The columns that are not there

Slide 1 / 22

What this lesson covers

• Start with the schema, not the data
• The principle, stated as a rule about columns
• What the coarsening costs, measured
• The one that is genuinely a loss
• The 62 blank age bands
• What to write in the methodology note
• What comes next

Slide 2 / 22

Start with the schema, not the data — In Python

import pandas as pd

referrals = pd.read_csv("protection-referrals-2024.v1.csv")
print(referrals.columns.tolist())
print(f"{len(referrals):,} cases, {len(referrals.columns)} columns")

Slide 3 / 22

Start with the schema, not the data — In R

library(dplyr)

referrals |> glimpse()

Slide 4 / 22

Start with the schema, not the data

Slide 5 / 22

Start with the schema, not the data

• Each row of that table is a decision someone made, and each one can be defended by naming the analysis it does not…

Slide 6 / 22

The principle, stated as a rule about columns

• Collect the minimum necessary. If no indicator needs a field, holding it is pure risk.
• Share less than you hold. The analysis extract is not the case file. Each step outward drops fields.
• Never share incident-level data outside the case management agency. Aggregate or do not send.
• Consent governs use. A person consenting to a referral has not consented to a research dataset.
• Safety first, always. Where an analysis and a survivor's safety conflict, the analysis loses.

Slide 7 / 22

The principle, stated as a rule about columns — In Python

minimum = {
    "who": ["age_band", "sex", "disability_reported"],
    "where": ["admin1", "admin2"],
    "when": ["referral_month"],
    "what": ["case_category", "service_requested"],
    "pathway": ["consent_to_refer", "referral_made", "referral_accepted",
                "days_to_first_service", "case_status"],
}
print({k: len(v) for k, v in minimum.items()})

Slide 8 / 22

The principle, stated as a rule about columns — In R

# Five groups of fields, each earning its place by an indicator that needs it.

Slide 9 / 22

The principle, stated as a rule about columns

• Every field in this extract maps to an indicator — days_to_first_service exists because the 72-hour clinical standard…

Slide 10 / 22

What the coarsening costs, measured — In Python

print(referrals["age_band"].value_counts(dropna=False))
print(referrals["referral_month"].nunique(), "distinct periods")
print(referrals["admin2"].nunique(), "areas")

Slide 11 / 22

What the coarsening costs, measured — In R

referrals |> count(age_band)
referrals |> summarise(periods = n_distinct(referral_month),
                       areas = n_distinct(admin2))

Slide 12 / 22

What the coarsening costs, measured

• Timeliness: measured in days from referral to service, which is already a duration and does not need a calendar…
• Seasonality: monthly is the finest resolution any protection trend is read at, and a weekly series on 1,850 cases…
• Age disaggregation: the bands are the reporting categories. A child, an adolescent, a working-age adult and an…
• So the coarsening costs nothing that anyone asked for — That is what makes it defensible, and it is why the argument…

Slide 13 / 22

The one that is genuinely a loss

• Dropping incident type means you cannot say which forms of violence the referral pathway serves worst — That is a real…

Slide 14 / 22

The one that is genuinely a loss — In Python

print("Question: does the pathway serve some incident types worse than others?")
print("Where it can be answered: inside the case management agency")
print("What leaves the agency: the finding, not the disaggregation")

Slide 15 / 22

The one that is genuinely a loss — In R

# The right answer to "we need incident type" is often "we will run it and
# send you the result", not "no".

Slide 16 / 22

The one that is genuinely a loss

• "You cannot have the file, and here is the answer to your question" is the professional response — and it is available…

Slide 17 / 22

The 62 blank age bands — In Python

missing = referrals["age_band"].isna()
print(f"{missing.sum()} cases with no age band ({missing.mean():.1%})")
print(referrals.loc[missing, "case_category"].value_counts())

Slide 18 / 22

The 62 blank age bands — In R

referrals |> filter(is.na(age_band)) |> count(case_category)

Slide 19 / 22

What to write in the methodology note — Example

Data handling

  Analysis extract holds 14 fields for 1,850 cases. It contains no name,
  contact detail, free text, incident date, location below admin2, exact age,
  incident type or perpetrator detail.

  Age is recorded in five bands and time in months. Both are sufficient for
  every indicator reported here, and both reduce the risk that a row can be
  matched to a person.

  Incident-level fields remain in the case management system. Analyses
  requiring them are run by the case management agency and reported as
  findings rather than shared as data.

  62 cases (3.4%) have no age band. Age-disaggregated figures are computed on
  1,788 cases and are labelled accordingly.

Slide 20 / 22

What to write in the methodology note

• Write this before the results, not in an annex — A protection report whose data handling section is at the back has…

Slide 21 / 22

What comes next

• Removing columns reduces risk and does not eliminate it.

Slide 22 / 22

Where this goes next

Read the full lesson, with runnable code Back to the lesson