2024年2月15日发(作者：鲁苒苒)
以著名的精神病测量量表《简明症状问卷18》（Brief Symptoms Inventory 18,
BSI-18）为例。BSI-18量表的各条目的测量采用Likert量表的5级评分法：
0——无；1——轻度；2——中度；3——偏重；4——严重。
尽管Likert量表的测量实际上是次序测量（ordinal measure），但在CFA以及其他统计模型分析中，有时会将5个以及更多点的次序测量作为连续变量分析。
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
ANALYSIS: ESTIMATOR = ML; TYPE=GENERAL;
MODEL=NOMEANSTRUCTURE; INFORMATION=EXPECTED;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
程序说明：
LISTWISE=ON; 设定了缺失值的删除方法：如果一个个案涉及的任何变量有缺失值，则该个案将整个从分析过程中提出。
MISSING= ALL (-9); 设定数据中的缺失值。
TECH4：输出潜变量的方差、协方差和相关系数。
程序输出：
DEP BY
X5 0.831 0.025 32.905 0.000
X2 0.760 0.031 24.343 0.000
X8 0.881 0.021 41.994 0.000
X11 0.751 0.033 23.008 0.000
X14 0.573 0.047 12.136 0.000
X17 0.384 0.058 6.625 0.000
RMSEA (Root Mean Square Error Of Approximation)
Estimate 0.074
90 Percent C.I. 0.063 0.085
Probability RMSEA <= .05 0.000
条目的可靠度（item reliability）：反映与具体题目相应的因子能够解释该条目变异的百分比。
R-SQUARE
Observed Two-Tailed
Variable Estimate S.E. Est./S.E. P-Value
X1 0.496 0.053 9.343 0.000
X2 0.578 0.048 12.172 0.000
X3 0.486 0.053 9.210 0.000
X4 0.369 0.056 6.624 0.000
X5 0.691 0.042 16.452 0.000
X6 0.518 0.052 9.965 0.000
X7 0.554 0.051 10.838 0.000
X8 0.776 0.037 20.997 0.000
X9 0.469 0.054 8.679 0.000
X10 0.505 0.053 9.487 0.000
X11 0.564 0.049 11.504 0.000
X12 0.503 0.053 9.483 0.000
X13 0.333 0.055 6.020 0.000
X14 0.329 0.054 6.068 0.000
X15 0.457 0.054 8.530 0.000
X16 0.575 0.050 11.402 0.000
X17 0.148 0.045 3.313 0.001
X18 0.422 0.054 7.790 0.000
量表可靠性（scale reliability）：当测量误差不相关时，对于标准化结果，有
(i)2i(i)i2ii
MODEL MODIFICATION INDICES
NOTE: Modification indices for direct effects of observed dependent variables
regressed on covariates may not be included. To include these, request
MODINDICES (ALL).
Minimum M.I. value for printing the modification index 10.000
M.I. E.P.C. Std E.P.C. StdYX E.P.C.
BY Statements
SOM BY X8 10.717 -0.398
WITH Statements
X3 WITH X1 14.089 0.167
X8 WITH X5 64.074 0.392
X10 WITH X4 25.443 0.249
X12 WITH X9 33.178 0.273
X14 WITH X5 16.098 -0.225
X14 WITH X11 14.209 0.220
X16 WITH X13 11.891 0.221
X17 WITH X9 11.945 0.103
X17 WITH X14 11.101 0.123
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
X5 with X8;
X9 with X12;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
-0.272 -0.220
0.167 0.273 0.392 0.927 0.249 0.374 0.273 0.435
-0.225 -0.307
0.220 0.269 0.221 0.262 0.103 0.236
0.123 0.221
RMSEA (Root Mean Square Error Of Approximation)
Estimate 0.054
90 Percent C.I. 0.041 0.066
Probability RMSEA <= .05 0.297
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
ANALYSIS: TYPE=MISSING;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
TITLE: Testing nonnormality;
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES ARE X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES ARE X1-X18;
CLASSES=C(1);
ANALYSIS: TYPE = MIXTURE;
MODEL: %Overall%
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH13;
TECH13：数据的非正态性检验，需要同CLASSES=C(1)和TYPE = MIXTURE一起使用。
TECHNICAL 13 OUTPUT
SKEW AND KURTOSIS TESTS OF MODEL FIT
TWO-SIDED MULTIVARIATE SKEW TEST OF FIT
Sample Value
Mean
Standard Deviation
P-Value
TWO-SIDED MULTIVARIATE KURTOSIS TEST OF FIT
Sample Value
Mean
Standard Deviation
P-Value
TITLE: CFA with indicators treated as ordinal.
DATA: FILE = BSI_;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
CATEGORICAL = X1-X18;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
85.942
27.779
1.219
0.0000
481.889
357.265
3.174
0.0000
2024年2月15日发(作者：鲁苒苒)
以著名的精神病测量量表《简明症状问卷18》（Brief Symptoms Inventory 18,
BSI-18）为例。BSI-18量表的各条目的测量采用Likert量表的5级评分法：
0——无；1——轻度；2——中度；3——偏重；4——严重。
尽管Likert量表的测量实际上是次序测量（ordinal measure），但在CFA以及其他统计模型分析中，有时会将5个以及更多点的次序测量作为连续变量分析。
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
ANALYSIS: ESTIMATOR = ML; TYPE=GENERAL;
MODEL=NOMEANSTRUCTURE; INFORMATION=EXPECTED;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
程序说明：
LISTWISE=ON; 设定了缺失值的删除方法：如果一个个案涉及的任何变量有缺失值，则该个案将整个从分析过程中提出。
MISSING= ALL (-9); 设定数据中的缺失值。
TECH4：输出潜变量的方差、协方差和相关系数。
程序输出：
DEP BY
X5 0.831 0.025 32.905 0.000
X2 0.760 0.031 24.343 0.000
X8 0.881 0.021 41.994 0.000
X11 0.751 0.033 23.008 0.000
X14 0.573 0.047 12.136 0.000
X17 0.384 0.058 6.625 0.000
RMSEA (Root Mean Square Error Of Approximation)
Estimate 0.074
90 Percent C.I. 0.063 0.085
Probability RMSEA <= .05 0.000
条目的可靠度（item reliability）：反映与具体题目相应的因子能够解释该条目变异的百分比。
R-SQUARE
Observed Two-Tailed
Variable Estimate S.E. Est./S.E. P-Value
X1 0.496 0.053 9.343 0.000
X2 0.578 0.048 12.172 0.000
X3 0.486 0.053 9.210 0.000
X4 0.369 0.056 6.624 0.000
X5 0.691 0.042 16.452 0.000
X6 0.518 0.052 9.965 0.000
X7 0.554 0.051 10.838 0.000
X8 0.776 0.037 20.997 0.000
X9 0.469 0.054 8.679 0.000
X10 0.505 0.053 9.487 0.000
X11 0.564 0.049 11.504 0.000
X12 0.503 0.053 9.483 0.000
X13 0.333 0.055 6.020 0.000
X14 0.329 0.054 6.068 0.000
X15 0.457 0.054 8.530 0.000
X16 0.575 0.050 11.402 0.000
X17 0.148 0.045 3.313 0.001
X18 0.422 0.054 7.790 0.000
量表可靠性（scale reliability）：当测量误差不相关时，对于标准化结果，有
(i)2i(i)i2ii
MODEL MODIFICATION INDICES
NOTE: Modification indices for direct effects of observed dependent variables
regressed on covariates may not be included. To include these, request
MODINDICES (ALL).
Minimum M.I. value for printing the modification index 10.000
M.I. E.P.C. Std E.P.C. StdYX E.P.C.
BY Statements
SOM BY X8 10.717 -0.398
WITH Statements
X3 WITH X1 14.089 0.167
X8 WITH X5 64.074 0.392
X10 WITH X4 25.443 0.249
X12 WITH X9 33.178 0.273
X14 WITH X5 16.098 -0.225
X14 WITH X11 14.209 0.220
X16 WITH X13 11.891 0.221
X17 WITH X9 11.945 0.103
X17 WITH X14 11.101 0.123
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
X5 with X8;
X9 with X12;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
-0.272 -0.220
0.167 0.273 0.392 0.927 0.249 0.374 0.273 0.435
-0.225 -0.307
0.220 0.269 0.221 0.262 0.103 0.236
0.123 0.221
RMSEA (Root Mean Square Error Of Approximation)
Estimate 0.054
90 Percent C.I. 0.041 0.066
Probability RMSEA <= .05 0.297
TITLE: CFA with indicators treated as continuous.
DATA: FILE = BSI_;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
ANALYSIS: TYPE=MISSING;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
TITLE: Testing nonnormality;
DATA: FILE = BSI_;
LISTWISE=ON;
VARIABLE:
NAMES ARE X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES ARE X1-X18;
CLASSES=C(1);
ANALYSIS: TYPE = MIXTURE;
MODEL: %Overall%
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH13;
TECH13：数据的非正态性检验，需要同CLASSES=C(1)和TYPE = MIXTURE一起使用。
TECHNICAL 13 OUTPUT
SKEW AND KURTOSIS TESTS OF MODEL FIT
TWO-SIDED MULTIVARIATE SKEW TEST OF FIT
Sample Value
Mean
Standard Deviation
P-Value
TWO-SIDED MULTIVARIATE KURTOSIS TEST OF FIT
Sample Value
Mean
Standard Deviation
P-Value
TITLE: CFA with indicators treated as ordinal.
DATA: FILE = BSI_;
VARIABLE:
NAMES = X1-X18 gender white age edu crack id;
MISSING= ALL (-9);
USEVARIABLES = X1-X18;
CATEGORICAL = X1-X18;
MODEL:
SOM BY X1 X4 X7 X10 X13 X16; !Somatization;
DEP BY X5 X2 X8 X11 X14 X17; !Depression;
ANX BY X3 X6 X9 X12 X15 X18; !Anxiety;
OUTPUT: SAMPSTAT TECH1 TECH4 STDYX MOD;
85.942
27.779
1.219
0.0000
481.889
357.265
3.174
0.0000