# Worked example: from a Google Form to a report

This end-to-end example takes a survey export all the way to results pasted into Word. It
mirrors a real workflow; adapt the column names to your own study.

Imagine a short survey with questions including: age, monthly income, a test score, an
overall satisfaction rating (1–5), whether the person passed an exam (Yes/No), gender,
region, and a "which features do you use? (select all that apply)" checkbox question.

## 1. Import the export

1. **File ▸ Open…** and select the Forms `.xlsx`.
2. Open the **data viewer** to confirm the columns loaded and check their **types** (numbers
   as Numeric, text as Nominal). Note that the app added an **ID** column at the front.

```{figure} _static/img/worked-1-import.png
:alt: The imported survey data in the data viewer
:width: 100%

Step 1 — the imported data in the data viewer, with column types and the added ID column.
```

## 2. Clean the data

Add the processing steps you need (each becomes a card in the chain):

- **Impute Missing** on income with **Median**, to fill the few skipped answers.
- **Filter** to drop test rows — for example, keep only rows where age **is not empty**.
- **Outliers** (IQR) on income, if you want to exclude extreme values. Toggle it off later
  to compare.
- **Response Quality** to drop careless respondents — for example, those who straightlined the
  rating grid or skipped most questions.

Preview any step to see its effect (removed rows show in red). Steps that remove rows (Filter,
the outlier steps, Response Quality) list the flagged IDs as checkboxes — untick any to keep
that respondent.

```{figure} _static/img/worked-2-clean.png
:alt: Data-processing steps stacked as cards in the chain
:width: 100%

Step 2 — cleaning steps stacked as cards in the chain; each feeds the next.
```

## 3. Build derived variables

- **One-hot encoding** on *region* → 0/1 indicator columns, so region can enter a
  regression.
- **Split Multi-Select** on the features question → one 0/1 column per feature.
- **Calculate Scale** if you have several items that form a scale (Sum or Mean).

## 4. Run the analyses

Pick modules in the settings panel on the right and choose the relevant columns:

- **Descriptive Statistics** on age, income, score (add a histogram and box plot).
- **Correlation** among age, income, score (Pearson, with confidence intervals).
- **T-test/ANOVA** of score by *passed* (Yes/No).
- **Regression** predicting score from age, income, and the region indicators.
- **Multiple Response** on the split feature indicators.

Each result appears as a card on the left and updates if you change the cleaning steps.

```{figure} _static/img/worked-3-analyse.png
:alt: An analysis configured in the settings panel, with the assembled result cards
:width: 100%

An analysis configured on the right, with the resulting cards (tables and plots) assembled on
the left — ready to copy or export together.
```

## 5. Get it into your document

1. Choose your **Language** (**Settings ▸ Language**) and **Plot theme**
   (**Settings ▸ Plot theme**).
2. **File ▸ Copy All Results**, then **paste into Word** — tables arrive as editable tables,
   figures as images. (See {doc}`results-and-export` for the paste tips, including removing
   Word's paragraph indent so tables line up.)
3. Or **File ▸ Export Report (HTML)…** for a single self-contained file you can archive or
   print to PDF.

## 6. Save the project

**File ▸ Save As…** to a `.sp` file so you can reopen the entire session — data, steps, and
analyses — later.

---

That's the whole loop: **import → clean → derive → analyse → export**. Once it's set up,
re-running with new responses is just re-importing and letting the chain recompute.