高莉菁
(暨南大學(xué) 經(jīng)濟(jì)學(xué)院統(tǒng)計(jì)學(xué)系,廣州 510632)
由于社會(huì)經(jīng)濟(jì)現(xiàn)象在不斷的發(fā)展變化,一次性抽樣調(diào)查已經(jīng)不足以及時(shí)反映總體的變化和發(fā)展,因此出現(xiàn)了連續(xù)性抽樣調(diào)查,通過(guò)連續(xù)性抽樣調(diào)查能夠得出對(duì)總體數(shù)字特征的更加可靠的估計(jì)。在連續(xù)性抽樣調(diào)查中,樣本輪換方法被廣泛采用,由于其既吸收了固體樣本調(diào)查和重復(fù)樣本調(diào)查的優(yōu)點(diǎn),又克服了兩者的缺點(diǎn),能夠反映總體單位特點(diǎn)的變化和總體結(jié)構(gòu)的變化。本文介紹幾種在輔助信息的總體均值未知的情況下,利用二階段輪換樣本連續(xù)抽樣來(lái)估計(jì)當(dāng)期總體均值的估計(jì)量,并通過(guò)數(shù)值模擬來(lái)比較其性質(zhì),并給出一些合理建議。
定理2Ti(i=1,2)的均方誤差線性近似為:
為了最小化式(14)和式(15)給出的均方誤差,需要考慮最優(yōu)的樣本輪換率。最小化式(14)和式(15)式中得到如下的二次方程,相應(yīng)的μi(i=1,2)如下:
表1 最優(yōu)的樣本輪換率以及相對(duì)效率
表1 最優(yōu)的樣本輪換率以及相對(duì)效率
0.9 0.576 207.17 147.54 0.62 287.3 204.61 0.7547 516.89 368.12 0.5247 195.77 139.43 0.5502 262.84 187.19 0.5647 422.18 300.67 0.4763 186.85 133.07 0.4972 245.87 175.1 0.4872 379.92 270.57 ρxz ρyz 0.7 0.8 ρyx μ^1 1 E(1)0.7 2 μ^1 E(1)1 E(2)2 0.9 1 μ^1 E(2)E(1)2 μ^1 E(1)0.71 E(2)E(2)2 E(1)μ^1 0.8 0.81 E(2)2 E(1)μ^1 0.91 E(2)2 μ^1 0.7 E(1)1 E(2)2 0.8 0.9 E(1)μ^11 E(2)2 μ^1 0.9 E(1)1 E(2)2 0.5 0.577 141.15 131.5 0.4745 182.64 170.15 0.3678 271.73 253.15 0.5441 138.66 129.18 0.4483 178.43 166.23 0.3443 263.85 245.81 0.5194 136.28 126.97 0.4287 174.54 162.61 0.3281 256.74 239.19 0.6 0.8514 148.1 132.99 0.6853 194.34 174.51*----0.6911 145.94 131.04 0.5459 189.58 170.23 0.4392 282.58 253.74 0.6167 143.36 128.73 0.492 184.87 166.01 0.375 273.6 245.68 0.7 0.0909 166.87 142.53 0.3046 214.27 183.02 0.3794 323.65 276.45*----*----0.3346 308.69 263.67*----0.9436 195.97 167.39 0.1919 296.63 253.37 0.8 0.4321 180.56 143.71 0.473 239.01 190.23 0.4744 372.87 296.77 0.3686 174.44 138.84 0.4231 227.5 181.07 0.427 346.99 276.18 0.2696 169.56 134.96 0.3573 218.19 173.66 0.3879 327.41 260.59
表2 最優(yōu)的樣本輪換率以及相對(duì)效率
表2 最優(yōu)的樣本輪換率以及相對(duì)效率
ρxz ρyz 0.7 ρyx μ^2 E(1)2 E(2)2 μ^2 0.7 0.8 E(1)2 E(2)2 0.9 2 μ^2 E(2)E(1)2 μ^2 0.7 E(1)2 E(2)2 E(1)μ^2 0.8 0.82 E(2)2 μ^2 0.9 E(1)2 E(2)2 μ^2 0.7 E(1)2 E(2)2 0.8 0.9 E(1)μ^2 2 E(2)2 μ^2 0.9 E(1)2 E(2)2 0.5 0.4166 152.86 142.41 0.3848 191.49 178.4 0.3249 277.33 258.38 0.4084 149.99 139.74 0.375 187.02 174.23 0.3136 269.3 250.89 0.4008 147.31 137.24 0.366 182.91 170.41 0.3036 262.07 244.15 0.6 0.444 162.41 145.84 0.4142 204.78 183.88 0.3563 299.13 268.61 0.4336 158.81 142.61 0.4015 199.05 178.74 0.3409 288.53 259.09 0.4242 155.5 139.63 0.3901 193.89 174.11 0.3277 279.25 250.75 0.7 0.4797 174.81 149.32 0.4542 222.55 190.09 0.4021 329.69 281.61 0.4659 170.05 145.25 0.4365 214.73 183.41 0.379 314.46 268.59 0.4536 165.77 141.59 0.4212 207.89 177.57 0.3601 301.7 257.7 0.8 0.5303 192.22 152.99 0.5147 248.91 198.11 0.4817 379.51 302.06 0.5102 185.32 147.5 0.4868 236.86 188.52 0.4391 353.34 281.23 0.4929 179.35 142.75 0.4641 226.93 180.61 0.408 333.51 265.45 0.9 0.6149 220.87 157.3 0.634 298.85 212.83 0.755 524.56 373.58 0.5788 208.71 148.64 0.5729 273.61 194.86 0.5679 429.28 305.73 0.5505 199.1 141.79 0.5314 256.04 182.35 0.4936 386.63 275.35
綜上所述,本文介紹的估計(jì)量能夠在估計(jì)精度上有較大提高,同時(shí)也可以減少調(diào)查成本,如果能夠在實(shí)踐中合理利用,將有利于推動(dòng)我國(guó)統(tǒng)計(jì)調(diào)查體系的改進(jìn)和完善,降低調(diào)查成本,提高統(tǒng)計(jì)數(shù)據(jù)的質(zhì)量,更好的為社會(huì)各界服務(wù)。
[1]Singh G N,Majhi D,Prasad S.Estimation of Population Mean in two-phase Successive Sampling[C].Dhanbad:Recent Advances in Information Technology(RAIT).1st International Conference on.2012.
[2]Singh G N,Priyanka K.On the Use of Chain-type Ratio to Difference Estimator in Successive Sampling[J].IJAMAS,2006,(5).
[3]Singh G N Kumari P.On the Use of Auxiliary information in Search of Good Rotation Patterns on Successive Occasions.[J].International Journal of Statistics&Economics,2007,(1).
[4]Singh G N,Priyanka K.Search of Good Rotation Patterns to Improve the Precision of Estimates at Current Occasion[J].Communications in Statistics—Theory and Methods,2008.37(3).
[5]Singh G N,Karna J P.Estimation of Population Mean on Current Occasion in two-occasion Successive Sampling[J].METRON,2009,LXVII(1).
[6]Singh G N,Prasad S.Some Estimators of Population Mean in two-occasion Rotation Patterns[J].Association For The Advancement Of Modeling And Simulation Techniques In Enterprises,2010,(12).
[7]Singh G N,Prasad S,Karna J.Some Classes of Estimators for population Mean at Current Occasion in two-occasion Successive Sampling[J].Journal of Statistical Research,2011,45(21).
[8]Singh G N,Prasad S.Some Rotation Patterns in two-Phase Sampling[J].Statistics in Transition,2011,(12).