为什么用不同的交互指令做出来的结果有差异？

1. . clear
   
2. 
   
3. . use hh3
   
4. 
   
5. . regress logg1018 a2000 labor2 children retire hh_income house houseloan e2002 c7001 savings a2003 age a2012 a2015 a2024 f2021 wo
   
6. > rk a3003 workyear worktime f1001 f2001 f3001 cinsurance region a2022rural [pweight=swgt]
   
7. (sum of wgt is 1.2329e+08)
   
8. 
   
9. Linear regression Number of obs = 791
   
10. F( 26, 764) = 7.23
    
11. Prob > F = 0.0000
    
12. R-squared = 0.2804
    
13. Root MSE = 1.0826
    
14. 
    
15. ------------------------------------------------------------------------------
    
16. | Robust
    
17. logg1018 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
    
18. -------------+----------------------------------------------------------------
    
19. a2000 | -.0883657 .0890625 -0.99 0.321 -.2632019 .0864706
    
20. labor2 | .0474589 .0907285 0.52 0.601 -.1306478 .2255656
    
21. children | .1295626 .1023531 1.27 0.206 -.071364 .3304892
    
22. retire | .1748803 .1194552 1.46 0.144 -.059619 .4093797
    
23. hh_income | 1.38e-06 6.53e-07 2.11 0.035 9.84e-08 2.66e-06
    
24. house | -.4324905 .1565911 -2.76 0.006 -.7398905 -.1250906
    
25. houseloan | -.2520211 .1589961 -1.59 0.113 -.5641422 .0601
    
26. e2002 | .0779077 .1429656 0.54 0.586 -.2027443 .3585596
    
27. c7001 | .5523863 .1269436 4.35 0.000 .3031866 .801586
    
28. savings | -.0677573 .0622234 -1.09 0.277 -.1899065 .0543918
    
29. a2003 | .165831 .1305863 1.27 0.205 -.0905196 .4221816
    
30. age | .003721 .0075394 0.49 0.622 -.0110795 .0185215
    
31. a2012 | .1097514 .0450151 2.44 0.015 .0213834 .1981194
    
32. a2015 | -.0821773 .0582571 -1.41 0.159 -.1965404 .0321858
    
33. a2024 | .0510015 .05941 0.86 0.391 -.0656247 .1676278
    
34. f2021 | -.1628977 .0615866 -2.65 0.008 -.2837968 -.0419986
    
35. work | .1516389 .1904025 0.80 0.426 -.2221353 .5254131
    
36. a3003 | -.1163044 .218655 -0.53 0.595 -.5455404 .3129316
    
37. workyear | .016017 .0067794 2.36 0.018 .0027084 .0293255
    
38. worktime | -.0000854 .0006073 -0.14 0.888 -.0012775 .0011067
    
39. f1001 | .057384 .0754563 0.76 0.447 -.0907424 .2055103
    
40. f2001 | -.1228524 .1916797 -0.64 0.522 -.4991339 .253429
    
41. f3001 | .2420348 .1267654 1.91 0.057 -.006815 .4908846
    
42. cinsurance | .11353 .1491954 0.76 0.447 -.1793515 .4064116
    
43. region | .0349508 .0868177 0.40 0.687 -.1354787 .2053803
    
44. a2022rural | -.1901757 .1768726 -1.08 0.283 -.5373896 .1570382
    
45. _cons | 7.041723 .7055126 9.98 0.000 5.65675 8.426697
    
46. ------------------------------------------------------------------------------
    
47. 
    
48. . xi: regress logg1018 a2000 labor2 children retire hh_income house houseloan e2002 c7001 savings a2003 age a2012 a2015 a2024 f202
    
49. > 1 work a3003 workyear worktime f1001 f2001 f3001 cinsurance region i.rural*a2022 [pweight=swgt]
    
50. i.rural _Irural_0-1 (naturally coded; _Irural_0 omitted)
    
51. i.rural*a2022 _IrurXa2022_# (coded as above)
    
52. (sum of wgt is 1.2329e+08)
    
53. 
    
54. Linear regression Number of obs = 791
    
55. F( 28, 762) = 6.91
    
56. Prob > F = 0.0000
    
57. R-squared = 0.2882
    
58. Root MSE = 1.0782
    
59. 
    
60. -------------------------------------------------------------------------------
    
61. | Robust
    
62. logg1018 | Coef. Std. Err. t P>|t| [95% Conf. Interval]
    
63. --------------+----------------------------------------------------------------
    
64. a2000 | -.0820923 .0845383 -0.97 0.332 -.2480478 .0838632
    
65. labor2 | .036054 .0886534 0.41 0.684 -.1379799 .2100879
    
66. children | .1264729 .0993262 1.27 0.203 -.0685125 .3214583
    
67. retire | .160383 .1172399 1.37 0.172 -.0697687 .3905346
    
68. hh_income | 1.42e-06 6.71e-07 2.12 0.034 1.06e-07 2.74e-06
    
69. house | -.3882593 .1641245 -2.37 0.018 -.7104493 -.0660694
    
70. houseloan | -.2305242 .1524171 -1.51 0.131 -.5297315 .0686832
    
71. e2002 | .0854729 .140618 0.61 0.543 -.1905718 .3615176
    
72. c7001 | .5196565 .1277123 4.07 0.000 .2689468 .7703662
    
73. savings | -.0660838 .0594367 -1.11 0.267 -.182763 .0505954
    
74. a2003 | .1802882 .1279389 1.41 0.159 -.0708664 .4314428
    
75. age | .0047518 .0071083 0.67 0.504 -.0092025 .018706
    
76. a2012 | .1280514 .0537575 2.38 0.017 .022521 .2335818
    
77. a2015 | -.0654256 .0591491 -1.11 0.269 -.1815402 .050689
    
78. a2024 | .0365402 .0595249 0.61 0.539 -.0803121 .1533925
    
79. f2021 | -.1597722 .0606721 -2.63 0.009 -.2788764 -.0406679
    
80. work | .178329 .1886879 0.95 0.345 -.1920808 .5487387
    
81. a3003 | -.1254408 .2191348 -0.57 0.567 -.5556205 .3047389
    
82. workyear | .0171254 .0067552 2.54 0.011 .0038644 .0303863
    
83. worktime | -.000036 .0005968 -0.06 0.952 -.0012077 .0011356
    
84. f1001 | .0530592 .0734673 0.72 0.470 -.0911632 .1972816
    
85. f2001 | -.1327798 .1901254 -0.70 0.485 -.5060116 .240452
    
86. f3001 | .24055 .1261088 1.91 0.057 -.007012 .488112
    
87. cinsurance | .0904654 .1478294 0.61 0.541 -.1997359 .3806667
    
88. region | .0639742 .0800329 0.80 0.424 -.0931369 .2210854
    
89. _Irural_1 | -.4517703 .303656 -1.49 0.137 -1.047872 .1443315
    
90. a2022 | .0788755 .1641699 0.48 0.631 -.2434036 .4011545
    
91. _IrurXa2022_1 | .1830922 .3539682 0.52 0.605 -.5117764 .8779608
    
92. _cons | 6.752954 .7906833 8.54 0.000 5.200778 8.30513
    
93. -------------------------------------------------------------------------------

复制代码

上面是你第一个命令和第二个命令的结果，从结果的变量个数你都可以看出，变量个数
都是不想等的，所以结果必然不同。第一个表只有a2022rural ，而第二个表有
_Irural_1，a2022，_IrurXa2022_1。你可以通过看数据文件，里面可以看到_Irural_1
和_IrurXa2022_1的具体数值。

下面两个语句是等价的：

1. . regress logg1018 a2000 labor2 children retire hh_income house houseloan e2002 c7001 savings a2003 age a2012 a2015 a2024 f2021 work a3003 workyear worktime f1001 f2001 f3001 cinsurance region a2022#rural [pweight=swgt]
   
2. (sum of wgt is   1.2329e+08)
   
3. 
   
4. Linear regression                                      Number of obs =     791
   
5.                                                        F( 28,   762) =    6.91
   
6.                                                        Prob > F      =  0.0000
   
7.                                                        R-squared     =  0.2882
   
8.                                                        Root MSE      =  1.0782
   
9. 
   
10. ------------------------------------------------------------------------------
    
11.              |               Robust
    
12.     logg1018 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
    
13. -------------+----------------------------------------------------------------
    
14.        a2000 |  -.0820923   .0845383    -0.97   0.332    -.2480478    .0838632
    
15.       labor2 |    .036054   .0886534     0.41   0.684    -.1379799    .2100879
    
16.     children |   .1264729   .0993262     1.27   0.203    -.0685125    .3214583
    
17.       retire |    .160383   .1172399     1.37   0.172    -.0697687    .3905346
    
18.    hh_income |   1.42e-06   6.71e-07     2.12   0.034     1.06e-07    2.74e-06
    
19.        house |  -.3882593   .1641245    -2.37   0.018    -.7104493   -.0660694
    
20.    houseloan |  -.2305242   .1524171    -1.51   0.131    -.5297315    .0686832
    
21.        e2002 |   .0854729    .140618     0.61   0.543    -.1905718    .3615176
    
22.        c7001 |   .5196565   .1277123     4.07   0.000     .2689468    .7703662
    
23.      savings |  -.0660838   .0594367    -1.11   0.267     -.182763    .0505954
    
24.        a2003 |   .1802882   .1279389     1.41   0.159    -.0708664    .4314428
    
25.          age |   .0047518   .0071083     0.67   0.504    -.0092025     .018706
    
26.        a2012 |   .1280514   .0537575     2.38   0.017      .022521    .2335818
    
27.        a2015 |  -.0654256   .0591491    -1.11   0.269    -.1815402     .050689
    
28.        a2024 |   .0365402   .0595249     0.61   0.539    -.0803121    .1533925
    
29.        f2021 |  -.1597722   .0606721    -2.63   0.009    -.2788764   -.0406679
    
30.         work |    .178329   .1886879     0.95   0.345    -.1920808    .5487387
    
31.        a3003 |  -.1254408   .2191348    -0.57   0.567    -.5556205    .3047389
    
32.     workyear |   .0171254   .0067552     2.54   0.011     .0038644    .0303863
    
33.     worktime |   -.000036   .0005968    -0.06   0.952    -.0012077    .0011356
    
34.        f1001 |   .0530592   .0734673     0.72   0.470    -.0911632    .1972816
    
35.        f2001 |  -.1327798   .1901254    -0.70   0.485    -.5060116     .240452
    
36.        f3001 |     .24055   .1261088     1.91   0.057     -.007012     .488112
    
37.   cinsurance |   .0904654   .1478294     0.61   0.541    -.1997359    .3806667
    
38.       region |   .0639742   .0800329     0.80   0.424    -.0931369    .2210854
    
39.              |
    
40. a2022#rural |
    
41.         0 1  |  -.4517703    .303656    -1.49   0.137    -1.047872    .1443315
    
42.         1 0  |   .0788755   .1641699     0.48   0.631    -.2434036    .4011545
    
43.         1 1  |  -.1898026   .2035583    -0.93   0.351    -.5894043    .2097991
    
44.              |
    
45.        _cons |   6.752954   .7906833     8.54   0.000     5.200778     8.30513
    
46. ------------------------------------------------------------------------------
    
47. 
    
48. . regress logg1018 a2000 labor2 children retire hh_income house houseloan e2002 c7001 savings a2003 age a2012 a2015 a2024 f2021 work a3003 workyear worktime f1001 f2001 f3001 cinsurance region i.a2022#i.rural [pweight=swgt]
    
49. (sum of wgt is   1.2329e+08)
    
50. 
    
51. Linear regression                                      Number of obs =     791
    
52.                                                        F( 28,   762) =    6.91
    
53.                                                        Prob > F      =  0.0000
    
54.                                                        R-squared     =  0.2882
    
55.                                                        Root MSE      =  1.0782
    
56. 
    
57. ------------------------------------------------------------------------------
    
58.              |               Robust
    
59.     logg1018 |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
    
60. -------------+----------------------------------------------------------------
    
61.        a2000 |  -.0820923   .0845383    -0.97   0.332    -.2480478    .0838632
    
62.       labor2 |    .036054   .0886534     0.41   0.684    -.1379799    .2100879
    
63.     children |   .1264729   .0993262     1.27   0.203    -.0685125    .3214583
    
64.       retire |    .160383   .1172399     1.37   0.172    -.0697687    .3905346
    
65.    hh_income |   1.42e-06   6.71e-07     2.12   0.034     1.06e-07    2.74e-06
    
66.        house |  -.3882593   .1641245    -2.37   0.018    -.7104493   -.0660694
    
67.    houseloan |  -.2305242   .1524171    -1.51   0.131    -.5297315    .0686832
    
68.        e2002 |   .0854729    .140618     0.61   0.543    -.1905718    .3615176
    
69.        c7001 |   .5196565   .1277123     4.07   0.000     .2689468    .7703662
    
70.      savings |  -.0660838   .0594367    -1.11   0.267     -.182763    .0505954
    
71.        a2003 |   .1802882   .1279389     1.41   0.159    -.0708664    .4314428
    
72.          age |   .0047518   .0071083     0.67   0.504    -.0092025     .018706
    
73.        a2012 |   .1280514   .0537575     2.38   0.017      .022521    .2335818
    
74.        a2015 |  -.0654256   .0591491    -1.11   0.269    -.1815402     .050689
    
75.        a2024 |   .0365402   .0595249     0.61   0.539    -.0803121    .1533925
    
76.        f2021 |  -.1597722   .0606721    -2.63   0.009    -.2788764   -.0406679
    
77.         work |    .178329   .1886879     0.95   0.345    -.1920808    .5487387
    
78.        a3003 |  -.1254408   .2191348    -0.57   0.567    -.5556205    .3047389
    
79.     workyear |   .0171254   .0067552     2.54   0.011     .0038644    .0303863
    
80.     worktime |   -.000036   .0005968    -0.06   0.952    -.0012077    .0011356
    
81.        f1001 |   .0530592   .0734673     0.72   0.470    -.0911632    .1972816
    
82.        f2001 |  -.1327798   .1901254    -0.70   0.485    -.5060116     .240452
    
83.        f3001 |     .24055   .1261088     1.91   0.057     -.007012     .488112
    
84.   cinsurance |   .0904654   .1478294     0.61   0.541    -.1997359    .3806667
    
85.       region |   .0639742   .0800329     0.80   0.424    -.0931369    .2210854
    
86.              |
    
87. a2022#rural |
    
88.         0 1  |  -.4517703    .303656    -1.49   0.137    -1.047872    .1443315
    
89.         1 0  |   .0788755   .1641699     0.48   0.631    -.2434036    .4011545
    
90.         1 1  |  -.1898026   .2035583    -0.93   0.351    -.5894043    .2097991
    
91.              |
    
92.        _cons |   6.752954   .7906833     8.54   0.000     5.200778     8.30513
    
93. ------------------------------------------------------------------------------

复制代码

1. Title
   
2. 
   
3.     [U] 11.4.3 Factor variables
   
4. 
   
5. 
   
6. Description
   
7. 
   
8.     Factor variables are extensions of varlists of existing variables.  When a command allows factor variables, in addition to typing variable names from your data, you can
   
9.     type factor variables, which might look like
   
10. 
    
11.         i.varname
    
12. 
    
13.         i.varname#i.varname
    
14. 
    
15.         i.varname#i.varname#i.varname
    
16. 
    
17.         i.varname##i.varname
    
18. 
    
19.         i.varname##i.varname##i.varname
    
20. 
    
21.     Factor variables create indicator variables from categorical variables, interactions of indicators of categorical variables, interactions of categorical and continuous
    
22.     variables, and interactions of continuous variables (polynomials).  They are allowed with most estimation and postestimation commands, along with a few other commands.
    
23. 
    
24.     There are four factor-variable operators:
    
25. 
    
26.          Operator  Description
    
27.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
28.          i.        unary operator to specify indicators
    
29.          c.        unary operator to treat as continuous
    
30.          #         binary operator to specify interactions
    
31.          ##        binary operator to specify factorial interactions
    
32.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
33. 
    
34.     The indicators and interactions created by factor-variable operators are referred to as virtual variables.  They act like variables in varlists but do not exist in the
    
35.     dataset.
    
36. 
    
37.     Categorical variables to which factor-variable operators are applied must contain nonnegative integers with values in the range 0 to 32,740, inclusive.
    
38. 
    
39.     Factor variables may be combined with the L. and F. time-series operators.
    
40. 
    
41. 
    
42. Remarks
    
43. 
    
44.     Remarks are presented under the following headings:
    
45. 
    
46.         Basic examples
    
47.         Base levels
    
48.         Selecting levels
    
49.         Applying operators to a group of variables
    
50. 
    
51. 
    
52. Basic examples
    
53. 
    
54.     Here are some examples of use of the operators:
    
55. 
    
56.          Factor            
    
57.          specification     Result
    
58.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
59.          i.group           indicators for levels of group
    
60. 
    
61.          i.group#i.sex     indicators for each combination of levels of group and sex, a two-way interaction
    
62. 
    
63.          group#sex         same as i.group#i.sex
    
64. 
    
65.          group#sex#arm     indicators for each combination of levels of group, sex, and arm, a three-way interaction
    
66. 
    
67.          group##sex        same as i.group i.sex group#sex
    
68. 
    
69.          group##sex##arm   same as i.group i.sex i.arm group#sex group#arm sex#arm group#sex#arm
    
70. 
    
71.          sex#c.age         two variables -- age for males and 0 elsewhere, and age for females and 0 elsewhere; if age is also in the model, one of the two virtual variables
    
72.                              will be treated as a base
    
73. 
    
74.          sex##c.age        same as i.sex age sex#c.age
    
75. 
    
76.          c.age             same as age
    
77. 
    
78.          c.age#c.age       age squared
    
79. 
    
80.          c.age#c.age#c.age age cubed
    
81.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
82. 
    
83. 
    
84. Base levels
    
85. 
    
86.     You can specify the base level of a factor variable by using the ib. operator.  The syntax is
    
87. 
    
88.            Base         
    
89.            operator(*)    Description
    
90.            ------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
91.            ib#.           use # as base, #=value of variable
    
92.            ib(##).        use the #th ordered value as base (**)
    
93.            ib(first).     use smallest value as base (the default)
    
94.            ib(last).      use largest value as base
    
95.            ib(freq).      use most frequent value as base
    
96.            ibn.           no base level
    
97.            ------------------------------------------------------------------------------------------------------------------------------------------------------------------
    
98.             (*) The i may be omitted.  For instance, you may type ib2.group or b2.group.
    
99.            (**) For example, ib(#2). means to use the second value as the base.
    
100. 
     
101.     If you want to use group==3 as the base in a regression, you can type,
     
102. 
     
103.         . regress y  i.sex ib3.group
     
104. 
     
105.     You can also permanently set the base levels of categorical variables by using the fvset command.
     
106. 
     
107. 
     
108. Selecting levels
     
109. 
     
110.     You can select a range of levels -- a range of virtual variables -- by using the i(numlist). operator.
     
111. 
     
112.          Examples          Description
     
113.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
     
114.          i2.cat            a single indicator for cat==2
     
115. 
     
116.          2.cat             same as i2.cat
     
117. 
     
118.          i(2 3 4).cat      three indicators, cat==2, cat==3, and cat==4;
     
119.                              same as i2.cat i3.cat i4.cat
     
120. 
     
121.          i(2/4).cat        same as i(2 3 4).cat
     
122. 
     
123.          2.cat#1.sex       a single indicator that is 1 when cat==2 and sex==1, and is 0 otherwise
     
124. 
     
125.          i2.cat#i1.sex     same as 2.cat#1.sex
     
126.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
     
127. 
     
128. 
     
129. Applying operators to a group of variables
     
130. 
     
131.     Factor-variable operators may be applied to groups of variables by using parentheses.
     
132. 
     
133.     In the examples that follow, variables group, sex, arm, and cat are categorical, and variables age, wt, and bp are continuous:
     
134. 
     
135.          Examples                  Expansion
     
136.          --------------------------------------------------------------------------------------------------------------------------------------------------------------------
     
137.          i.(group sex arm)         i.group i.sex i.arm
     
138. 
     
139.          group#(sex arm cat)       group#sex group#arm group#cat
     
140. 
     
141.          group##(sex arm cat)      i.group i.sex i.arm i.cat group#sex group#arm group#cat
     
142. 
     
143.          group#(c.age c.wt c.bp)   i.group group#c.age group#c.wt group#c.bp
     
144. 
     
145.          group#c.(age wt bp)       same as group#(c.age c.wt c.bp)

复制代码

具体用法你可以详见

终于明白了！太感谢了！！那再请问一般那些发表的文章中，哪一种指令更为常用？

jingleqq 发表于 2013-10-16 14:19
终于明白了！太感谢了！！那再请问一般那些发表的文章中，哪一种指令更为常用？

语句没有关系，你能实现自己目的就行。
比如在生成虚拟变量时，你有两种方法：
1、直接在命令前面加xi
xi:reg y x i.year
2、可以先利用tab命令生成虚拟变量，然后加入
tab year,gen(yeardum)
reg y x yeardum*

恩，我一直使用的是xi的方法。

我还有点困惑：在发表的文章当中，我看见凡是有交互项的回归分析表，其交互项的相关数据coef.、标准差或t值）都只有一行，难道他们都是采用诸如“a2022rural”这一种指令吗？假如用xi：i.rural*a2022或 rural#a2022这两种指令，相关数据会有三行（如下所示），那么文章中的回归分析表是否该全部列出这几行数据，还是只需要列出最后一行的数据即可？

1.        a2022#rural |
2.                0 1  |  -.4517703    .303656    -1.49   0.137    -1.047872    .1443315
3.                1 0  |   .0788755   .1641699     0.48   0.631    -.2434036    .4011545
4.                1 1  |  -.1898026   .2035583    -0.93   0.351    -.5894043    .2097991

1.        _Irural_1 | -.4517703 .303656 -1.49 0.137 -1.047872 .1443315
2.        a2022 | .0788755 .1641699 0.48 0.631 -.2434036 .4011545
3.        _IrurXa2022_1 | .1830922 .3539682 0.52 0.605 -.5117764 .8779608

jingleqq 发表于 2013-10-16 14:43
恩，我一直使用的是xi的方法。

我还有点困惑：在发表的文章当中，我看见凡是有交互项的回归分析表，其交 ...

楼主，这要看你自己的研究设计。
实际中你可以见到的更多是两个连续变量的交叉项。
如果文章中就是设计包含3项的，那结果都要给出的，不能只给出一项。

好的，我明白了！您真是非常专业和耐心！