Build a performance matrix from rated matrix items

The third collection path: respondents rate every alternative on every criterion with ordinary matrix items, one item per criterion with the alternatives as its rows, and the decision matrix is the per-cell aggregate. No new item type is needed. Per-cell counts and standard deviations are kept so the report can show how firm each cell is.

Usage

sframe_rated_matrix(data, instrument, items, statistic = c("mean", "median"))

Arguments

data

A data frame of responses.

instrument

An sframe instrument declaring every id in items.

items

Character vector of "matrix" item ids, one per criterion, in the intended criterion order. Every item must declare the same matrix_items (the alternatives) in the same order. When a decision block also collects weights through a weights_item whose criterion names differ from these ids, the items pair with its criteria in this order, and the result notes each pairing.

statistic

"mean" or "median".

Value

A list with matrix (alternatives x criteria, with dimnames), n, sd, alternatives, criteria, and statistic.

See also

sframe_collected_weights()

Examples

q5    <- sf_choices("q5", 1:5,
           c("Very poor", "Poor", "Fair", "Good", "Excellent"))
price <- sf_item("rate_price", "Rate each supplier: value",
                 type = "matrix", matrix_items = c("Alpha", "Basilica"),
                 choice_set = "q5")
study <- sf_instrument("Supplier selection", components = list(q5, price))
responses <- data.frame(rate_price__Alpha = c(3, 4), rate_price__Basilica = c(5, 5))
rm <- sframe_rated_matrix(responses, study, "rate_price")
rm$matrix
#>          rate_price
#> Alpha           3.5
#> Basilica        5.0