Launch app App
Documentation Chapters

Metric Data Export

The Metric Data (CSV) export extracts any configured metric from the workspace metric library as CSV. It supports both Long (tidy) and Wide layouts, optional temporal filtering, and a sidecar provenance codebook.

AOIs are identified by their displayed name throughout: AOIs sharing a displayed name are merged, within a stimulus and across stimuli, matching the behavior everywhere else in the application. The names No_AOI and Any_Fixation are reserved for the synthetic slots (fixations outside every AOI, and the all-fixations aggregate).

Layout Formats

Long (tidy) Format

Best suited for analysis in modern statistical languages (R, Python Pandas, Julia) and software like JASP or jamovi. Emits one row per observation. Active dimension columns are dynamically included only when needed by the selected metrics:

  • Scalar metrics: stimulus × participant rows.
  • AOI-vector metrics: adds the AOI column (stimulus × participant × slot).
  • AOI-pair-matrix metrics: adds the From_AOI and To_AOI columns (stimulus × participant × cell).
  • Windowed metrics: adds the Window_Start and Window_End columns (stimulus × participant × window). Windows cover each participant’s own recording, so a participant who stopped recording early has fewer window rows, never padded zeros.
  • Participant-pair-matrix metrics: adds the Participant_B column (stimulus × unordered pair).

Example CSV Output

Participant_ID,Participant,Stimulus,AOI,Metric,Unit,Value
1,P01,Stimulus1,Logo,Fixation count,,3
1,P01,Stimulus1,No_AOI,Fixation count,,1
1,P01,Stimulus1,,Average fixation duration,ms,250.5

Fields are quoted only when they contain the delimiter, quotes, or line breaks.

Wide Format

Best suited for case-per-row analysis in traditional software like IBM SPSS. Emits one row per participant × stimulus case. Values are expanded into separate columns:

  • Scalar metrics → <Label> column.
  • AOI-vector metrics → <Label>_<AOI> column per slot (including No_AOI and Any_Fixation).
  • AOI-pair-matrix metrics → <Label>_<From>_to_<To> column per cell.
  • Participant-pair metrics (e.g. scanpath similarity) → <Label>_<Participant> column per participant in the group. Each row holds participant A’s line of the similarity matrix, so the wide file IS the full matrix grid, diagonal included.

Spaces in names become underscores (e.g. Absolute_dwell_time_Face), and a header that would repeat gets a _2, _3 suffix.

Temporal Exclusion: Wide format excludes windowed metrics, as mixed-grain wide tables are statistically unsound. Export time series in the long format instead.

Example CSV Output

Participant_ID,Participant,Stimulus,Fixation_count_Logo,Fixation_count_No_AOI,Fixation_count_Any_Fixation,Recurrence_rate
1,P01,Stimulus1,3,1,4,12.5
2,P02,Stimulus1,5,0,5,8.3

Missing Values

Missing values (for example, a participant who never fixated on an AOI when computing Time to First Fixation) are exported as empty cells.

No Sentinel Values: Sentinel values like -1 are never used, preventing downstream statistical models (such as group means) from being silently poisoned by negative numbers.

Sidecar Codebook

Checking the Include metric codebook option downloads the export as a ZIP archive containing three files, named after the loaded data file (data below; GazePlotter when there is none):

  1. data-MetricData.csv: the dataset itself.
  2. data-MetricData-codebook.csv: a sidecar table detailing the operational parameters and provenance of every exported column.
  3. data-MetricData.analysis.json: the workspace’s Analysis: the renames, merges, selections, and metric definitions behind the table, which the command line replays on other recordings.

The codebook contains the following columns:

ColumnDescription
MetricThe column label used in the data CSV
Metric_IDThe metric instance’s id, as it appears in the Analysis JSON
Base_MetricThe metric’s original registry name
Base_IdThe unique recipe identifier
UnitMeasurement units (e.g. ms, %, or empty for dimensionless)
Measurement_ClassStatistical class of the metric: extensive (counts, summed durations), intensive (rates, averages, latencies), proportion (binary indicators), or relational (participant-pair)
ParametersActive parameters, as shown in the metric library (e.g. Fixation pairs, Step 2)
ProjectionSelected projection (e.g. max across AOIs)
WindowWindowing parameters (e.g. 1000 ms window, 100 ms step)
Output_ShapeReshaped dimension classification
Time_RangeRange restriction (e.g. t ∈ [500, 2000] ms or full)
AOIsThe AOIs the values were computed on (All, or their names), noting when No_AOI was left out
ParticipantsList of participants included
StimuliList of stimuli included
AOI_MissingDisplays true if any stimulus lacked the AOI referenced by a projection

Exporting Data

The modal guides you through five collapsible steps, one open at a time. Every collapsed step shows a one-line summary of its current selection, and its badge turns into a check when the step is complete.

  1. Open the Export menu in the header and select Metric Data (CSV). Or click Download data in a metric plot’s pane: the steps then start preset to that plot’s metrics, stimulus, participant selection, time range, AOI selection and Hide No AOI data setting.
  2. Choose metrics: check the metrics to include. On longer lists a search bar appears; Select Found and Deselect Found apply the current matches to the selection in one click. Right-click any item, here or in the later lists, for Select only this or Select all but this.
  3. Choose stimuli and Choose participants: check items the same way. The selection chips above the participant list select or deselect an entire participant selection at once; individual checkmarks refine the result. Right-click a chip for Select only these or Select all but these (its members).
  4. Choose AOIs: tick the AOIs to compute on (all by default). The list shows only the AOIs of the stimuli chosen above. Your AOI selections appear as chips above the list and work like the participant chips. The metrics are computed exactly as in a plot whose AOI selection holds the ticked AOIs: only they get rows or columns, and fixations on the other AOIs count as No_AOI. The step is left out when the chosen stimuli have no AOIs. Untick Include No AOI data to leave out the No_AOI rows and columns, as a plot’s Hide No AOI data toggle does; no other value changes, and Download data presets it from the plot.
  5. Configure the file: choose the Format (Long (tidy) or Wide), and the Delimiter and Decimal Separator according to your regional system settings. Each CSV export starts from the delimiter and decimal separator of your last CSV export, remembered in this browser, and the collapsed step names them when they are not comma and dot. Optional: enter Time Start (ms) and Time End (ms) boundary limits to crop the evaluation window (leave them empty for the full recording). The Format and Include metric codebook choices are remembered the same way; Download data on a plot whose metrics only fit the long format starts in Long (tidy). The step ends with a preview of the exact columns the file will contain. Switching to Wide deselects time-series metrics and says so.
  6. Click Export Metrics.

Sliding Windows: Sliding-window metrics evaluate dense overlaps and can generate very large files. Opt-in to windowed metrics explicitly when exporting.