Error of changing covariates when creating siena data by using R package RSiena

My data are collected from 200 people at 2 time points, which include

A1: a 200×200 network matrix for time 1 (dependent variable)
A2: a 200×200 network matrix for time 2 (dependent variable)
C1: a 200×200 network matrix for time 1
C2: a 200×200 network matrix for time 2
S: a 200×1 vector
Y: a 200×1 vector
P: a 200×1 vector
N: a 200×1 vector

The codes I use are as follows.

> A <- sienaDependent(array(c(A1, A2), dim=c(200, 200, 2)))
> C <- varDyadCovar(array(c(C1, C2), dim=c(200, 200, 2)))
> S <- coCovar(S)
> Y <- coCovar(Y)
> P <- sienaDependent(P, type = "behavior" )
> N <- sienaDependent(N, type = "behavior" )

The error shows when I try to create the siena data object with error message like

> MyData <- sienaDataCreate(A, S, Y, C, P, N)

    Error in sienaDataCreate(A, S, Y, C, ：
    Changing covariates are not possible with only two waves

Does the error shows because I only have 2 matrices at time 1 and time 2? Do I need more observations from more than 3 time points to create a varying dyadic variable? Thank you.

Solution

As you guessed, you need more than 3 observations of the same network to achieve meaningful results for influence effects. The RSiena manual states for individual covariates : "Changing covariates, as independent variables, are meaningful only if there are 3 or more observa- tion moments, because for 2 observation moments the distinction between constant and changing covariates is not meaningful." (RSiena Manual p. 27) In general I can recommend using the RSiena Manual especially when you are new to RSiena: http://www.stats.ox.ac.uk/~snijders/siena/RSiena_Manual.pdf