1. Overview

The Long format is designed for detailed, question-by-question analysis. It includes all the data stored in the platform and is identical to the data used internally for reporting. Each row represents a single respondent's answer to a single survey question. This format is particularly useful when you want to:

• Explore how different questions were answered across respondents.
• Pivot, filter, and group responses in tools like Excel, R, or Python.
• Work with multi-choice and grid questions without needing to manage many columns.

2. File Structure & Layout

Each row corresponds to one answer to one option of one question from a respondent. Respondents therefore have multiple rows, one for each question they encountered.

Example (first 5 rows):

respondent_id	status	question	reporting_id	type	response	raw_response	responded	timestamp	weight
00044f3b-ec63-2e17-9b5c-970e0efd5a8b	Terminated	How old are you?	Age	NumericQuestion	35-44	36	1	2025-07-30 16:27:06.025	0.638
00044f3b-ec63-2e17-9b5c-970e0efd5a8b	Terminated	What is your gender?	Gender	MultiChoiceQuestion	Male	Male	1	2025-07-30 16:27:06.068	0.638
00044f3b-ec63-2e17-9b5c-970e0efd5a8b	Terminated	What is your gender?	Gender	MultiChoiceQuestion	Female	Female	0	2025-07-30 16:27:06.068	0.638

3. Key Columns

• respondent_id – Unique identifier for each participant.
• status – Final survey status (e.g., Completed, Terminated).
• question – Full wording of the question asked.
• reporting_id – The labeled identifier for the question as set in the dashboard (e.g., Age, Gender).
• line_number – The line number of the question in the survey script.
• type – Type of question (NumericQuestion, MultiChoiceQuestion, OpenEnd, etc.).
• response – The recoded, human-readable response category (e.g., 35-44).
• raw_response – The raw value stored (e.g., 36).
• responded – Indicates whether and in what order the respondent selected the option. 0 = not selected, 1 = selected first, 2 = selected second, and so on.
• timestamp – Time when the answer was submitted.
• weight – Weighting factor applied to this respondent’s answers for statistical adjustment.

4. Data Representation

Single-choice questions

Stored as one row with responded=1.

Multi-choice questions

Stored as multiple rows per respondent per option. The chosen options have responded>0, with the number indicating the order in which the options were chosen. Unchosen options have responded=0.

Example: Multi-choice question

Question: Which of the following fruits do you like? (Select all that apply)

respondent_id	question	response	responded
r1	Fruits	Apple	1
r1	Fruits	Banana	0
r1	Fruits	Orange	2

Here, the respondent chose Apple first, Orange second, and did not select Banana.

Numeric questions

Both response (bucketed/cleaned category, e.g. 35-44) and raw_response (e.g. 36) are provided.

Open-end questions

The full text appears in response and raw_response.

5. Missing & Special Values

• Non-responses may appear with responded=0 and empty raw_response.
• "Prefer not to say" or similar options appear as normal response categories.
• Terminated respondents may have partial rows depending on where they dropped out.

6. Weighting

• Apply the weight column in analysis to ensure results reflect population targets.

7. Best Practices

• Use pivot tables (Excel) or groupby (Python/Pandas) to aggregate responses.
• For multi-choice questions, include all rows where responded>0 to capture all selected options. Use the order number if you need to analyze sequence of selection.
• When comparing across formats, match on reporting_id (long) to variable codes (wide/SPSS).

8. When to Use Long Format

• For deep exploratory analysis.
• When handling **multi-select ** or grid questions where wide format becomes cumbersome.
• When exporting data into R/Python for custom cleaning, text analysis, or advanced visualization.