remstats (version 4.1.0)

active_effects: Active-state statistics for Duration Relational Event Models

Description

Constructor functions for active-state effects used inside formulas passed to remstats (when reh is a remify_durem object). They capture properties of the currently active event network — events that have started but not yet ended — at each time point in a duration relational event sequence.

Usage

activeTie(scaling = c("none", "std"), consider_type = "ignore")
activeReciprocalTie(scaling = c("none", "std"), consider_type = "ignore")
activeOutdegreeSender(scaling = c("none", "std"), consider_type = "ignore")
activeIndegreeReceiver(scaling = c("none", "std"), consider_type = "ignore")
activeTotaldegreeSender(scaling = c("none", "std"), consider_type = "ignore")
activeTotaldegreeReceiver(scaling = c("none", "std"), consider_type = "ignore")
activeTotaldegreeDyad(scaling = c("none", "std"), consider_type = "ignore")
activeSharedPartners_otp(scaling = c("none", "std"), consider_type = "ignore")
activeSharedPartners_itp(scaling = c("none", "std"), consider_type = "ignore")
activeSharedPartners_osp(scaling = c("none", "std"), consider_type = "ignore")
activeSharedPartners_isp(scaling = c("none", "std"), consider_type = "ignore")
activeDegreeMin(scaling = c("none", "std"), consider_type = "ignore")
activeDegreeMax(scaling = c("none", "std"), consider_type = "ignore")
activeDegreeDyad(scaling = c("none", "std"), consider_type = "ignore")
activeSharedPartners(scaling = c("none", "std"), consider_type = "ignore")

Value

A named list with elements effect, scaling, and

consider_type, consumed by duremstats.

Arguments

scaling

Scaling applied to the raw statistic before returning:

"none"

Raw counts (default).

"std"

Per-time-point standardisation: \((x - \bar{x}) / \mathrm{sd}(x)\), computed over the \(D\) dyads in the fixed risk set.

consider_type

Character (or logical). How event types are handled:

"ignore" or FALSE

Aggregate over all event types (default). Counts all currently active events regardless of type.

"separate" or TRUE

Compute one statistic per event type. Only active events of that type contribute. Output effect names are suffixed with the type label, e.g. activeOutdegreeSender.X.start.

"interact"

Compute one statistic per (past-event type x dyad type) combination (\(C^2\) slices). Requires extend_riskset_by_type = TRUE in the remify call; otherwise silently coerced to "separate".

Details

Each function returns a configuration list consumed internally. The functions are passed inside a formula:

  remstats(reh,
    start_effects = ~ activeTie() + activeOutdegreeSender(scaling = "std"),
    end_effects   = ~ activeOutdegreeSender())

Directed-network effects:

activeTie()

Whether there is currently an active event from actor \(i\) to actor \(j\) (binary, 0/1). Note: in end_effects this is always 1 by definition (only currently active dyads are at risk of ending) and should not be included as a predictor.

activeReciprocalTie()

Whether there is currently an active event from actor \(j\) to actor \(i\) (binary, 0/1). Captures reciprocity in the active network.

activeOutdegreeSender()

Number of currently active events in which actor \(i\) (sender) is involved as sender (out-degree in the active-event network).

activeIndegreeReceiver()

Number of currently active events in which actor \(j\) (receiver) is involved as receiver.

activeTotaldegreeSender()

Total active degree of actor \(i\): active events in which \(i\) appears as either sender or receiver.

activeTotaldegreeReceiver()

Total active degree of actor \(j\).

activeTotaldegreeDyad()

Sum of the total active degrees of actors \(i\) and \(j\): \(\deg(i) + \deg(j)\).

activeSharedPartners_otp()

(Advanced) Number of actors \(h\) for whom \(i \to h\) and \(h \to j\) are both currently active (outgoing two-path). Rarely informative when the active network is sparse.

activeSharedPartners_itp()

(Advanced) Incoming two-path: actors \(h\) with \(h \to i\) and \(j \to h\) both active.

activeSharedPartners_osp()

(Advanced) Outgoing shared partners: actors \(h\) with \(i \to h\) and \(j \to h\) both active.

activeSharedPartners_isp()

(Advanced) Incoming shared partners: actors \(h\) with \(h \to i\) and \(h \to j\) both active.

Undirected-network effects:

activeTie()

Whether there is currently an active event between actors \(i\) and \(j\). Note: in end_effects this is always 1 by definition and should not be included as a predictor.

activeDegreeMin()

Minimum of the active degrees of \(i\) and \(j\): \(\min(\deg(i), \deg(j))\).

activeDegreeMax()

Maximum of the active degrees of \(i\) and \(j\): \(\max(\deg(i), \deg(j))\).

activeDegreeDyad()

Sum of the active degrees of \(i\) and \(j\): \(\deg(i) + \deg(j)\).

activeSharedPartners()

(Advanced) Number of actors \(h\) for whom both \((i,h)\) and \((j,h)\) are currently active. Rarely informative when the active network is sparse.

See Also

remstats for computing the statistics, and tie_effects / actor_effects for the standard (non-duration) effect overviews. Active-state effects are only available for the tie-oriented model of events with a duration.