Regresses diff(x)_t on a constant, x_{t-1} and lags lagged differences; the statistic is the t-ratio on x_{t-1}. The null is a unit root, so a statistic BELOW the critical value rejects non-stationarity. Critical values are MacKinnon's (1991) response surface for the constant-only regression, tau = b0 + b1/n + b2/n^2; they are returned rather than a p-value, because interpolating a p-value would need a table the book does not print.
Examples
V <- c(1, 2, 3, 4, 5, 6, 7, 8)
morie_adfur(V)
#> $stat
#> [1] 1.619352
#>
#> $gamma
#> [1] 1.285829e-13
#>
#> $se
#> [1] 7.940388e-14
#>
#> $lags
#> [1] 1
#>
#> $n
#> [1] 6
#>
#> $crit1
#> [1] -4.986511
#>
#> $crit5
#> [1] -3.460868
#>
#> $crit10
#> [1] -2.901198
#>
#> $stationary5
#> [1] FALSE
#>