Re-identification risk from small groups and combined details

Name:

Date:

Class:

Quick Questions

Shade in the A or B circle for each question.

1. A dataset has no names, but only one student is aged 13, lives in Elmwood and is in Year 8. Is this student safe from re-identification?

A No, they are a group of one, so someone could work out who they are

B Yes, because there is no name in the data

2. Which group size hides a person best?

A A group of 1 student who has unique details

B A group of 30 students who share the same details

3. Age alone is a weak identifier, but age + suburb + year level together can point to one person. Which statement is correct?

A Combining indirect identifiers makes the matching group smaller

B Combining details always makes the group bigger

4. A school survey shows only one Year 7 student from the suburb of Rosehill. What is the best way to lower the risk?

A Keep the exact suburb and add the student's birth date

B Widen the detail, such as showing a region instead of the suburb

5. Which change makes groups bigger and re-identification harder?

A Changing an exact age of 13 into the age band 12-14

B Changing an age band of 12-14 into the exact age of 13

Match the Pairs

Draw a line connecting each item on the left to its match on the right.

Left column		Right column
1SELECT COUNT(*) FROM students WHERE yearLevel = 8 AND house = 'Blue';		ACounts how many students match a mix of details. If the answer is 1, that person is a group of one and could be re-identified.
2IF(B2<=12,"11-12","13-14")		BPuts exact ages into wider age bands so more people share the same value and each group is bigger.
3Age, suburb and year level may each look harmless, but together they can match only one student.		CSELECT COUNT(*) FROM students WHERE yearLevel = 7 AND house = 'Red' AND points = 120;
4Removing a detail from a dataset makes the groups bigger and lowers the re-identification risk.		DA group of one means the details match just one person, so that person can be re-identified even without a name.
5Only the name column is a risk, so keeping age, suburb and year level is safe.		ESELECT yearLevel, house FROM students;
6A group of one student is fine because the dataset has no names.		FIndirect identifiers such as age, suburb and year level can be combined to match a very small group, even a group of one, so they are risky too.

Definitions

re-identification

Fill in the Blanks

Complete each line of code by filling in the missing parts.

SELECT _____(*) FROM students WHERE yearLevel = 7;

SELECT COUNT(*) FROM students WHERE yearLevel = 8 ___ house = 'Blue';

SELECT yearLevel, house, COUNT(*) FROM students GROUP BY yearLevel, _____;

IF groupSize = _ THEN widen the detail ENDIF

SELECT yearLevel, COUNT(*) FROM students GROUP BY _________;

Spot the Error

Shade in the letter of the line that contains an error.

Question 1

A age ← 13

B suburb ← "Hill Park"

C groupSize ← 1

D IF groupSize > 5 THEN

E risk ← "high"

Question 2

A SELECT yearLevel, house, COUNT(*)

B FROM products

C GROUP BY yearLevel, house

D HAVING COUNT(*) = 1;

Question 3

A Age 13: 40 students

B Age 13 + suburb: 60 students

C Age+suburb+year: 1 student

Question 4

A SELECT * FROM students

B WHERE yearLevel = 8

C AND house = 'Red';

Question 5

A age ← 13

B IF age BETWEEN 12 AND 14

C ageBand ← "12-14"

D SHOW age

Answer Key

Re-identification risk from small groups and combined details

Section A: Quick Questions

A — No, they are a group of one, so someone could work out who they are

B — A group of 30 students who share the same details

A — Combining indirect identifiers makes the matching group smaller

B — Widen the detail, such as showing a region instead of the suburb

A — Changing an exact age of 13 into the age band 12-14

Section B: Match the Pairs

SELECT COUNT(*) FROM students WHERE yearLevel = 8 AND house = 'Blue'; → Counts how many students match a mix of details. If the answer is 1, that person is a group of one and could be re-identified.

IF(B2<=12,"11-12","13-14") → Puts exact ages into wider age bands so more people share the same value and each group is bigger.

Age, suburb and year level may each look harmless, but together they can match only one student. → SELECT COUNT(*) FROM students WHERE yearLevel = 7 AND house = 'Red' AND points = 120;

Removing a detail from a dataset makes the groups bigger and lowers the re-identification risk. → SELECT yearLevel, house FROM students;

Only the name column is a risk, so keeping age, suburb and year level is safe. → Indirect identifiers such as age, suburb and year level can be combined to match a very small group, even a group of one, so they are risky too.

A group of one student is fine because the dataset has no names. → A group of one means the details match just one person, so that person can be re-identified even without a name.

Section C: Definitions

re-identification — Working out who a person is by matching a mix of details, like age, suburb and year level, that fits only a very small group, even a group of one. Bigger groups hide people better.

Section D: Fill in the Blanks

SELECT COUNT(*)
FROM students
WHERE yearLevel = 7;

SELECT COUNT(*)
FROM students
WHERE yearLevel = 8 AND house = 'Blue';

SELECT yearLevel, house, COUNT(*)
FROM students
GROUP BY yearLevel, house;

IF groupSize = 1 THEN
    widen the detail
ENDIF

SELECT yearLevel, COUNT(*)
FROM students
GROUP BY yearLevel;

Section E: Spot the Error

D — Small groups are the risky ones, because fewer people share your details. The test should be groupSize < 5, so a group of one is marked high risk.

B — yearLevel and house are columns in the students table, not in products. The query should read FROM students to find groups of one.

B — Each extra detail you combine makes the matching group smaller, not bigger. Age 13 plus suburb must be fewer than 40 students, such as 6.

A — To check the size of a group you need COUNT(*). SELECT * lists every column about each student instead of counting the matches.

D — The exact age puts the person back in a tiny group. It should show ageBand, the wider range, so more people share the same detail.