|
|
| Message/Author |
|
|
| jan mod posted on Monday, September 09, 2019 - 1:53 pm
|
|
|
1) I have a continuous X (mean=0, standard deviation=1), a binary M and a continuous Y (which is standardized). I give two values of the continuous X (-1
SD and +1 SD).
Is the following a correct interpretation of the Total natural indirect effect?
“there is a “z” standard deviation decrease of Y when X goes from -1 to 1 standard deviation”
2) When X, M and Y are continuous, I can interpret the total indirect effect as what happens with Y when X increases with one unit.
3) If so, does the interpretation of 2) stay the same with a standardized M? |
|
|
|
|
1) Yes, the interpretation is correct - if you give the term (x1-x0) with x1=1SD and x0=-1SD. Also note that typically one uses x1-x0 = 1SD, not 2SDs. Typically, x0 is the mean and x1 1SD above the mean.
2) Yes, unless you specify x1-x0 as something different than 1.
3)only if TNIE is the same for standardized and unstandardized M (I assume you mean standardization before analysis) - check it. |
|
|
| jan mod posted on Monday, September 09, 2019 - 4:13 pm
|
|
|
Thank you.
Sorry, 2) and 3)(standardization before analysis) were about indirect effects (ind statement), as in chapter 2 of your book. I wondered whether 2 and 3 were correct in that context (indirect effects without binary variables). |
|
|
|
|
| I think 2) is ok. Not sure if 3) is. You can check by comparing to a run where the variable is not pre-standardized, looking at the standardized TNIE which simply divides by the Y SD. |
|
|
| Back to top |
|
|
