Collapses a metadata tibble (one row per component result) to one row per study result, which is the unit that means "one run by one participant". Counts of people, duplicate detection and quality flags belong at this level; a participant who reloaded a component has several component results but one study result.
Arguments
- metadata
A metadata tibble from
jatos_results_metadata(),jatos_read_metadata()orjatos_flatten_metadata().- participant
Name of the column that identifies a participant, for
n_runs;NULLusesquery_prolific_pidwhen present.
Value
A tibble with the study and study-result columns of metadata,
plus n_component_results, n_components (distinct component ids),
n_finished (component results in state FINISHED), data_size
(bytes, summed), n_files, first_component_start,
last_component_end, n_runs (when a participant key is available;
NA where the key is NA), any extra columns of metadata, and
conflicts (list of column names; only present when metadata has
extra columns).
Details
Columns that are not part of the metadata contract (for example
participant_id added by jatos_extract_fields(), or a column you
joined yourself) are carried over with their first non-NA value per
study result. When the component results of one study result disagree
on such a column, the value becomes NA, the column name is listed in
conflicts, and a warning names the study results concerned. The
<field>_status columns of jatos_extract_fields() (recorded in the
jatosr_extracted attribute) and the local file, file_size and
status columns are left out.
Prolific returners and reloaded links produce several study results per
person. When a participant key is available, n_runs counts the study
results per key value, so duplicates are visible before N is counted:
participant names the column (a query_* column from
jatos_url_query(), a field from jatos_extract_fields(), or
worker_id); with participant = NULL the column query_prolific_pid
is used when present, and n_runs is left out otherwise. A message
reports how many keys have more than one run.
Examples
meta <- jatos_flatten_metadata(
system.file("extdata", "metadata.json", package = "jatosr")
)
jatos_study_results(meta)[, c("study_result_id", "n_component_results", "data_size")]
#> # A tibble: 4 × 3
#> study_result_id n_component_results data_size
#> <int> <int> <dbl>
#> 1 9001 1 2048
#> 2 9002 2 1536
#> 3 9003 1 0
#> 4 9004 2 1000
jatos_study_results(meta, participant = "worker_id")[, c("study_result_id", "worker_id", "n_runs")]
#> # A tibble: 4 × 3
#> study_result_id worker_id n_runs
#> <int> <int> <int>
#> 1 9001 501 1
#> 2 9002 502 1
#> 3 9003 503 1
#> 4 9004 504 1