Skip to contents

Indicators of automated or careless responding for every session of a results table

dq_flags_by_session.Rd

Applies dq_flags() to each row of a results table: the data_quality columns of the row are read as its page records, the agent_probe columns as what was typed into the hidden fields, and the optional columns with request headers are passed on when they exist.

Usage

dq_flags_by_session(
  results,
  pages = "^dq_p[0-9]+$",
  probes = "^dq_agent_p[0-9]+$",
  srv_ua = "dq_srv_ua",
  sig_agent = "dq_sig_agent",
  ch_platform = "dq_ch_platform",
  ch_mobile = "dq_ch_mobile",
  session = "session",
  flagged_only = TRUE
)

Arguments

results

a data frame of survey results with one row per session, for example from formr_results().

pages

a regular expression that matches the names of the data_quality columns.

probes

a regular expression that matches the names of the agent_probe columns.

srv_ua, sig_agent, ch_platform, ch_mobile

the names of the columns that hold the User-Agent header (a browser item), the Signature-Agent header and the two client hints (server items). A column that does not exist is ignored; NULL leaves it out.

session

the name of the column that identifies a session. Row numbers are used when there is no such column.

flagged_only

return only the indicators that are flagged (the default), or all of them.

Value

A data frame with the columns session, group, indicator, value and flag, one row per session and indicator.

Details

The defaults match the column names used in formr's example survey (dq_p1, dq_p2, ... for the records, dq_agent_p1, ... for the hidden fields). Use your own names or patterns if you named the items differently.

See also

dq_rrt() for the randomised-response question.

Examples

results <- data.frame(
  session = c("a", "b"),
  dq_p1 = c(
    '{"v":2,"ld":1,"keys":95,"txt_ch":90,"pd":5,"mv":160}',
    '{"v":2,"ld":1,"keys":0,"txt_ch":90,"fld_nokey":1,"pd":5,"pd_ctr":5}'
  ),
  dq_agent_p1 = c(NA, "yes, a language model")
)
dq_flags_by_session(results)
#>   session       group                                             indicator
#> 1       b environment                              agent_probe field filled
#> 2       b       input      text answers mostly not typed or pasted (fields)
#> 3       b       input share of text typed via keys (keys / chars in fields)
#> 4       b       input                       mouse presses / at exact centre
#>                   value flag
#> 1 yes, a language model TRUE
#> 2                     1 TRUE
#> 3                     0 TRUE
#> 4                 5 / 5 TRUE
# how many indicators are flagged per session
table(dq_flags_by_session(results)$session)
#> 
#> b 
#> 4