斯托克,沃森計(jì)量經(jīng)濟(jì)學(xué)第七章實(shí)證練習(xí)stata

E7.1user,c:\Users\as口3\口總￡吒匸0總1心百30user,c:\Users\as口3\口總￡吒匸0總1心百30日-dica^teotessaih.Eaue^t:□上:口田亡"建立日血己對(duì)日共的叵歸eatstoremlrsgxessmtimag@feraaletiaCheloxrarutium匸//.逹立同n己對(duì)sgm『female^UtKachslox9\f[h]J0eatstoreTt2eattBb朮丄m2uaingl.docfatari*0.10**D.05***0.01)regilace//導(dǎo)出以上砲丁回舊：>井且得到七統(tǒng)計(jì)量(表丄)ouxxsgiusing"'S.docr\eKcalellevel|S5)replace//-^出回歸中汩巳系癡的05去登‘專區(qū)舊」(疾2Jourx^g^[n2n2]口sing3-dac^adcLsrat(A2us^e，：1R2*mi二二n；i■Fz^plmciie/f尋出回.!■日fltf標(biāo)；隹羞?R-sq-uazsjustsdR2(^3?E7.2u日亡"C:Wasr^Xa^taaXDc5ktciF-\TenchlngRatin^j,dcBuyegress匚口beauty,rcb-uat/Z^^covsir3s^TalXd'beaHty&\J[ElSeatstoreraiLo?trrg2usinij.doc"rexcelci“兀丄t勺創(chuàng)replace//出Ba口ty對(duì)_miiE共二帀丄繪應(yīng)的9S%SfaEIWTigresscur^^valbeautyminorityagefewalecneaT^ditintronnenijliah,工ot<udty/'建立csour：jit_u￥d丄對(duì)■復(fù)他控剖多年1丁]■回.!啟石eatstorem2esttabmlm2uaing2,docrscar110Bi0xr0.05TTT0-01)乂口“匚匕"導(dǎo)出H上兩個(gè)回歸丁并且袒到t堂Tigresscur^^valbeautyminorityfernaleonecreditnnen^lJ-3hrrcbu3t//^Z￡coT3r35!_5!val7d'JSlJacre^incioD^Hj^Ho?treg2UBing.doc"rexcelci勺印replace//出回歸井導(dǎo)到合理的9￡電孟?虧區(qū)間E7.3use"■匚：\Users\abub\DebkrcpXCDllegaDistancewesr?dtanregxe^sdiat^robust己呂匸storerxlregresseddist.fsmoleblackh.ispsdc:bytestdadcc11nam匚口J_]Lovnh口m舌urban匚ue305t-wm±gB口t.uiti口口i.n匚口mehireststorer%2regresseildisur&maletilack:Ms&aKlebyies匚dadccllraomcollD-wnhomecue30弓匚wmCgEOincomehl,edIjue匚satstoren3xegzeas已口listb^rescdadacllj?jm.collcneEDzulcionlncoinj&lLizob-asr-^itort厲弓Tab-catE7.4use"'C:\.User□\aaua\DeaktopXGrowtti?dtandxopifcoLuitrynanie&sFrHalca"/./剔球馬其他的敎兎口二己口uHtgg匚cwthtrfldeshaxEyeffixaschciolrevc-oupafxnatxonsjgdp6<froLust//導(dǎo)出M歸咱關(guān)數(shù)捋E7.1(1)ahe(2)aheVARIABLESaheage0.605***0.585***age0.585***(15.02)(16.02)(0.514-0.657)female-3.664***female-3.664***(-4.071--3.257)(-17.65)bachelor8.083***(7.666-8.500)bachelor8.083***Constant-0.636(38.00)(-2.759-1.487)_cons1.082-0.636Observations7,711(0.93)(-0.59)R-squared0.200(表2)Robustciinparentheses***p<0.01,**p<0.05,*p<0.1N77117711tstatisticsinparentheses*p<0.10,**p<0.05,***p<0.01(表1)建立ahe對(duì)age的回歸。截距估計(jì)值是1.082,斜率估計(jì)值是0.605。①建立ahe對(duì)age，female和bachelor的回歸oAge對(duì)收入的效應(yīng)的估計(jì)值是0.585。②age回歸系數(shù)的95%置信區(qū)間：(0.514,0.657)⑶設(shè)匕臥⑵嘰⑴=0日鞏，⑵咆⑴工0由表3,得SE,SE(Ba,(2)-Ba,⑴)二”(0.0403)2+(0.0365)2二0.054t=(0.605-0.585)/0.054=0.37<1.96所以不拒絕原假設(shè)，即在5%顯著水平下age對(duì)ahe的效應(yīng)估計(jì)沒有顯著差異，所以(1)中的回歸沒有遭遇遺漏變量偏差。(4)Bob'spredictedahe=0.585x26-3.664x0+8.083x0-0.636二$14.574Alexis'spredictedahe=0.585x30-3.664x1+8.083x1-0.636二$21.3335）VARIABLES⑴m1ahe(2)m2aheage0.605***0.585***(0.0403)(0.0365)female-3.664***(0.208)bachelor8.083***(0.213)Constant1.082-0.636(1.167)(1.083)Observations7,7117,711R-squared0.0290.200AjustedR20.1990.199Robuststandarderrorsinparentheses

***p<0.01,**p<0.05,*p<0.1（表3）通過比較，（2）回歸的標(biāo)準(zhǔn)誤更小，且R-squared和AjustedR2更接近1,因此（2）中回歸的擬合效果更好。⑵中R-squared和AjustedR2如此相近是因?yàn)闃颖玖孔銐虼螅╪二7711）。（6）對(duì)FeiTia?Ho：BFemale=°H1:^Female^0由表1得，t=-17.65<-1.96,所以在5%的顯著水平下拒絕原假設(shè)，即不可以剔除female。對(duì)Bachelor.H0：PBachelor=°H1：^Bachelor^0由表1得,t=38.00>1.96,所以在5%的顯著水平下拒絕原假設(shè),即不可以剔除bachelor。對(duì)Female和Bachelor:H0：PFemale+^Bachelor=0H1:^Female+^Bachelor^0由表3得SE,SE（pFemale+pBachelor）=V（0.208）2+（0.213）2=0.0886t=（8.083-3.664）/0.0886=49.88>1.96所以在5%的顯著水平下拒絕原假設(shè)，即回歸中不能同時(shí)剔除Female和Bachelor。（7）這兩個(gè)條件是.至少有一個(gè)回歸變量必須與遺漏變量相關(guān)；遺漏變量必須是因變量丫的決定因素。遺漏變量year（從業(yè)時(shí)間），它既與回歸變量age（年齡）有關(guān)，又與因變量ahe（平均每小時(shí)收入）有關(guān)。所以這兩個(gè)條件成立,這個(gè)多元回歸遭遇了遺漏變量偏差。

E7.2VARIABLES(1)courseeval(1)(2)course_evalcourse_evalbeauty0.133***(0.0695-0.197)beauty0.133***0.159***Constant3.998***(4.12)(5.19)(3.948-4.048)minority-0.169**Observations463(-2.50)R-squared0.036-0.00195age（表1）Robustciinparentheses(-0.75)***p<0.01,**p<0.05,*p<0.1female-0.183***(-3.51)onecredit0.633***(5.87)intro0.00795(0.14)nnenglish-0.244**(-2.54)_cons3.998***(157.73)4.169***(29.98)N463463tstatisticsinparentheses(表2)*p<0.10,**p<0.05,***p<0.01⑴建立course_eval對(duì)beauty的回歸，course_eval=0.133*beauty+3.998由表1得，Beauty對(duì)course_eval效應(yīng)的95%置信區(qū)間：(0.0695,0.197)⑵對(duì)age：%：際=0H.際工。由表2得，-1.96<t=-0.75v1.96,所以在5%的顯著水平下不能拒絕原假設(shè)，即可以剔除age。對(duì)intro:H0：Pintro=0H1：Pintro^0由表2得,t=0.14<1.96,所以在5%的顯著水平下不能拒絕原假設(shè)，即可以剔除intro。對(duì)age和intro:%:際+際ro=°H1:際+際「。工0SE(Page+Pintro)=V(0?0026218)2+(0.0565)2=0.0567t=(0.00795-0.00195)/0.0567=0.106<1.96所以在5%的顯著水平下不能拒絕原假設(shè)，即回歸中可以同時(shí)剔除age和intro。我認(rèn)為應(yīng)該包含在回歸中的控制變量:beauty,minority,female,onecredit,nnenglishVARIABLES⑴courseevalbeauty0.166***(0.104-0.228)minority-0.165**(-0.299--0.0309)female-0.174***(-0.271--0.0769)onecredit0.641***(0.451-0.831)nnenglish-0.248***(-0.431--0.0655)Constant4.072***(4.008-4.136)Observations463R-squared0.155Robustciinparentheses***p<0.01,**p<0.05,*p<0.1(表3Beauty對(duì)course_eval效應(yīng)的合理的95%置信區(qū)間是(0.104,0.228)conscons13.86***9.233***9.453***9.755***E7.3(1)ed(2)ed(3)ed(4)eddist-0.0677***-0.0418**-0.0517***-0.0402**(-3.57)(-2.10)(-2.75)(-2.09)female0.0547(0.55)0.0614(0.62)black0.09830.133(0.54)(0.76)hispanic0.1870.213*(1.63)(1.85)bytest0.0738***0.0751***0.0723***(11.40)(11.67)(11.62)dadcoll0.478***0.477***0.445***(3.55)(3.56)(3.32)momcoll0.368**0.368**0.358**(2.24)(2.26)(2.21)ownhome0.1970.179(1.55)(1.41)urban-0.0811(-0.63)cue800.0444**0.0433**0.0476**(1.97)(1.97)(2.10)stwmfg800.0237-0.0305(0.54)(-0.76)tuition-0.528**-0.511**(-2.18)(-2.39)incomehi0.410***0.436***0.408***(3.39)(3.60)(3.47)(201.37)(17.76)(17.98)(28.41)943943943943(201.37)(17.76)(17.98)(28.41)943943943943tstatisticsinparentheses*p<0.10,**p<0.05,***p<0.01⑴建立ed對(duì)dist的回歸，得ed=-0.0677*dist+13.86回歸中dist的系數(shù)的95%置信區(qū)間為：-0.0677±1.96x0.0189724,即（-0.1048934，-0.0304273）根據(jù)該組織的聲稱，它贊同的dist的系數(shù)為-0.15/2=-0.075，-0.075包含于回歸中dist的系數(shù)的95%置信區(qū)間（-0.1048934，-0.0304273），因此該組織倡導(dǎo)的言論與回歸估計(jì)一致。2）上表中，（1）的構(gòu)造是簡(jiǎn)單設(shè)定形式，（4）的構(gòu)造是基礎(chǔ)設(shè)定形式，（2）和（3）的設(shè)定是基礎(chǔ)設(shè)定形式的幾種修正形式?；A(chǔ)設(shè)定形式中dist對(duì)ed的效應(yīng)為-0.0402,與簡(jiǎn)單設(shè)定形式的-0.0677差異較大?；A(chǔ)設(shè)定形式的修正形式中，即加入或剔除某些控制變量，dist對(duì)ed的效應(yīng)變化不大。構(gòu)造的dist對(duì)ed的效應(yīng)的95%置信區(qū)間為（-0.0778381,-0.0024972）,（2）和⑶構(gòu)造的dist的系數(shù)包含于內(nèi)。⑶這個(gè)結(jié)論與⑵中建立的回歸一致。因?yàn)棰茦?gòu)建的回歸中Bblack和Hispanic都為正數(shù)，且數(shù)值不小。E7.4LinearregressiongrowthCoef.St.Err.t-valuep-value[95%ConfInterval]Sigtradeshare1.3410.8821.520.134-0.4253.106yearsschool0.5640.1294.360.0000.3050.823***rev_coups-2.1500.875-2.460.017-3.901-0.400**assasinations0.3230.3800.850.400-0.4391.084rgdp600.0000.000-3.800.000-0.0010.000***Constant0.6270.8690.720.474-1.1132.367Meandependentvar1.869SDdependentvar1.816R-squared0.291Numb