【摘要】：從不均等選擇概率的角度,提出兩類常見的權(quán)數(shù)調(diào)整類型及其調(diào)整方法:一是規(guī)模調(diào)整,使得樣本單元權(quán)數(shù)之和等于總體規(guī)模;二是結(jié)構(gòu)調(diào)整,使得樣本結(jié)構(gòu)和總體結(jié)構(gòu)一致,并構(gòu)造出加權(quán)調(diào)整的設(shè)計(jì)效應(yīng)模型,應(yīng)用于復(fù)雜樣本設(shè)計(jì)。案例分析顯示,加權(quán)調(diào)整往往導(dǎo)致設(shè)計(jì)效應(yīng)變大,帶來負(fù)的效應(yīng),但校準(zhǔn)調(diào)整能降低設(shè)計(jì)效應(yīng),提高估計(jì)精度。
[Abstract]:From the point of probability of unequal selection, two kinds of weight adjustment types and their adjustment methods are put forward: first, the size adjustment makes the sum of the weights of sample units equal to the overall size; The second is structural adjustment, which makes the sample structure and the overall structure consistent, and constructs a design effect model of weighted adjustment, which is applied to complex sample design. Case analysis shows that weighted adjustment often leads to large design effect and negative effect, but calibration adjustment can reduce design effect and improve estimation accuracy.
【作者單位】： 廣東工業(yè)大學(xué)管理學(xué)院;廣東工業(yè)大學(xué)大數(shù)據(jù)戰(zhàn)略研究院;
【基金】：廣東省哲學(xué)社會(huì)科學(xué)“十二五”規(guī)劃2015年度學(xué)科共建項(xiàng)目《基于大數(shù)據(jù)統(tǒng)計(jì)化的政府統(tǒng)計(jì)設(shè)計(jì)研究》(GD15XGL09)
【分類號(hào)】：O212

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 金勇進(jìn);張U，

本文編號(hào)：2451664