5. Application to real data
A real data set is used to illustrate the efficiency of μˆ YMRSSk relative to μˆ YSRSk in estimating the population mean of the
heights (Y) of 348 students using the first and third quartiles of the weights (X). As we mentioned in the previous sections, in
this study the ranking is performed on the variable X. As shown in Table 7, the data are asymmetrically distributed, where
the skewnesses of X and Y are 1.1954 and 0.0289, respectively. Results are summarized in Table 8 for m = 2, 3, 4 with
q1 = 39 and q3 = 57.
Table 8 reveals that the MRSS estimators are more efficient than their counterparts obtained using SRS in estimating the
mean height of 348 students.
A.I. Al-Omari / Statistics and Probability Letters 82 (2012) 1883–1890 1889
Table 5
The efficiency of μˆ ′
YMRSSk with respect to μˆ ′
YSRSk and bias values for 2 ≤ n ≤ 6 using q1 = 1.32551.
ρ n = 2 n = 3 n = 4 n = 5 n = 6
0.99
Efficiency 1.5236 1.8815 2.0990 2.2571 2.3115
Bias (MRSS) 0.0276 0.0104 0.0055 0.0026 0.0013
Bias (SRS) 0.0402 0.0244 0.0200 0.0140 0.0110
0.90
Efficiency 1.3461 1.3485 1.3956 1.4309 1.4327
Bias (MRSS) 0.0275 0.0155 0.0086 0.0061 0.0029
Bias (SRS) 0.0666 0.0335 0.0239 0.0204 0.0252
0.80
Efficiency 1.2921 1.3461 1.3778 1.3783 1.3966
Bias (MRSS) 0.0444 0.0202 0.0094 0.0088 0.0042
Bias (SRS) 0.0800 0.0494 0.0266 0.0265 0.0214
0.70
Efficiency 1.2992 1.3741 1.4000 1.4210 1.4320
Bias (MRSS) 0.0569 0.0250 0.0149 0.0072 0.0067
Bias (SRS) 0.0909 0.0566 0.0421 0.0436 0.0316
−0.99
Efficiency 1.9125 2.6452 3.2399 4.0361 4.6400
Bias (MRSS) 0.2516 0.1056 0.0629 0.0368 0.0251
Bias (SRS) 0.4054 0.2361 0.1725 0.1400 0.1190
−0.90
Efficiency 1.7705 2.5588 3.0793 3.6510 4.1586
Bias (MRSS) 0.2299 0.0962 0.0526 0.0345 0.0234
Bias (SRS) 0.3697 0.2332 0.1701 0.1324 0.1134
−0.80
Efficiency 1.7062 2.4676 2.9151 3.3607 3.6982
Bias (MRSS) 0.2322 0.0981 0.0571 0.0373 0.0228
Bias (SRS) 0.3441 0.2288 0.1668 0.1161 0.1147
−0.70
Efficiency 1.6562 2.3426 2.6434 3.0677 3.3403
Bias (MRSS) 0.2135 0.0882 0.0532 0.0352 0.0236
Bias (SRS) 0.3354 0.2143 0.1649 0.1361 0.1134
−0.50
Efficiency 1.6094 2.0749 2.3528 2.6503 2.8724
Bias (MRSS) 0.1906 0.0852 0.0497 0.0335 0.0199
Bias (SRS) 0.2321 0.0978 0.1352 0.1093 0.0913
Table 6
The efficiency of μˆ ′
YMRSSk with respect to μˆ ′
YSRSk and bias values for 2 ≤ n ≤ 6 using q3 = 2.67449.
ρ n = 2 n = 3 n = 4 n = 5 n = 6
0.99
Efficiency 1.2264 1.4496 1.5148 1.5980 1.6649
Bias (MRSS) −0.0107 −0.0053 −0.0012 −0.0028 −0.0015
Bias (SRS) −0.0150 −0.0087 −0.0070 −0.0058 −0.0050
0.90
Efficiency 1.0231 1.0355 1.0377 1.0392 1.0399
Bias (MRSS) −0.0067 −0.0007 −0.0018 −0.0013 −0.0011
Bias (SRS) −0.0052 −0.0040 −0.0041 −0.0029 −0.0009
0.80
Efficiency 1.0152 1.0307 1.0367 1.0387 1.0395
Bias (MRSS) 0.0017 0.0023 0.0036 0.0009 0.0003
Bias (SRS) 0.0057 0.0041 0.0027 0.0019 0.0017
0.70
Efficiency 1.0364 1.0547 1.0594 1.0630 1.0651
Bias (MRSS) 0.0138 0.0069 0.0052 0.0025 −0.0010
Bias (SRS) 0.0153 0.0119 0.0079 0.0076 0.0075
−0.99
Efficiency 1.5725 2.4069 3.0392 3.6582 4.3594
Bias (MRSS) 0.1394 0.0602 0.0351 0.0227 0.0173
Bias (SRS) 0.2218 0.1447 0.1050 0.0794 0.0690
−0.90
Efficiency 1.5361 2.2719 2.7531 3.2682 3.6968
Bias (MRSS) 0.1312 0.0581 0.0329 0.0210 0.0139
Bias (SRS) 0.1925 0.1320 0.0940 0.0839 0.0575
−0.80
Efficiency 1.5094 2.1275 2.5226 2.9004 3.2423
Bias (MRSS) 0.1218 0.0561 0.0320 0.0225 0.0154
Bias (SRS) 0.1297 0.0461 0.0291 0.0281 0.0159
−0.70
Efficiency 1.4492 1.9612 2.3304 2.5769 2.8721
Bias (MRSS) 0.1200 0.0534 0.0327 0.0234 0.0146
Bias (SRS) 0.1805 0.1219 0.0757 0.0567 0.0607
−0.50
Efficiency 1.4043 1.7865 1.9721 2.1393 2.3026
Bias (MRSS) 0.1033 0.0422 0.0254 0.0222 0.0114
Bias (SRS) 0.1562 0.1041 0.0746 0.0555 0.0587
1890 A.I. Al-Omari / Statistics and Probability Letters 82 (2012) 1883–1890
Table 7
Summary statistics of data for 348 students.
Variable Mean Variance Skewness q1 q3
Weight (X) in kg 50.3017 275.7560 1.1954 39.0 57.0
Height (Y) in cm 152.3250 131.0560 0.0289
Ratio 3.0282
Correlation coefficient 0.6775
Table 8
The efficiency of estimating the population mean of the heights of 348 students.
Without double sampling With double sampling
m = 2 m = 3 m = 4 m = 2 m = 3 m = 4
Results with third quartile q1 = 39
Efficiency 1.23 1.73 1.87 1.26 1.72 1.85
MSE MRSS 172.19 86.10 61.71 171.35 86.84 61.36
MSE SRS 212.244 148.965 115.17 216.20 149.16 113.46
Results with third quartile q3 = 57
Efficiency 1.18 1.62 1.68 1.192 1.59 1.69
MSE MRSS 117.37 59.86 44.17 116.36 60.58 43.66
MSE SRS 139.18 96.01 74.39 138.70 96.60 73.70
6. Conclusions
In this paper, the problem of estimating the population mean is investigated for when the mean of the auxiliary variable
is known or unknown. Two SRS ratio estimators as well as two MRSS estimators are suggested, based on the first and third
quartiles of the auxiliary variable. We proved that for the first degree of approximation, the SRS and MRSS estimators are
unbiased. Also, the MSE of the MRSS estimators is less than the MSE of the SRS estimators, on the basis of the same sample
size. It turns out that when ρ < 0 the efficiency values are greater than when ρ > 0. Also, the efficiency is greater for q1
than for q3.