理解LMS对应的mplus语句

标签:
研究 |
分类: 知无涯 |

LMS是latent moderated structural
equations的缩写,长相就是上面那样。出自论文Preacher, K. J., Zhang, Z., & Zyphur,
M. J. (2016). Multilevel structural equation models for assessing
moderation within and across levels of analysis. Psychological
methods, 21, 189-205. 图在第11页。这个模型我不确定会不会用,跟一个朋友交流说周边从没有人用过,只能安慰自己说是重新熟悉一下mplus语句,语句的出处是这篇论文在线的supplementary
materials,preacher网站上可以找到。写下中文标注(在语句下方,用黑色)加深点印象+存档。
MODEL:
%WITHIN%
sesw BY ses@1;
sesw*.436; ses@.01;
用single indicator来搞latent
variable它的标准语句就是 x@1; x@a,所以在这个模型里需要把indicator “ses”的 loading fix
为1,然后根据ses的reliability来估算它的误差项(error term
)并且也进行标定,不标定model不identify。error term的公式是Var(ses)*(1 -
rho),reliability越高,相应的error
term也越小[注1]。作者应该是进行了猜测,所以选ses@.01接近于0。
sesw*.436就是赋予它variance的初始值,根据Mplus
Statistical Analysis With Latent Variables User Guide 6
(以下简称”用户手册“)p.594“y1*.5; is interpreted as freeing the
variance/residual variance of y1 to be estimated with a starting
value of 0.5.” 不过这个初始值好像完全是根据模型最后得出的结果(Table2 给出了各参数的估算值)来给的。
mathachw BY
mathach@1; mathachw*36.6; mathach@.01;
b1 | mathachw
ON sesw;
根据https://www.statmodel.com/language.html (以下简称“一览表”),这里"|"
就是用来标定随机效应变量names and defines random effect variables example: s |
y1 ON x1; "b1"对应着图里的s1j
sessizew |
sesw XWITH size;
根据用户手册p.617这是在搞乘积项,XWITH and the |
symbol to name and define an interaction
mathachw ON
sessizew*.58;
TWEEN%
sesb BY ses@1;
sesb*.15; ses@.01;
mathach*2.06;
sessizeb | sesb XWITH size;
mathach ON
size*-.11 sesb*7.1 sessizeb*-.52;
[ses@0
sesb*-.01 mathach*12.81 b1*1.6];
这个[ ]符号有点陌生,根据一览表指的是mean的估值(“[list
of variables]; refers to means, intercepts,
thresholds example: [f1,
y1-y9]:)这些对应图上三角(代表常量项1)出来的箭头。另,个人理解:b1在between层面的mean应该就是在within层面的slope,在table
2里给的结果也是sesw-->mathw 1.607
b1*.6; mathach
WITH b1*-.23;
sesb WITH
size*-.03; b1 WITH sesb*.07;
OUTPUT: TECH1
TECH3;
https://www.researchgate.net/post/Did_you_use_single-item_measures_in_your_SEM_models_How_did_it_work_out
后一篇:重温《红色》