離散選擇模型-文庫(kù)吧資料
2024-10-08 19:01本頁(yè)面
【正文】 z P|z| [95% Conf. Interval] ? + ? potexp| .0325594 ? exp2 | .9973167 .0005639 .9962121 .9984225 ? grade | .9320947 .0299594 .8751866 .9927031 ? married | .2208633 .7632141 ? high | .4798005 ? 皮爾遜擬合優(yōu)度檢驗(yàn) ? lfit ? Logistic model for union, goodnessoffit test ? number of observations = 1000 ? number of covariate patterns = 556 ? Pearson chi2(550)= ? Prob chi2 = ? 表明模型充分地?cái)M合了數(shù)據(jù)。因變量 depvar。if you specify oute(),you must specify one new variable。 ? score:first is dLnLj/dln(xjb)。Prob chi2 = Ordered probit model ? oprobit depvar [varlist][weight][if exp][in range][,table robust cluster(varname) score(newvarlist) level() maximize_options] ? predict [type] newvarname [if exp][in range][,{p|xb|stdp}oute(oute) nooffset] ? Options:k:number of categories ? table showing how probability are puted from the fitted equation. ? cluster : specifies observations are independent across groups(clusters) but not necessary within groups。 t 檢驗(yàn)和 F檢驗(yàn) ? test potexp=grade ? ( 1) potexp grade = 0 ? chi2( 1) = 。綠色是指數(shù)偏效應(yīng) (probit)系數(shù) 。 ? 在樣本均值處的邊際概率效應(yīng)等于相應(yīng)的估計(jì)系數(shù)值乘以相應(yīng)的在樣本均值處的概率密度函數(shù)值 . 以教育為例的結(jié)果 ? scalar pdfxallbar=normden(e(xbar)) (注依次是 dprobit 之后沒有 at(.)。當(dāng)然可以指定其他值 。 ? e(par):fraction of success observed in data. ? e(dfdx):marginal effects. ? e(se_dfdx):standard error of marginal effects. 例子 :參與工會(huì)模型 ? 因變量 :參加工會(huì) union ? 自變量 :潛在經(jīng)驗(yàn) potexp,工作經(jīng)驗(yàn)的平方exp2,受教育年數(shù) grade,婚否 married,工會(huì)化程度 high. ? 分別采用線性概率模型 ,probit(包括mfx),dprobit,logit 模型進(jìn)行分析 ,結(jié)果如下 : LPM results ? reg union potexp exp2 grade married high /*LPM*/ ? ? union | Coef. Std. Err. t P|t| [95% Conf. Interval] ? + ? potexp| .0202288 .0038969 .0123916 .0276859 ? exp2 | .0000819 ? grade| .0051005 ? married| .0133428 .030001 .0722153 ? high| .1439396 .0256785 .0935492 .1943299 ? _cons| .1021368 .0749337 .2491832 ? Probit results ? Probit estimates Number of obs = 1000 LR chi2(5)= Prob chi2 = Log likelihood = Pseudo R2 = ? ? union | Coef. Std. Err. z P|z| [95% Conf. Interval] ? + ? potexp | .0835091 .0156087 .0529166 .1141016 ? exp2 | .0003179 ? grade | .0189089 married | .0622516 .1125836 .2829115 ? high | .5612953 .099662 .3659613 .7566292 ? _cons | .2958112 ? using mfx after probit estimation ? Marginal effects after probit 。e(b)指的是對(duì)應(yīng)的分布函數(shù)值 。 ? if for dummy,use ? Prediction: ? P:probability of a positive oute. ? Xb: calculate the linear predictioin. ? stdp:standard error of linear prediction. xibbxf )(? ? )( 01 bxbx ???Output after dprobit ? ereturn list (在 dprobit之后使用 )輸出各種結(jié)果 。問題在于當(dāng)多個(gè)選擇相似時(shí) ,例如交通方式的選擇 ,坐車 ,乘船 ,坐飛機(jī) 。 ? 因此要進(jìn)行標(biāo)準(zhǔn)化 : ? ? ? ???????????????????????????????????????????????????????????iiiiiiiiiixxxpxypyN I Dxy112*i2*i11*i2*||1。.1)1(11變化若干單位的影響當(dāng)然也可以研究發(fā)生比率的變化率不會(huì)變化無論怎么變化發(fā)生比都生比會(huì)相應(yīng)地減少每增加一個(gè)單位值時(shí)發(fā)生比會(huì)相應(yīng)地增加每增加一個(gè)單位值時(shí)發(fā)時(shí)的發(fā)生比之比時(shí)的發(fā)生比與實(shí)際上是又稱為調(diào)整發(fā)生比率所變化的倍數(shù)增加一個(gè)單位時(shí)發(fā)生比表示在的偏系數(shù)稱為自變量個(gè)發(fā)生比率發(fā)生比的變化率就是一連續(xù)自變量的發(fā)生比率xexexexexxeeea d j u s t e dA O Rxep a rt
語(yǔ)文相關(guān)推薦
鄂ICP備17016276號(hào)-1