BasicRSiena

Tom Snijders

August 9, 2020

An example of a basic sequence of commands for estimating a basic model by function siena07() of RSiena. With a lot of use of help pages; this can be skipped as you like. Note that lines starting with # are comment lines, not commands.

What is your current working directory?

getwd()
#> [1] "C:/Users/Tom.Snijders/AppData/Local/Temp/RtmpUXNCro/Rbuild429870dc3d8d/RSiena/vignettes"

If you wish it to be different, change it by

# setwd()

In my case:

# setwd("C:\\Users\\tom.snijders\\Documents\\Siena\\s50_script")

Note the double backslashes used for R.

Define data sets

If you have internet access, you can download the data from the Siena website (“Data sets” tab) http://www.stats.ox.ac.uk/~snijders/siena/s50_data.zip and unzip it in your working directory. The data description is at http://www.stats.ox.ac.uk/~snijders/siena/s50_data.htm

Then you can read the data files by the commands (this can be replaced by using the internal data set, see below)

# friend.data.w1 <- as.matrix(read.table("s50-network1.dat"))
# friend.data.w2 <- as.matrix(read.table("s50-network2.dat"))
# friend.data.w3 <- as.matrix(read.table("s50-network3.dat"))
# drink <- as.matrix(read.table("s50-alcohol.dat"))
# smoke <- as.matrix(read.table("s50-smoke.dat"))

But without internet access, the data can be obtained from within RSiena (see below), because this is an internal data set.

library(RSiena)
# Now we use the internally available s50 data set.
# Look at its description:
?s50
# 3 waves, 50 actors
# Look at the start and end of the first wave matrix
head(s501)
#>   V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12 V13 V14 V15 V16 V17 V18 V19 V20 V21
#> 1  0  0  0  0  0  0  0  0  0   0   1   0   0   1   0   0   0   0   0   0   0
#> 2  0  0  0  0  0  0  1  0  0   0   1   0   0   0   0   0   0   0   0   0   0
#> 3  0  0  0  1  0  0  0  0  1   0   0   0   0   0   0   0   0   0   0   0   0
#> 4  0  0  1  0  0  0  0  0  1   0   0   0   0   0   0   0   0   0   0   0   0
#> 5  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 6  0  0  0  0  0  0  0  1  0   0   0   0   0   0   0   0   0   0   0   0   0
#>   V22 V23 V24 V25 V26 V27 V28 V29 V30 V31 V32 V33 V34 V35 V36 V37 V38 V39 V40
#> 1   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 2   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 3   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 4   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 5   0   0   0   0   0   0   0   0   0   0   1   0   0   0   0   0   0   0   0
#> 6   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#>   V41 V42 V43 V44 V45 V46 V47 V48 V49 V50
#> 1   0   0   0   0   0   0   0   0   0   0
#> 2   0   0   0   0   0   0   0   0   0   0
#> 3   0   0   0   0   0   0   0   0   0   0
#> 4   0   0   0   0   0   0   0   0   0   0
#> 5   0   0   0   0   0   0   0   0   0   0
#> 6   0   0   0   0   0   0   0   0   0   0
tail(s501)
#>    V1 V2 V3 V4 V5 V6 V7 V8 V9 V10 V11 V12 V13 V14 V15 V16 V17 V18 V19 V20 V21
#> 45  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 46  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 47  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 48  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 49  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#> 50  0  0  0  0  0  0  0  0  0   0   0   0   0   0   0   0   0   0   0   0   0
#>    V22 V23 V24 V25 V26 V27 V28 V29 V30 V31 V32 V33 V34 V35 V36 V37 V38 V39 V40
#> 45   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1
#> 46   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   1
#> 47   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 48   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 49   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#> 50   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0   0
#>    V41 V42 V43 V44 V45 V46 V47 V48 V49 V50
#> 45   0   0   0   0   0   1   1   0   0   0
#> 46   0   0   0   0   1   0   0   0   1   0
#> 47   0   0   0   0   0   0   0   0   0   0
#> 48   0   0   0   0   0   1   0   0   1   0
#> 49   0   0   0   0   0   1   0   1   0   0
#> 50   0   0   0   0   0   0   0   0   0   0
# and at the alcohol variable
s50a
#>    V1 V2 V3
#> 1   3  1  3
#> 2   2  2  2
#> 3   2  3  3
#> 4   2  3  2
#> 5   3  3  4
#> 6   4  4  4
#> 7   4  4  3
#> 8   4  5  4
#> 9   2  2  2
#> 10  4  5  4
#> 11  5  5  5
#> 12  5  5  5
#> 13  3  2  2
#> 14  3  4  3
#> 15  4  4  5
#> 16  4  5  4
#> 17  2  4  4
#> 18  4  3  3
#> 19  3  5  5
#> 20  2  3  3
#> 21  1  1  3
#> 22  3  2  3
#> 23  4  4  2
#> 24  3  3  3
#> 25  3  4  4
#> 26  4  4  3
#> 27  2  2  3
#> 28  2  2  4
#> 29  3  3  3
#> 30  1  3  4
#> 31  4  4  4
#> 32  4  2  4
#> 33  3  4  3
#> 34  2  2  2
#> 35  3  3  4
#> 36  4  4  4
#> 37  2  2  3
#> 38  3  3  4
#> 39  2  2  3
#> 40  1  1  1
#> 41  4  3  4
#> 42  4  5  5
#> 43  2  2  4
#> 44  5  5  5
#> 45  2  2  2
#> 46  2  2  2
#> 47  2  2  2
#> 48  2  3  4
#> 49  1  2  3
#> 50  1  2  3
# Now define the objects with the same names as above
# (this step is superfluous if you read the data already).
        friend.data.w1 <- s501
        friend.data.w2 <- s502
        friend.data.w3 <- s503
        drink <- s50a
        smoke <- s50s

Now the data must be given the specific roles of variables in an RSiena analysis.

Dependent variable

?sienaDependent
# First create a 50 * 50 * 3 array composed of the 3 adjacency matrices
friendshipData <- array( c( friend.data.w1, friend.data.w2, friend.data.w3 ),
           dim = c( 50, 50, 3 ) )
# and next give this the role of the dependent variable:
    friendship <- sienaDependent(friendshipData)

What did we construct?

friendship

We also must prepare the objects that will be the explanatory variables.

Actor covariates

We use smoking for wave 1 as a constant actor covariate:

smoke1 <- coCovar( smoke[ , 1 ] )
# A variable actor covariate is defined for drinking:
alcohol <- varCovar( drink )
# (This choice is purely for the purpose of illustration here.)

Put the variables together in the data set for analysis

?sienaDataCreate
mydata <- sienaDataCreate( friendship, smoke1, alcohol )
# Check what we have
mydata
#> Dependent variables:  friendship 
#> Number of observations: 3 
#> 
#> Nodeset                  Actors 
#> Number of nodes              50 
#> 
#> Dependent variable friendship      
#> Type               oneMode         
#> Observations       3               
#> Nodeset            Actors          
#> Densities          0.046 0.047 0.05
#> 
#> Constant covariates:  smoke1 
#> Changing covariates:  alcohol

You can get an outline of the data set with some basic descriptives from

print01Report( mydata, modelname="s50")
# For the model specification we need to create the effects object
myeff <- getEffects( mydata )
# All the effects that are available given the structure
# of this data set can be seen from
effectsDocumentation(myeff)
# For a precise description of all effects, see Chapter 12 in the RSiena manual.
# A basic specification of the structural effects:
?includeEffects
myeff <- includeEffects( myeff, transTrip, cycle3)
#>   effectName          include fix   test  initialValue parm
#> 1 transitive triplets TRUE    FALSE FALSE          0   0   
#> 2 3-cycles            TRUE    FALSE FALSE          0   0
# and some covariate effects:
myeff <- includeEffects( myeff, egoX, altX, simX, interaction1 = "alcohol" )
#>   effectName         include fix   test  initialValue parm
#> 1 alcohol alter      TRUE    FALSE FALSE          0   0   
#> 2 alcohol ego        TRUE    FALSE FALSE          0   0   
#> 3 alcohol similarity TRUE    FALSE FALSE          0   0
myeff <- includeEffects( myeff, simX, interaction1 = "smoke1" )
#>   effectName        include fix   test  initialValue parm
#> 1 smoke1 similarity TRUE    FALSE FALSE          0   0
myeff
#>    effectName                          include fix   test  initialValue parm
#> 1  constant friendship rate (period 1) TRUE    FALSE FALSE    4.69604   0   
#> 2  constant friendship rate (period 2) TRUE    FALSE FALSE    4.32885   0   
#> 3  outdegree (density)                 TRUE    FALSE FALSE   -1.46770   0   
#> 4  reciprocity                         TRUE    FALSE FALSE    0.00000   0   
#> 5  transitive triplets                 TRUE    FALSE FALSE    0.00000   0   
#> 6  3-cycles                            TRUE    FALSE FALSE    0.00000   0   
#> 7  smoke1 similarity                   TRUE    FALSE FALSE    0.00000   0   
#> 8  alcohol alter                       TRUE    FALSE FALSE    0.00000   0   
#> 9  alcohol ego                         TRUE    FALSE FALSE    0.00000   0   
#> 10 alcohol similarity                  TRUE    FALSE FALSE    0.00000   0

Create object with algorithm settings Accept defaults but specify name for output file (which you may replace by any name you prefer)

?sienaAlgorithmCreate
myalgorithm <- sienaAlgorithmCreate( projname = 's50' )
#> If you use this algorithm object, siena07 will create an output file s50.txt .

Estimate parameters

?siena07
ans <- siena07( myalgorithm, data = mydata, effects = myeff)
ans
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                    Estimate   Standard   Convergence 
#>                                                 Error      t-ratio   
#> 
#> Rate parameters: 
#>   0.1      Rate parameter period 1  6.6488  ( 1.1597   )             
#>   0.2      Rate parameter period 2  5.2851  ( 0.8996   )             
#> 
#> Other parameters: 
#>   1.  eval outdegree (density)     -2.7440  ( 0.1256   )   -0.0886   
#>   2.  eval reciprocity              2.4506  ( 0.2155   )   -0.0398   
#>   3.  eval transitive triplets      0.6692  ( 0.1474   )   -0.0578   
#>   4.  eval 3-cycles                -0.0989  ( 0.2951   )   -0.0545   
#>   5.  eval smoke1 similarity        0.2023  ( 0.2061   )   -0.0258   
#>   6.  eval alcohol alter           -0.0130  ( 0.0674   )    0.0215   
#>   7.  eval alcohol ego              0.0505  ( 0.0714   )   -0.0321   
#>   8.  eval alcohol similarity       0.7377  ( 0.3033   )   -0.0772   
#> 
#> Overall maximum convergence ratio:    0.1803 
#> 
#> 
#> Total of 2355 iteration steps.

This gives results from a random starting point. To use a fixed starting point, use the “seed” parameter:

# myalgorithm <- sienaAlgorithmCreate( projname = 's50', seed=435123 )

For checking convergence, look at the ‘Overall maximum convergence ratio’ mentioned under the parameter estimates.

It can also be shown by requesting 0.1803416

If this is less than 0.25, convergence is good. If convergence is inadequate, estimate once more, using the result obtained as the “previous answer” from which estimation continues:

ans <- siena07( myalgorithm, data = mydata, effects = myeff, prevAns=ans)
ans
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                    Estimate   Standard   Convergence 
#>                                                 Error      t-ratio   
#> 
#> Rate parameters: 
#>   0.1      Rate parameter period 1  6.6232  ( 1.1692   )             
#>   0.2      Rate parameter period 2  5.3234  ( 0.8868   )             
#> 
#> Other parameters: 
#>   1.  eval outdegree (density)     -2.7325  ( 0.1249   )   -0.0159   
#>   2.  eval reciprocity              2.4283  ( 0.2174   )    0.0045   
#>   3.  eval transitive triplets      0.6415  ( 0.1459   )   -0.0087   
#>   4.  eval 3-cycles                -0.0376  ( 0.2937   )    0.0165   
#>   5.  eval smoke1 similarity        0.2010  ( 0.1984   )    0.0139   
#>   6.  eval alcohol alter           -0.0117  ( 0.0790   )    0.0190   
#>   7.  eval alcohol ego              0.0555  ( 0.0799   )   -0.0265   
#>   8.  eval alcohol similarity       0.7300  ( 0.2858   )    0.0214   
#> 
#> Overall maximum convergence ratio:    0.2360 
#> 
#> 
#> Total of 2454 iteration steps.
# If convergence is good, you can look at the estimates.
# More extensive results
summary(ans)
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                    Estimate   Standard   Convergence 
#>                                                 Error      t-ratio   
#> 
#> Rate parameters: 
#>   0.1      Rate parameter period 1  6.6232  ( 1.1692   )             
#>   0.2      Rate parameter period 2  5.3234  ( 0.8868   )             
#> 
#> Other parameters: 
#>   1.  eval outdegree (density)     -2.7325  ( 0.1249   )   -0.0159   
#>   2.  eval reciprocity              2.4283  ( 0.2174   )    0.0045   
#>   3.  eval transitive triplets      0.6415  ( 0.1459   )   -0.0087   
#>   4.  eval 3-cycles                -0.0376  ( 0.2937   )    0.0165   
#>   5.  eval smoke1 similarity        0.2010  ( 0.1984   )    0.0139   
#>   6.  eval alcohol alter           -0.0117  ( 0.0790   )    0.0190   
#>   7.  eval alcohol ego              0.0555  ( 0.0799   )   -0.0265   
#>   8.  eval alcohol similarity       0.7300  ( 0.2858   )    0.0214   
#> 
#> Overall maximum convergence ratio:    0.2360 
#> 
#> 
#> Total of 2454 iteration steps.
#> 
#> Covariance matrix of estimates (correlations below diagonal)
#> 
#>        0.016       -0.017       -0.005        0.002       -0.001        0.000       -0.001       -0.007
#>       -0.615        0.047        0.007       -0.020        0.001        0.000        0.002       -0.003
#>       -0.277        0.215        0.021       -0.036       -0.001        0.000       -0.001       -0.001
#>        0.053       -0.306       -0.850        0.086       -0.002       -0.001        0.002        0.009
#>       -0.060        0.030       -0.037       -0.041        0.039        0.005        0.002       -0.010
#>        0.016       -0.015        0.016       -0.057        0.320        0.006       -0.003       -0.002
#>       -0.090        0.116       -0.122        0.073        0.152       -0.450        0.006        0.001
#>       -0.201       -0.042       -0.012        0.102       -0.170       -0.070        0.033        0.082
#> 
#> Derivative matrix of expected statistics X by parameters:
#> 
#>      282.325      223.142      469.493      150.503       26.051       18.528       30.955       19.725
#>      126.367      135.084      239.293       79.339       13.147        4.709        6.197        8.885
#>      327.361      296.361     1093.473      350.068       32.110       57.228       67.494       16.403
#>      161.703      153.738      529.010      177.714       16.165       25.211       26.342        7.165
#>       22.788       18.520       38.906       12.394       40.152      -52.788      -47.420        5.213
#>       19.771       24.132       93.886       32.200      -50.955      334.078      229.139        1.636
#>       25.256       20.419      106.420       32.227      -43.279      233.814      315.603       -0.626
#>       19.092       15.906       13.796        4.216        5.427       -3.235       -2.663       14.792
#> 
#> Covariance matrix of X (correlations below diagonal):
#> 
#>      445.621      387.594     1054.657      340.705       39.678       48.694       62.428       29.355
#>        0.927      392.404     1032.751      338.141       34.833       56.519       63.321       26.448
#>        0.789        0.823     4012.814     1304.428       87.018      219.494      230.301       51.250
#>        0.779        0.824        0.994      429.332       27.670       73.177       73.481       16.251
#>        0.272        0.255        0.199        0.193       47.662      -70.644      -66.307        6.610
#>        0.107        0.132        0.160        0.163       -0.473      468.227      378.248        1.521
#>        0.140        0.152        0.172        0.168       -0.456        0.829      444.356        1.746
#>        0.349        0.335        0.203        0.197        0.240        0.018        0.021       15.913

Still more extensive results are given in the output file s50.out in the current directory.

Note that by putting an R command between parentheses (….), the result will also be printed to the screen. Next add the transitive reciprocated triplets effect, an interaction between transitive triplets and reciprocity,

(myeff <- includeEffects( myeff, transRecTrip))
#>   effectName                  include fix   test  initialValue parm
#> 1 transitive recipr. triplets TRUE    FALSE FALSE          0   0
#>    effectName                          include fix   test  initialValue parm
#> 1  constant friendship rate (period 1) TRUE    FALSE FALSE    4.69604   0   
#> 2  constant friendship rate (period 2) TRUE    FALSE FALSE    4.32885   0   
#> 3  outdegree (density)                 TRUE    FALSE FALSE   -1.46770   0   
#> 4  reciprocity                         TRUE    FALSE FALSE    0.00000   0   
#> 5  transitive triplets                 TRUE    FALSE FALSE    0.00000   0   
#> 6  transitive recipr. triplets         TRUE    FALSE FALSE    0.00000   0   
#> 7  3-cycles                            TRUE    FALSE FALSE    0.00000   0   
#> 8  smoke1 similarity                   TRUE    FALSE FALSE    0.00000   0   
#> 9  alcohol alter                       TRUE    FALSE FALSE    0.00000   0   
#> 10 alcohol ego                         TRUE    FALSE FALSE    0.00000   0   
#> 11 alcohol similarity                  TRUE    FALSE FALSE    0.00000   0
(ans1 <- siena07( myalgorithm, data = mydata, effects = myeff, prevAns=ans))
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                        Estimate   Standard   Convergence 
#>                                                     Error      t-ratio   
#> 
#> Rate parameters: 
#>   0.1      Rate parameter period 1      6.2277  ( 1.0265   )             
#>   0.2      Rate parameter period 2      5.0783  ( 0.8455   )             
#> 
#> Other parameters: 
#>   1.  eval outdegree (density)         -2.9396  ( 0.1480   )    0.0078   
#>   2.  eval reciprocity                  2.8939  ( 0.2620   )    0.0092   
#>   3.  eval transitive triplets          0.9081  ( 0.1525   )   -0.0089   
#>   4.  eval transitive recipr. triplets -0.9095  ( 0.2708   )   -0.0219   
#>   5.  eval 3-cycles                     0.5079  ( 0.2953   )   -0.0214   
#>   6.  eval smoke1 similarity            0.1663  ( 0.2136   )    0.0006   
#>   7.  eval alcohol alter               -0.0234  ( 0.0785   )    0.0208   
#>   8.  eval alcohol ego                  0.0440  ( 0.0835   )    0.0226   
#>   9.  eval alcohol similarity           0.7201  ( 0.2985   )   -0.0481   
#> 
#> Overall maximum convergence ratio:    0.1236 
#> 
#> 
#> Total of 2677 iteration steps.
# If necessary, repeat the estimation with the new result:
(ans1 <- siena07( myalgorithm, data = mydata, effects = myeff, prevAns=ans1))
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                        Estimate   Standard   Convergence 
#>                                                     Error      t-ratio   
#> 
#> Rate parameters: 
#>   0.1      Rate parameter period 1      6.2272  ( 1.0380   )             
#>   0.2      Rate parameter period 2      5.1382  ( 0.8566   )             
#> 
#> Other parameters: 
#>   1.  eval outdegree (density)         -2.9398  ( 0.1493   )    0.0056   
#>   2.  eval reciprocity                  2.8891  ( 0.2599   )    0.0548   
#>   3.  eval transitive triplets          0.8849  ( 0.1479   )    0.0591   
#>   4.  eval transitive recipr. triplets -0.8820  ( 0.2745   )    0.0828   
#>   5.  eval 3-cycles                     0.5357  ( 0.3010   )    0.0659   
#>   6.  eval smoke1 similarity            0.1632  ( 0.2199   )   -0.0139   
#>   7.  eval alcohol alter               -0.0222  ( 0.0795   )    0.0636   
#>   8.  eval alcohol ego                  0.0425  ( 0.0754   )    0.0151   
#>   9.  eval alcohol similarity           0.7210  ( 0.3068   )   -0.0575   
#> 
#> Overall maximum convergence ratio:    0.2006 
#> 
#> 
#> Total of 2448 iteration steps.

This might still not have an overall maximum convergence ratio less than 0.25. If not, you could go on once more.

Inspect the file s50.txt in your working directory and understand the meaning of its contents.

To have a joint test of the three effects of alcohol:

?Multipar.RSiena
Multipar.RSiena(ans1, 7:9)

Focusing on alcohol similarity, the effect is significant; diluting the effects of alcohol by also considering ego and alter, the three effects simultaneously are not significant.

Assignment 1

1a.

Drop the effect of smoke1 similarity and estimate the model again. Do this by the function setEffects() using the <> parameter. Give the changed effects object and the new answer object new names, such as effects1 and ans1, to distinguish them.

1b.

Change the three effects of alcohol to the single effect of alcohol similarity, and estimate again.

Networks and behavior study

Now we redefine the role of alcohol drinking as a dependent behaviour variable.

# Once again, look at the help file
?sienaDependent
# now paying special attention to the <<type>> parameter.
drinking <- sienaDependent( drink, type = "behavior" )

Put the variables together in the data set for analysis

NBdata <- sienaDataCreate( friendship, smoke1, drinking )
NBdata
#> Dependent variables:  friendship, drinking 
#> Number of observations: 3 
#> 
#> Nodeset                  Actors 
#> Number of nodes              50 
#> 
#> Dependent variable friendship      
#> Type               oneMode         
#> Observations       3               
#> Nodeset            Actors          
#> Densities          0.046 0.047 0.05
#> 
#> Dependent variable drinking
#> Type               behavior
#> Observations       3       
#> Nodeset            Actors  
#> Range              1 - 5   
#> 
#> Constant covariates:  smoke1
NBeff <- getEffects( NBdata )
effectsDocumentation(NBeff)
NBeff <- includeEffects( NBeff, transTrip, transRecTrip )
#>   effectName                  include fix   test  initialValue parm
#> 1 transitive triplets         TRUE    FALSE FALSE          0   0   
#> 2 transitive recipr. triplets TRUE    FALSE FALSE          0   0
NBeff <- includeEffects( NBeff, egoX, egoSqX, altX, altSqX, diffSqX,
                         interaction1 = "drinking" )
#>   effectName             include fix   test  initialValue parm
#> 1 drinking alter         TRUE    FALSE FALSE          0   0   
#> 2 drinking squared alter TRUE    FALSE FALSE          0   0   
#> 3 drinking ego           TRUE    FALSE FALSE          0   0   
#> 4 drinking squared ego   TRUE    FALSE FALSE          0   0   
#> 5 drinking diff. squared TRUE    FALSE FALSE          0   0
NBeff <- includeEffects( NBeff, egoX, altX, simX, interaction1 = "smoke1" )
#>   effectName        include fix   test  initialValue parm
#> 1 smoke1 alter      TRUE    FALSE FALSE          0   0   
#> 2 smoke1 ego        TRUE    FALSE FALSE          0   0   
#> 3 smoke1 similarity TRUE    FALSE FALSE          0   0
NBeff
#>    name       effectName                          include fix   test 
#> 1  friendship constant friendship rate (period 1) TRUE    FALSE FALSE
#> 2  friendship constant friendship rate (period 2) TRUE    FALSE FALSE
#> 3  friendship outdegree (density)                 TRUE    FALSE FALSE
#> 4  friendship reciprocity                         TRUE    FALSE FALSE
#> 5  friendship transitive triplets                 TRUE    FALSE FALSE
#> 6  friendship transitive recipr. triplets         TRUE    FALSE FALSE
#> 7  friendship smoke1 alter                        TRUE    FALSE FALSE
#> 8  friendship smoke1 ego                          TRUE    FALSE FALSE
#> 9  friendship smoke1 similarity                   TRUE    FALSE FALSE
#> 10 friendship drinking alter                      TRUE    FALSE FALSE
#> 11 friendship drinking squared alter              TRUE    FALSE FALSE
#> 12 friendship drinking ego                        TRUE    FALSE FALSE
#> 13 friendship drinking squared ego                TRUE    FALSE FALSE
#> 14 friendship drinking diff. squared              TRUE    FALSE FALSE
#> 15 drinking   rate drinking (period 1)            TRUE    FALSE FALSE
#> 16 drinking   rate drinking (period 2)            TRUE    FALSE FALSE
#> 17 drinking   drinking linear shape               TRUE    FALSE FALSE
#> 18 drinking   drinking quadratic shape            TRUE    FALSE FALSE
#>    initialValue parm
#> 1     4.69604   0   
#> 2     4.32885   0   
#> 3    -1.46770   0   
#> 4     0.00000   0   
#> 5     0.00000   0   
#> 6     0.00000   0   
#> 7     0.00000   0   
#> 8     0.00000   0   
#> 9     0.00000   0   
#> 10    0.00000   0   
#> 11    0.00000   0   
#> 12    0.00000   0   
#> 13    0.00000   0   
#> 14    0.00000   0   
#> 15    0.70571   0   
#> 16    0.84939   0   
#> 17    0.32237   0   
#> 18    0.00000   0
# For including effects also for the dependent behaviour variable, see
?includeEffects
NBeff <- includeEffects( NBeff, avAlt, name="drinking",
                         interaction1 = "friendship" )
#>   effectName             include fix   test  initialValue parm
#> 1 drinking average alter TRUE    FALSE FALSE          0   0
NBeff
#>    name       effectName                          include fix   test 
#> 1  friendship constant friendship rate (period 1) TRUE    FALSE FALSE
#> 2  friendship constant friendship rate (period 2) TRUE    FALSE FALSE
#> 3  friendship outdegree (density)                 TRUE    FALSE FALSE
#> 4  friendship reciprocity                         TRUE    FALSE FALSE
#> 5  friendship transitive triplets                 TRUE    FALSE FALSE
#> 6  friendship transitive recipr. triplets         TRUE    FALSE FALSE
#> 7  friendship smoke1 alter                        TRUE    FALSE FALSE
#> 8  friendship smoke1 ego                          TRUE    FALSE FALSE
#> 9  friendship smoke1 similarity                   TRUE    FALSE FALSE
#> 10 friendship drinking alter                      TRUE    FALSE FALSE
#> 11 friendship drinking squared alter              TRUE    FALSE FALSE
#> 12 friendship drinking ego                        TRUE    FALSE FALSE
#> 13 friendship drinking squared ego                TRUE    FALSE FALSE
#> 14 friendship drinking diff. squared              TRUE    FALSE FALSE
#> 15 drinking   rate drinking (period 1)            TRUE    FALSE FALSE
#> 16 drinking   rate drinking (period 2)            TRUE    FALSE FALSE
#> 17 drinking   drinking linear shape               TRUE    FALSE FALSE
#> 18 drinking   drinking quadratic shape            TRUE    FALSE FALSE
#> 19 drinking   drinking average alter              TRUE    FALSE FALSE
#>    initialValue parm
#> 1     4.69604   0   
#> 2     4.32885   0   
#> 3    -1.46770   0   
#> 4     0.00000   0   
#> 5     0.00000   0   
#> 6     0.00000   0   
#> 7     0.00000   0   
#> 8     0.00000   0   
#> 9     0.00000   0   
#> 10    0.00000   0   
#> 11    0.00000   0   
#> 12    0.00000   0   
#> 13    0.00000   0   
#> 14    0.00000   0   
#> 15    0.70571   0   
#> 16    0.84939   0   
#> 17    0.32237   0   
#> 18    0.00000   0   
#> 19    0.00000   0
# Define an algorithm with a new project name
myalgorithm1 <- sienaAlgorithmCreate( projname = 's50_NB' )
#> If you use this algorithm object, siena07 will create an output file s50_NB.txt .

# Estimate again, using the second algorithm right from the start.
NBans <- siena07( myalgorithm1, data = NBdata, effects = NBeff)
# You may improve convergence (considering the overall maximum
# convergence ratio) by repeated estimation in the same way as above.

# Look at results
NBans
#> Estimates, standard errors and convergence t-ratios
#> 
#>                                                Estimate   Standard   Convergence 
#>                                                             Error      t-ratio   
#> Network Dynamics 
#>    1. rate constant friendship rate (period 1)  6.3315  ( 1.1202   )    0.0392   
#>    2. rate constant friendship rate (period 2)  5.0594  ( 0.8609   )   -0.0319   
#>    3. eval outdegree (density)                 -2.8298  ( 0.2408   )    0.0138   
#>    4. eval reciprocity                          2.8344  ( 0.3002   )    0.0500   
#>    5. eval transitive triplets                  0.8913  ( 0.1597   )    0.0633   
#>    6. eval transitive recipr. triplets         -0.5034  ( 0.2548   )    0.0797   
#>    7. eval smoke1 alter                         0.0783  ( 0.1585   )   -0.0466   
#>    8. eval smoke1 ego                          -0.0124  ( 0.1615   )   -0.0618   
#>    9. eval smoke1 similarity                    0.2517  ( 0.2629   )    0.0480   
#>   10. eval drinking alter                      -0.0678  ( 0.1207   )   -0.0544   
#>   11. eval drinking squared alter              -0.1173  ( 0.1290   )   -0.0013   
#>   12. eval drinking ego                         0.0453  ( 0.1214   )   -0.0431   
#>   13. eval drinking squared ego                 0.2308  ( 0.1087   )   -0.0010   
#>   14. eval drinking diff. squared              -0.1038  ( 0.0529   )    0.0111   
#> 
#> Behavior Dynamics
#>   15. rate rate drinking (period 1)             1.3376  ( 0.3510   )    0.0363   
#>   16. rate rate drinking (period 2)             1.8235  ( 0.5306   )   -0.0178   
#>   17. eval drinking linear shape                0.4025  ( 0.2209   )   -0.0787   
#>   18. eval drinking quadratic shape            -0.5610  ( 0.2814   )    0.0427   
#>   19. eval drinking average alter               1.2454  ( 0.6931   )    0.0041   
#> 
#> Overall maximum convergence ratio:    0.1988 
#> 
#> 
#> Total of 3427 iteration steps.
# Make a nicer listing of the results
siena.table(NBans, type="html", sig=TRUE)

This produces an html file; siena.table can also produce a LaTeX file.

Assignment 2

2a.

Replace the average alter effect by average similarity (avSim) or total similarity (totSim) and estimate the model again.

2b.

Add the effect of smoking on drinking and estimate again.

Assignment 3

Read Sections 13.3 and 13.4 of the Siena Manual, download scripts SelectionTables.r and InfluenceTables.r from the Siena website, and make plots of the selection table and influence table for drinking.