	R	SAS	R	SAS	R	SAS	R	SAS
Estimate
0.025	1.5367	1.5367	1.5291	1.5291	50.8669	50.8669	45.1422	45.1422
0.05	1.5452	1.5452	1.5691	1.5691	54.1563	54.1563	48.1974	48.1974
0.1	1.5769	1.5769	1.5958	1.5958	57.7347	57.7347	56.6144	56.6144
0.5	1.6729	1.6729	1.6915	1.6915	76.5711	76.5711	77.7776	77.7776
0.9	1.7564	1.7564	1.7677	1.7677	95.8922	95.8922	100.7855	100.7855
0.95	1.7846	1.7846	1.8053	1.8053	103.0741	103.0741	110.0363	110.0363
0.975	1.8024	1.8024	1.8391	1.8391	109.5934	109.5934	117.7170	117.7170
SE
0.025	0.0065	0.0138	0.0252	0.0424	1.2007	1.5157	7.8304	NA
0.05	0.0056	0.0078	0.0148	0.0144	1.1715	1.1681	3.2904	8.0046
0.1	0.0134	0.0134	0.0108	0.0112	1.7917	1.7500	4.0130	4.2362
0.5	0.0083	0.0084	0.0103	0.0103	1.1980	1.1927	1.1935	1.2661
0.9	0.0084	0.0083	0.0084	0.0082	1.7410	1.5701	4.4117	3.6180
0.95	0.0080	0.0081	0.0164	0.0153	3.0127	3.0355	4.2546	4.2032
0.975	0.0189	0.0144	0.0183	0.0167	7.3211	7.2938	5.5476	NA

Note:

SAS does not provide some standard errors since the quantiles are too extreme.