3.1.4 加统计信息的一维表
. table male, contents (N wage mean wage sd wage p50 wage) 1 male 0 female N(wage) mean(wage) sd(wage) med(wage) female 4,219 10222.54 7379.265 8532.41 male 5,362 12572.05 8267.957 11119.54
. table male ownership, contents (N wage mean wage sd wage p50 wage) 1 male 0 ownership female SOE COE FOE POE female 2,699 385 83 903 11674.71 7213.589 12509.7 7317.338 7637.997 4694.328 7911.12 6144.484 10260 6060 11288.07 6000 male 3,769 299 127 1,031 13475 8926.686 16487.83 10147.28 7793.849 5751.082 9700.689 9459.863 12080.04 7472.5 13866 7900
tabulate varlist [,summarize]
. tab male, sum(wage) 1 male 0 Summary of wage female Mean Std. Dev. Freq. female 10222.544 7379.2647 4219 male 12572.054 8267.9574 5362 Total 11537.445 7974.3349 9581
统计表格tabstat
tabstat wage educ exper, statistics( count mean p50 skew kurt) column(statistics) . tabstat wage educ exper, stat(count mean p50 skew kurt) col(stat) variable N mean p50 skewness kurtosis wage 9581 11537.45 9996.24 2.562384 19.6333 educ 10105 11.42415 12 -.2533899 3.779897 exper 10031 20.11295 21 -.1663281 2.212653
3.1.5 统计检验
ttest 可以用来检验变量总体均值是否等于某一常数（H0: ?=?*），或检验两个均值是否相等（H0: ?1=?2）
26
. ttest wage, by(male)Two-sample t test with equal variances Group Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] female 4219 10222.54 113.6078 7379.265 9999.813 10445.28 male 5362 12572.05 112.9106 8267.957 12350.7 12793.4 combined 9581 11537.45 81.46837 7974.335 11377.75 11697.14 diff -2349.509 162.3521 -2667.754 -2031.265 diff = mean(female) - mean(male) t = -14.4717Ho: diff = 0 degrees of freedom = 9579 Ha: diff < 0 Ha: diff != 0 Ha: diff > 0 Pr(T < t) = 0.0000 Pr(|T| > |t|) = 0.0000 Pr(T > t) = 1.0000
3.1.6 数据画图
kdensity 和 twoway kdensity可以用来画变量的分布，比如工资的分布 kdensity wage
twoway kdensity wage, by(male)
kdensity wage.00002.00004.00006.00008femalemale0050000100000150000050000100000150000xGraphs by 1 male 0 female
Twoway (kdensity wage if male)(kdensity wage if !male), legend(label(1 male) label (2 female))
27
kdensity wage.00002.00004.00006.000080050000 malex100000 female150000
3.2
回归分析
3.2.1 相关分析
correlate 相关系数，仅对所有变量同时取值时，不考虑缺失值
pwcorr 相关系数，两两计算相关系数，考虑缺失值，可以加参数obs、sig（显著性）
. cor wage educ exper(obs=9518) wage educ exper wage 1.0000 educ 0.2747 1.0000 exper 0.1867 -0.2218 1.0000
. pwcorr wage educ exper, obs sig wage educ exper wage 1.0000 9581 educ 0.2754 1.0000 0.0000 9581 10105 exper 0.1867 -0.2036 1.0000 0.0000 0.0000 9518 10031 1003128
wage educ exper wage 1.0000 educ 0.2747 1.0000 exper 0.1867 -0.2218 1.0000. pwcorr wage educ exper, obs sig wage educ exper wage 1.0000 9581 educ 0.2754 1.0000 0.0000 9581 10105 exper 0.1867 -0.2036 1.0000 0.0000 0.0000 9518 10031 10031
3.2.2 线性回归
regress depvar [indepvars] [if] [in] [weight] [, options]
regress depvar [indepvars] [if] [in] [weight] [, options] options Description Model noconstant suppress constant term hascons has user-supplied constant tsscons compute total sum of squares with constant; seldom used SE/Robust vce(vcetype) vcetype may be ols, robust, cluster clustvar, bootstrap, jackknife, hc2, or hc3 Reporting level(#) set confidence level; default is level(95) beta report standardized beta coefficients eform(string) report exponentiated coefficients and label as string depname(varname) substitute dependent variable name; programmer's option display_options control column formats, row spacing, line width, and display of omitted variables and base and empty cells noheader suppress table header notable suppress coefficient header plus make table extendable mse1 force mean squared error to 1 coeflegend display legend instead of statistics
29
. reg lwage educ exper, vce(robust)Linear regression Number of obs = 9518 F( 2, 9515) = 780.95 Prob > F = 0.0000 R-squared = 0.1582 Root MSE = .69448 Robust lwage Coef. Std. Err. t P>|t| [95% Conf. Interval] educ .0875659 .0026086 33.57 0.000 .0824525 .0926793 exper .0233035 .0008106 28.75 0.000 .0217146 .0248924 _cons 7.641395 .0400242 190.92 0.000 7.562939 7.719851
ereturn list
help regress postestimation 3.2.3 假设检验 Wald test test educ
联合假设 test educ exper 有约束回归 cnsreg constraint 1 educ=.5
cnsreg lwage educ exper, vce(robust)
3.2.4 估计结果呈现
estimate store 可以将e()中的回归结果保存起来。
estimate table 可以将estimate store的结果以表格的形式呈现出来。 estimate stats 可以将样本容量和似然函数值统计出来。 qui: reg lwage educ exper expersq est store m1 est table m1
qui: reg lwage educ exper expersq male est store m2 est table m1 m2 est stats m1 m2
est table m1 m2, b(%9.4f) se stats(N r2 F ll)
30