The task and original case
Code three open-text responses while keeping coded mentions separate from respondents. The task produces a small codebook and a disagreement record. Allowing multiple labels is a deliberate choice; it means theme counts need not add up to the number of people.
| Case element | Value or requirement |
|---|---|
| Inputs | R 1 mentions price and support;R 2 price;R 3 support; allow multiple labels per response. |
| Build | A codebook, coded examples and a disagreement-resolution note. |
This is an original invented exercise. It is independent practice, with no employer assessment or open-vacancy claim. Download the inputs, attempt your own result, then open the separate answer key.
Data and worked inputs
Build a result another person can follow
R1 mentions high price and slow support. R2 mentions high price; R3 mentions slow support. Define price and support codes with inclusion rules and a short example for each. Allow both codes on a response, and retain the original text beside its response ID.
Store assignments as response–code pairs, with at most one occurrence of each code per response under this exercise's rule. Python's CSV documentation supports preserving quoted text during import; the interpretation rules are your authored codebook. Ask a second reviewer to code independently before comparing assignments. Record ambiguous wording and the rule used to resolve a disagreement rather than pretending the supplied labels are universal truth.
Sources[1]
Check the worked result
There are three responses and four coded mentions: price appears on R1 and R2; support appears on R1 and R3. Each theme has two mentions and occurs in two of three responses. Their respondent shares can sum beyond 100% because labels overlap.
A chart titled “Four customers” would be wrong. Use labels such as coded mentions or responses mentioning each theme. The small invented sample demonstrates the counting method; it cannot support a population-level claim about customer priorities.
Make the failure visible
Repeat the word price twice within R2. Under the one-code-per-response rule, price still receives one assignment for R2. Test a response that mentions neither code and retain it as uncoded. A keyword match can be a starting aid, but a sentence denying a price concern may require a different inclusion decision than simply containing the word.
Revise it and ask someone to reproduce it
Ask a peer to code an additional response saying “Support is fine, but the price is high.” Discuss whether positive support mentions belong in this codebook or need a sentiment distinction. Update the rule, recode affected responses and keep the previous version so changes in counts have an inspectable explanation.
- Original response text and code assignments remain visible.
- Four mentions are distinguished from three responses.
- Disagreements trigger a documented rule revision and recoding.
Describe the work accurately
After completing your own version, you could adapt this project bullet: “Created an independent multi-label text-coding exercise with retained responses, explicit code definitions and separate mention and respondent counts.”
Use Projects or Independent practice as the heading. Keep the input files, result and revision together. Describe only your own completed work; the invented numbers are model outputs, not employer results or measured business impact.