Skip to content

Output Format

Results are written as CSV files. There are two types of output files: rating output and questionnaire output.

Rating Output

For each subject there will be one rating output file. It is named as follows:

<participant_id>_<datetime>_<method>.csv

For example: 1_20250115-143055_ACR.csv

Columns

The CSV file contains one line per video rated, plus additional lines for breaks:

Column Description
video_position Zero-based index of the video in the playlist. -1 for BREAK entries.
video_name Filename of the video (e.g., video1.mp4). BREAK for break entries.
rating Integer rating value. For ACR: 0=Excellent, 1=Good, 2=Fair, 3=Poor, 4=Bad. For continuous: 0-100.
rated_at ISO8601 timestamp when the rating was submitted (e.g., 2025-01-15T14:32:05+01:00).
rating_duration Time in seconds the subject took to submit the rating (e.g., 2.345). Empty for time-continuous ratings.

Notes

  • The rating, rated_at, and rating_duration columns are empty for BREAK entries.
  • For time-continuous ratings, there will be one line per second of video playback, with the rating column containing the current rating value at that second, and the rated_at column indicating when that rating was recorded.

Why Track Rating Times?

The rating_duration column can be used to detect issues with single votes caused by observers struggling to decide on a rating.

For more information on using rating duration for quality assessment, see: Robitza, W., and Hlavacs, H. (2014). Assessing the validity of subjective QoE data through rating times and self-reported confidence. In 2014 Sixth International Workshop on Quality of Multimedia Experience (QoMEX) (pp. 297-302). Singapore: IEEE.

Questionnaire Output

If pre- and/or post-questionnaires are defined in the JSON config, answers are logged to separate CSV files:

<participant_id>_<datetime>_questionnaire_<type>.csv

For example: 1_20250115-143050_questionnaire_pre.csv

Columns

The CSV uses a tidy data format with one row per answer:

Column Description
question_number 1-based index of the question.
question_type Type of question: number, radio, multiple-choice, or text.
question The question text.
answer The answer value. For multiple-choice, each selected option produces a separate row.
answered_at ISO8601 timestamp when the question was answered.
answer_duration Time in seconds the subject took to answer.

Note

For multiple-choice questions, each selected option is written as a separate row with the same question_number, answered_at, and answer_duration. This tidy format makes analysis easier.