Step 1a: Define node set(s) and attributes
We’ve loaded a dataset that defines its nodes and their attributes as
a data frame.
tail(states)
#> label present regime gdp
#> 162 SLB FALSE NA NA
#> 165 VCT TRUE NA 1.306663e-05
#> 167 BLZ TRUE NA 2.807250e-05
#> 169 KNA TRUE NA 1.236664e-05
#> 170 BRN FALSE NA NA
#> 193 KIR FALSE NA NA
class(states)
#> [1] "data.frame"
We need to define them as nodes so that goldfish knows what to do
with them.
states <- defineNodes(states)
head(states)
#> Number of nodes: 6
#> Number of present nodes: 6
#>
#> First 6 rows
#> label present regime gdp
#> 1 AND TRUE NA NA
#> 2 DNK TRUE 10 0.006248947
#> 3 FRA TRUE 5 0.062651475
#> 4 DEU TRUE -9 NA
#> 5 ITA TRUE 10 0.040385288
#> 6 LIE TRUE NA NA
class(states)
#> [1] "nodes.goldfish" "data.frame"
As you can see, the structure is the same, but the added class helps
goldfish interpret the data frame correctly.
Next we want to link events to this nodeset to indicate how the
attributes change. There are three attributes associated with this node
set, and they all change:
head(sovchanges)
#> time node replace
#> 1165 1960-04-03 01:00:00 SEN TRUE
#> 99 1960-04-27 00:00:00 TGO TRUE
#> 1317 1960-05-31 01:00:00 COD TRUE
#> 1316 1960-05-31 01:00:00 MLI TRUE
#> 100 1960-06-20 00:00:00 MLI TRUE
#> 101 1960-06-26 00:00:00 MDG TRUE
head(regchanges)
#> time node replace
#> 814 1960-01-02 LAO -1
#> 361 1960-01-16 CUB NA
#> 410 1960-01-16 VNM -8
#> 1269 1960-04-04 SEN -1
#> 1185 1960-04-16 KOR NA
#> 1449 1960-04-27 TGO -6
head(gdpchanges)
#> time node replace
#> 202 1961-01-01 DNK 0.006933842
#> 203 1961-01-01 FRA 0.068346742
#> 205 1961-01-01 ITA 0.044842760
#> 207 1961-01-01 NLD 0.013493834
#> 208 1961-01-01 PRT 0.003417517
#> 211 1961-01-01 ESP 0.013834301
The first attribute, in the second column of the data frame, is
special. It holds the (reserved) “present” variable that indicates
whether nodes are available to send or receive ties.
head(states$present) # or states[,2]
#> [1] TRUE TRUE TRUE TRUE TRUE TRUE
Some state nodes are not present (do not hold sovereign status) at
the start of our study period. But we do have an additional list of
events detailing how these states gain (and some lose) sovereign status
during the study period. To make sure goldfish recognises these changes
at the appropriate point, we “link” these events to the node set and
attribute we have defined.
states <- linkEvents(states, sovchanges, attribute = "present")
# If you call the object now, what happens?
states
#> Number of nodes: 154
#> Number of present nodes: 137
#> Dynamic attribute(s): Linked events
#> present sovchanges
#>
#> First 6 rows
#> label present regime gdp
#> 1 AND TRUE NA NA
#> 2 DNK TRUE 10 0.006248947
#> 3 FRA TRUE 5 0.062651475
#> 4 DEU TRUE -9 NA
#> 5 ITA TRUE 10 0.040385288
#> 6 LIE TRUE NA NA
That’s because the event list is only linked to the node set, to
update it when necessary. You can see what is linked to which attribute
here:
str(states)
#> Classes 'nodes.goldfish' and 'data.frame': 154 obs. of 4 variables:
#> $ label : chr "AND" "DNK" "FRA" "DEU" ...
#> $ present: logi TRUE TRUE TRUE TRUE TRUE TRUE ...
#> $ regime : num NA 10 5 -9 10 NA 10 -9 -1 NA ...
#> $ gdp : num NA 0.00625 0.06265 NA 0.04039 ...
#> - attr(*, "events")= chr "sovchanges"
#> - attr(*, "dynamicAttributes")= chr "present"
We can continue on with our other two dynamic monadic variables. Try
this for yourself now.
states <- linkEvents(states, regchanges, attribute = "regime") |>
linkEvents(gdpchanges, attribute = "gdp")
str(states)
#> Classes 'nodes.goldfish' and 'data.frame': 154 obs. of 4 variables:
#> $ label : chr "AND" "DNK" "FRA" "DEU" ...
#> $ present: logi TRUE TRUE TRUE TRUE TRUE TRUE ...
#> $ regime : num NA 10 5 -9 10 NA 10 -9 -1 NA ...
#> $ gdp : num NA 0.00625 0.06265 NA 0.04039 ...
#> - attr(*, "events")= chr [1:3] "sovchanges" "regchanges" "gdpchanges"
#> - attr(*, "dynamicAttributes")= chr [1:3] "present" "regime" "gdp"
Step 1b: Define networks
Next we want to define the dyadic or network elements: the bilateral
fisheries agreements and contiguity between the state nodes.
Since international fisheries management began before 1960, there
were already several bilateral fisheries agreements in place at the
start of our time period. This is shown in the matrix
bilatnet[1:12, 1:12] # head(bilatnet, n = c(12, 12))
#> AND DNK FRA DEU ITA LIE NLD PRT RUS SMR ESP SWE
#> AND 0 0 0 0 0 0 0 0 0 0 0 0
#> DNK 0 0 0 1 0 0 0 0 0 0 0 0
#> FRA 0 0 0 0 0 0 0 0 0 0 2 0
#> DEU 0 1 0 0 0 0 0 0 0 0 0 0
#> ITA 0 0 0 0 0 0 0 0 0 0 0 0
#> LIE 0 0 0 0 0 0 0 0 0 0 0 0
#> NLD 0 0 0 0 0 0 0 0 0 0 0 0
#> PRT 0 0 0 0 0 0 0 0 0 0 0 0
#> RUS 0 0 0 0 0 0 0 0 0 0 0 0
#> SMR 0 0 0 0 0 0 0 0 0 0 0 0
#> ESP 0 0 2 0 0 0 0 0 0 0 0 0
#> SWE 0 0 0 0 0 0 0 0 0 0 0 0
Two things can be observed here. What are they?
So that goldfish knows how to handle this network, we define it
so:
bilatnet <- defineNetwork(bilatnet, nodes = states, directed = FALSE)
We specify the nodes so that goldfish can check for consistency, and
relate the network to that nodeset when needed.
class(bilatnet)
#> [1] "network.goldfish" "matrix" "array"
str(bilatnet)
#> 'network.goldfish' num [1:154, 1:154] 0 0 0 0 0 0 0 0 0 0 ...
#> - attr(*, "dimnames")=List of 2
#> ..$ : chr [1:154] "AND" "DNK" "FRA" "DEU" ...
#> ..$ : chr [1:154] "AND" "DNK" "FRA" "DEU" ...
#> - attr(*, "events")= chr(0)
#> - attr(*, "nodes")= chr "states"
#> - attr(*, "directed")= logi FALSE
bilatnet
#> Dimensions: 154 154
#> Number of ties (no weighted): 22
#> Nodes set(s): states
#> It is a one-mode and undirected network
#>
#> First 6 rows and columns
#> AND DNK FRA DEU ITA LIE
#> AND 0 0 0 0 0 0
#> DNK 0 0 0 1 0 0
#> FRA 0 0 0 0 0 0
#> DEU 0 1 0 0 0 0
#> ITA 0 0 0 0 0 0
#> LIE 0 0 0 0 0 0
By stating that this is an undirected network, any updates will be
forced to be symmetric.
Now that goldfish recognises the matrix as a network, we can also
associate an event list that updates it. We use precisely the same
function as before (since we have defined our objects, goldfish knows
what to do and how to check it).
bilatnet <- linkEvents(bilatnet, bilatchanges, nodes = states)
bilatnet
#> Dimensions: 154 154
#> Number of ties (no weighted): 22
#> Nodes set(s): states
#> It is a one-mode and undirected network
#> Linked events: bilatchanges
#>
#> First 6 rows and columns
#> AND DNK FRA DEU ITA LIE
#> AND 0 0 0 0 0 0
#> DNK 0 0 0 1 0 0
#> FRA 0 0 0 0 0 0
#> DEU 0 1 0 0 0 0
#> ITA 0 0 0 0 0 0
#> LIE 0 0 0 0 0 0
TASK: You should now be able to do the same with the initial
contiguity network and the event list changing it.
contignet <- defineNetwork(contignet, nodes = states, directed = FALSE) |>
linkEvents(contigchanges, nodes = states)
class(contignet)
#> [1] "network.goldfish" "matrix" "array"
contignet
#> Dimensions: 154 154
#> Number of ties (no weighted): 175
#> Nodes set(s): states
#> It is a one-mode and undirected network
#> Linked events: contigchanges
#>
#> First 6 rows and columns
#> AND DNK FRA DEU ITA LIE
#> AND 0 0 0 0 0 0
#> DNK 0 0 1 0 0 0
#> FRA 0 1 0 0 1 0
#> DEU 0 0 0 0 0 0
#> ITA 0 0 1 0 0 0
#> LIE 0 0 0 0 0 0
Step 1c: Define dependent events
The final step in defining the data objects is to identify the
dependent events. Since treaty dissolution is a rarer, more complex, and
idiosyncratic process, we will only look at those events that create a
treaty tie between countries.
createBilat <- defineDependentEvents(
events = bilatchanges[bilatchanges$increment == 1,],
nodes = states,
defaultNetwork = bilatnet
)
Note that we define the default network to which this dependent event
list applies, as well as the node set, so that structural effects can be
interpreted correctly.
class(createBilat)
#> [1] "dependent.goldfish" "data.frame"
createBilat
#> Number of events: 69
#> Nodes set(s): states
#> Default network: bilatnet
#>
#> First 6 rows
#> time sender receiver increment
#> 36 1960-03-19 FIN SWE 1
#> 37 1960-06-23 FIN RUS 1
#> 38 1960-11-15 FIN NOR 1
#> 39 1960-11-17 GBR NOR 1
#> 40 1961-03-11 GBR ISL 1
#> 41 1961-07-19 DEU ISL 1
