Skip to contents

Linear combination of panel waves, with honest between-wave variance

Source: R/variance-panel.R
panel_estimate.Rd

Estimates any linear combination psi = sum(contrast * theta_wave) of a statistic across panel waves – an annual average (contrast = rep(1/W, W)), a net change (contrast = c(-1, 1), i.e. what change_estimate() does), a semester contrast, etc. – and its variance a' Sigma a, where Sigma is the between-wave covariance matrix that the overlap of a rotating panel induces. That covariance is taken straight from the coordinated replicates of wave_bootstrap() or wave_jackknife(), so it is correct without any analytic covariance formula. Treating the waves as independent (the diagonal of Sigma only) understates the variance of an average and overstates that of a change.

Usage

panel_estimate(
  wb,
  statistic,
  contrast = NULL,
  waves = NULL,
  level = 0.95,
  ci_type = c("normal", "t"),
  df = NULL
)

panel_mean(wb, variable, contrast = NULL, waves = NULL, level = 0.95)

panel_total(wb, variable, contrast = NULL, waves = NULL, level = 0.95)

Arguments

wb

a weightflow_wave_boot from wave_bootstrap() or a weightflow_wave_jack from wave_jackknife().

statistic

a function function(w, data) returning one number, evaluated per wave.

contrast

numeric weights, one per wave, defining the combination; defaults to the equal-weight average rep(1/W, W). May be named by wave label (matched to waves).

waves

optional character vector of the waves to combine (defaults to all, in order).

level

confidence level for the normal-approximation interval.

ci_type

"normal" (default, z) or "t" (Student t with df df).

df

degrees of freedom for the "t" interval; NULL uses the object's design df.

variable

name of a numeric variable present in every wave.

Value

an object of class weightflow_panel_estimate with estimate, se, V, Vind (the independence-assuming variance), deff = V / Vind, the covariance matrix Sigma, the per-wave point estimates, the contrast, and the confidence interval.

See also

change_estimate(), wave_bootstrap(), wave_jackknife()

Examples

waves <- lapply(1:3, function(t)
  weighting_spec(subset(panel_ine, ola == t & disp == "R"), base_weights = w_base))
names(waves) <- c("T1", "T2", "T3")
wb <- wave_bootstrap(waves, replicates = 100, strata = "estrato", psu = "psu",
                     seed = 1, progress = FALSE)
rate <- function(w, d) weighted.mean(d$desocupado, w, na.rm = TRUE)
panel_estimate(wb, rate)                       # average unemployment level over the waves
#> <weightflow panel estimate>
#>   contrast   : T1=0.333333, T2=0.333333, T3=0.333333
#>   estimate   : 0.275714   SE 0.00797932
#>   95% CI    : [0.260075, 0.291353]
#>   V = 6.367e-05  vs  V(independent) = 3.59e-05  (ratio 1.773)
#>   between-wave covariance RAISES the variance (an average-type contrast).
panel_estimate(wb, rate, contrast = c(-1, 0, 1))  # T3 - T1 change
#> <weightflow panel estimate>
#>   contrast   : T1=-1, T2=0, T3=1
#>   estimate   : 0.00100547   SE 0.0127257
#>   95% CI    : [-0.0239364, 0.0259473]
#>   V = 0.0001619  vs  V(independent) = 0.0002196  (ratio 0.737)
#>   between-wave covariance LOWERS the variance (a change-type contrast).