Questionnaire scoring guide #

This guide walks through three real-world scoring scenarios, from simple to advanced. Each builds on the data transformation features, so you may want to read that page first.

The common starting point: you have a spreadsheet (e.g. from Google Forms) where each row is a participant and each column is a questionnaire item. The responses might be text labels (“Strongly agree”) or already numeric – either way, you need scale scores for analysis.

Example 1: PHQ-9 (simple sum) #

The PHQ-9 is a 9-item depression screening tool. Each item asks how often a symptom occurred over the past two weeks. There are no reverse-scored items – just recode and sum.

Response options: Not at all (0), Several days (1), More than half the days (2), Nearly every day (3). Total score range: 0–27.

Step 1: recode text to numbers #

If your data already has numeric values 0–3, skip this step.

1. Open Data transformation and click + Value recode
2. In the input panel, select all 9 PHQ items
3. The editor auto-fills the unique text values. Map them:

4. Under output options, choose Replace original values – the text is no longer needed
5. Name the rule (e.g. “PHQ-9 recode”) and click Save rule

Step 2: compute the total score #

1. Click + Formula
2. Select the same 9 items as input
3. Enter the formula:

sum(v1:v9)

4. Under output options, choose Create new variable(s) and name it PHQ9_Total
5. Save the rule

Done. You now have a PHQ9_Total column with scores from 0 to 27.

Example 2: Rosenberg Self-Esteem Scale (mixed item directions) #

The Rosenberg Self-Esteem Scale has 10 items on a 4-point scale. Items 3, 5, 8, 9, and 10 are negatively worded and need to be reversed before summing.

Response options: Strongly agree (3), Agree (2), Disagree (1), Strongly disagree (0). Total score range: 0–30.

Approach A: reverse at scoring time (recommended) #

This approach keeps the original item values intact – no recode step needed. The reversal happens inside the formula itself using the pattern (max - item), where max is 3 for a 0–3 scale:

1. Click + Formula
2. Select all 10 RSE items as input (in order)
3. Enter the formula:

v1 + v2 + (3-v3) + v4 + (3-v5) + v6 + v7 + (3-v8) + (3-v9) + (3-v10)

4. Create a new variable called RSE_Total
5. Save

The reversed items (v3, v5, v8, v9, v10) are flipped inline: a response of 0 becomes 3, 1 becomes 2, and so on. The original data stays unchanged, which makes it easier to verify individual items later – and it is what an IRT analysis of the same items needs: that module reverses the negatively keyed items itself, its careless-responding screen reads the answers as respondents gave them, and its keying detection cannot see a reversal made earlier.

Approach B: recode first, then sum #

If you prefer to have the reversed values stored explicitly:

1. + Standardize – select only the 5 reversed items (3, 5, 8, 9, 10), method Reverse scoring, scale range Min 0 and Max 3. Choose Create new variable(s) with a prefix such as r, so the reversed copies sit beside the originals. (A + Value recode mapping 0→3, 1→2, 2→1, 3→0 does the same by hand.)
2. + Formula – select the 5 regular items and the 5 reversed copies. Formula: sum(v1:v10). Create a new variable RSE_Total.

Both approaches produce identical results. Approach A is more compact; approach B leaves the reversed values visible in the data table. Don’t use Replace original values for items an IRT analysis may read – see keep originals intact.

Example 3: TIPI – Ten-Item Personality Inventory (multiple subscales) #

The TIPI measures the Big Five personality traits with just 10 items – two per trait, one regular and one reverse-scored. The scale runs from 1 (Disagree strongly) to 7 (Agree strongly), and each subscale is the average of its two items.

Here’s the structure:

Scoring with a multi-variable formula #

This is a perfect use case for the @ syntax – five subscales computed in a single rule.

1. Click + Formula
2. Select all 10 TIPI items as input, in item order (1–10)
3. Enter the formula:

@Extraversion = (v1 + (8-v6)) / 2
@Agreeableness = ((8-v2) + v7) / 2
@Conscientiousness = (v3 + (8-v8)) / 2
@EmotionalStability = ((8-v4) + v9) / 2
@Openness = (v5 + (8-v10)) / 2

4. Save the rule (output options are automatic with @ declarations)

The reversal pattern here is (8 - item) because the scale runs 1–7 (so max + 1 = 8). Each subscale score ranges from 1 to 7.

Why (8 - v) and not (7 - v)? The reversal formula is always (scaleMax + scaleMin) - value. For a 1–7 scale that’s (7 + 1) - v = 8 - v. For a 0–3 scale (like RSE above) it’s (3 + 0) - v = 3 - v. This preserves the original range.

General tips #

• Check your scale range. The reversal formula depends on it: (max + min) - value. For 1–5 Likert it’s 6 - v, for 0–4 it’s 4 - v, for 1–7 it’s 8 - v. See the formula reference for all available functions and syntax. The Standardize rule’s reverse scoring applies the same formula from a declared range, one rule for any number of items.
• Keep originals intact when possible. Using inline reversal (approach A) means you can always go back and inspect individual item responses without confusion about which values were recoded. Keep them intact whenever an IRT analysis is planned or may follow: the IRT module reverses the negatively keyed items itself, its careless-responding screen needs the answers as given, and its keying detection cannot see a reversal made in the data. The Reliability module likewise reverses items for the analysis alone.
• Name variables clearly. Use descriptive names like PHQ9_Total, RSE_Total, Extraversion – you’ll see these names in all analysis outputs.
• Rules apply in order. If you recode text to numbers in rule 1, rule 2’s formula will see the numeric values. You can verify intermediate results in the data preview at any point.
• Save as a project file. Your transformation rules are saved inside the project file (.json), so you can reopen the project later with all rules intact. You can also store rules in the rule library for reuse across projects.