Bootstrapp: unexpected difference between path-totaleffects

[erasmusvonot]

Hello everybody,

I did a bootstrapp and analyzed the samples. I got differences in total effects and path coefficients where I did not expect them.

I will make you an example (my model is a little bit more complex)

There are the five Variables: A, B, C, D and E
The pathconnections are:

A->B
A->E
B->C
B->D
B->E
C->E
D->E
There are no connections from A->C or A->D

It is clear that the path coeficients from A->D or B->D are different to the total effects but I did not expect a difference in the relation C->E and D->E

There are no problems after the algorithm but when I look at the bootstrapp results most of the coefficents are the same but sometimes their absolute value is the same but with different algebraic sign. I did not see a rule behind it. It seems to be random.

Whats the reason for this?