老师您好,我有个迫切需要解决的研究方法问题,该题的题目没有看明白,不知道该跑什么模型,希望能得到您的帮助,附上数据。谢谢!
Test if the data in file Q2.dat could be viewed as complying with a single-factor model for the 6 observed variables in it.Test then if there is evidence to suggest that the 6 intelligence tests in that data are (not) having the same units of measurement.
Hint: The single-factor model may not fit very well, on the surface of things, but to evaluate its fit use the lower endpoint of the 90%-confidence interval of the RMSEA; with it, you may be able to argue in favor of being possible to consider this a starting(full) model for the test of “same units of measurement.”
Response:
该题的意思是让你做两个test。第一个test是检验数据中的6个观测变量是否服从单因子模型;第二个test是检验这6个观测变量(智力测验)是否具有相同的测量单位。Hint里面是建议你使用RMSEA 90%CI的下限去判断模型的拟合状况,如果拟合尚可,可以但因种子模型为基准模型去进一步检验单位等值。
Test 1: Single-factor model for t1-t6?
INPUT:
data: file is q2.dat;
vari: names are t1-t6;
model:
f by t1-t6*;
f@1;
outp: stdyx;
OUTPUT:


从输出结果可以看出,RMSEA 90%CI的下限为0.055,故拟合良好(注:实际研究中很少有这种极端判别方法,因为下限判别倾向于接受研究模型,应结合RMSEA的值及其置信区间以及其他指数综合判断)。因此,单因子模型可以作为基准模型去检验单位等值问题。
INPUT:
data: file is q2.dat;
vari: names are t1-t6;
model:
f by t1-t6*(1);
f@1;
outp: stdyx;
OUTPUT:


从输出结果可以看出,该模型已限定各观测变量的测量单位等值(2.244)。检验这种等值限定是否合理,需要比较该模型与基准模型的模型拟合,如果等值限定没有显著的恶化模型拟合,则表明单位等值的假定可以被接受。这里,等值模型与基准模型是嵌套模型,我们利用卡方检验去处理这个问题。卡方间加减后构成的统计量仍然服从卡方分布,自由度为两个模型的自由度之差。该检验的卡方值为435.027(461.754-26.727),自由度为5(14-9),p<0.001。因此,增加等值限定显著地恶化了模型拟合,单位等值假定应予以拒绝。
Good luck with your research!

