Sampling and data collection AI SL
A sample is a small part of a population chosen so that it can speak for the whole group, and the method of choosing decides whether it can. This page shows you how to name each sampling method from a description, how to calculate a systematic or stratified sample, and how to explain bias in a survey. It then covers discrete and continuous data, how to judge a source, and the rule for spotting outliers. Start with the sampling methods, then move on to data types and collection.
Formulas at a glance
None of the sampling formulas, and not the outlier rule, is in the formula booklet, so learn them. The interquartile range that the outlier rule uses is in the booklet. Both sampling formulas are just direct proportion.
Stratified sampling
Stratum sample size stratum size
Not in the bookletSystematic sampling
Sampling interval
Not in the bookletInterpreting outliers
Outlier if more than from nearest quartile
Not in the bookletWhich method to use
Decide first whether the question asks you to name a method, to calculate a sample, or to judge whether a survey is fair. Each kind of question has its own short routine.
| The question has | Use | Notes |
|---|---|---|
| A description of how people were chosen, and asks you to name the method | Match the key feature: chance only (simple random), every th member (systematic), whoever is easiest to reach (convenience), a fixed number per group picked by the interviewer (quota), a proportional random share of each group (stratified) | Quota and stratified both use groups. In stratified sampling the members of each group are chosen at random |
| A population split into groups, and a total sample size | Stratified sampling: stratum size | Round each answer to a whole number, then check the groups still add up to the sample size |
| A numbered list of the whole population, and a sample size | Systematic sampling: interval | Choose a random start in the first interval, then add the interval each time |
| A survey, and asks why the results may be biased or unreliable | Name the group that is left out or over-represented, then say how that group would answer differently | A small sample is imprecise, but only the way it was chosen makes it biased |
| A quantity to classify | Counted values are discrete. Measured values are continuous | Number of goals and number of pets are discrete. Time, mass and height are continuous |
| A data set with an unusually large or small value | Find and with technology, then test and | An outlier may be a valid value or an error. Check the context before removing it |
| Two reports or sources that disagree | Compare how recent each is, how the data were collected, how large and representative the sample is, and whether the source has a stake in the result | Give specific reasons that refer to the situation |
Section 1 of 2 SL4.1
Sampling methods
A sampling method is the rule used to choose which members of a population are surveyed. Simple random, systematic, convenience, quota and stratified sampling each choose members differently, and the method decides how well the sample represents the population. It covers recognising each method from a description and the calculations for systematic and stratified samples.
Choosing a sampling method
Covered under IB syllabus reference SL4.1, which also lists convenience and quota sampling as effectiveness comparisons - but simple random, systematic and stratified are the three you'll actually have to calculate.
Simple random sampling
Number every member of the population, then use technology to generate random numbers and pick the sample - every individual and every combination has an equal chance of selection.
Not in the formula booklet - selection procedureSystematic sampling
Choose a random starting point within the first interval, then take every member at that interval after it.
Not in booklet - direct proportionStratified sampling
stratum size
Split the population into strata, then apply the same sampling fraction to each one so every group is represented proportionally.
Not in booklet - direct proportionThe reliability, bias and outlier definitions that go with this are in Data types and collection.
Key terms
What is a simple random sample?
Members are chosen purely by chance, so each member of the population has an equal chance of being selected. A list of the whole population is needed.
e.g. Giving each of 500 people a number and choosing 50 numbers with a random number generator.
What is convenience sampling?
Members are chosen because they are easy to reach, not by chance. It is quick and cheap, but the sample is often unrepresentative, so it is likely to be biased.
e.g. Surveying the first 30 people who walk past a shop on one morning.
What is quota sampling?
The population is divided into groups, and the interviewer must find a fixed number of people from each group to reflect the population. Within each group the people are chosen by whoever is available, not by chance.
e.g. To reflect a town that is 60% adults and 40% under 18s, an interviewer questioning 50 people asks 30 adults and 20 people under 18.
What is stratified sampling?
The population is divided into groups called strata, and a random sample is taken from each stratum. The number taken from each is proportional to the size of the stratum.
e.g. A school has 200 pupils in Year 7 and 300 in Year 8. A stratified sample of 50 takes 20 from Year 7 and 30 from Year 8.
Worked example, step by step
The question Calculator allowed [6 marks]
A veterinary group has 360 registered pets on one numbered list: 190 dogs, 120 cats and 50 other animals. The manager wants to survey the owners of a sample of 30 pets.
(a) A stratified sample of 30 pets is taken, using dogs, cats and other animals as the strata. Find the number of pets from each stratum.
(b) Give one reason why a stratified sample is better than a convenience sample made up of the first 30 pets to arrive at the surgery.
(c) Instead, a systematic sample of 30 pets is taken from the list. Find the sampling interval, and write down the first three pets chosen when the random start is the 5th pet on the list.
Worked solution: worked example
Answer: (a) 16 dogs, 10 cats and 4 other animals (b) it makes sure that each type of pet is included in proportion to its size (c) the interval is 12, and the first three pets are the 5th, 17th and 29th.
Step 1 - Find the sampling fraction.
The sample is 30 pets from a population of 360, so the sampling fraction is Every stratum is sampled at this same fraction.
Why? Using one fraction for every stratum is what makes a stratified sample proportional, so each kind of pet is represented in the right proportion.
Step 2 - Apply it to each stratum.
Dogs: . Cats: . Other animals: . Check: . M1 - Multiply a stratum size by the sampling fraction A1 - 16 dogs, 10 cats and 4 other animals
Why? A stratum sample must be a whole number of animals, so round each answer. Then add the three numbers to check that the rounded sample is still the right size.
Step 3 - Give a reason.
A convenience sample is whoever happens to arrive first, so it might contain no animals from the smallest group, or far too many dogs. A stratified sample takes each type in proportion to its size, so every group of owners is represented fairly. R1 - A reason that refers to representing each group in proportion (or to a group being missed or over-represented)
Why? A good reason names the group that could be missed or over-represented. The single word "biased" is not enough on its own.
Step 4 - Find the sampling interval.
M1 - Divide the population size by the sample size A1 - Interval
Why? In a systematic sample every 12th pet on the list is chosen. The interval is the spacing, found by dividing the list into 30 equal blocks.
Step 5 - List the first three pets.
The random start is the 5th pet, so the first three pets chosen are the 5th, the 17th () and the 29th (). A1 - 5th, 17th and 29th pets
Why? Only the starting point is random, and it must lie within the first interval (positions 1 to 12). After that, the spacing is fixed.
M1 Multiply a stratum size by the sampling fraction
A1 16 dogs, 10 cats and 4 other animals
R1 A reason that refers to representing each group in proportion (or to a group being missed or over-represented)
M1 Divide the population size by the sample size
A1 Interval
A1 5th, 17th and 29th pets
Now you try: enter your answers to be marked
These are real practice questions. Type your answer in each box and press Mark it. The worked solution and mark scheme appear straight away, and your first attempt is the one that counts. Use the keyboard for powers, roots and trig, and the working box for your notes.
Worked solution: practice question 1
Worked solution: practice question 2
Worked solution: practice question 3
Your score so far: 0 out of 8 marks
Common mistakes
- Mixing up systematic and random. Systematic sampling still needs a genuine random starting point inside the first interval - always taking position 1 as the start makes it a fixed, non-random rule.
- Not rounding stratified sample sizes sensibly. Stratified calculations often produce a decimal - round to the nearest whole number and check your rounded group sizes still add up to the total sample size.
- Calling convenience sampling "random". Choosing whoever is easiest to reach - people leaving a particular shop, say - is convenience sampling, not simple random sampling, even when the selection feels arbitrary.
Section 2 of 2 SL4.1
Data types and collection
A population is the whole group being studied, and a sample is the part of it that is actually surveyed. Data are discrete when they are counted and continuous when they are measured, and a data source is reliable only if its figures can be trusted. The topic covers these terms, how bias arises in sampling, and how to judge a source.
Outliers and data types
Covered under IB syllabus reference SL4.1. This part of the syllabus sets up the population/sample vocabulary, the discrete/continuous split, and the reliability and bias language used across the whole statistics course - plus a precise rule for spotting outliers.
The outlier rule
or
This is the exact boundary the IB uses to define an outlier. It flags a value as unusual - it doesn't automatically tell you to remove it.
Not in the formula booklet - syllabus definitionDiscrete vs continuous
Discrete data can only take separate, countable values - number of siblings, shoe size, goals scored. Continuous data can take any value in a range, usually from measuring - height, time, mass. The distinction decides which charts and calculations make sense.
Looking for sampling techniques instead? See Sampling methods.
Key terms
What is a population and what is a sample?
The population is the whole group you want information about. A sample is a smaller part of that group which is actually observed and used to draw conclusions.
e.g. All 1200 students in a school form the population, and 60 students who are questioned form a sample.
What is a random sample?
In a random sample every member of the population has an equal chance of being chosen, and the choice is made by chance rather than by the person collecting the data.
e.g. Numbering 500 customers and choosing 25 numbers with a random number generator gives a random sample of customers.
What is sampling bias?
Bias occurs when the way a sample is chosen makes it unrepresentative, so some members of the population are more likely to be included than others. Conclusions from a biased sample cannot safely be applied to the whole population.
e.g. Asking only people at a bus stop about public transport leaves out people who drive, so the results may be biased.
What makes a data source reliable?
A reliable source collects its data by a clear method from a representative sample, is up to date, and has no reason to favour a particular result. Poor method, out-of-date figures or a source with a stake in the answer all reduce reliability.
e.g. A national statistics office survey of a randomly chosen sample is usually more reliable than an online poll that anyone can vote in.
Worked example, step by step
The question Calculator needed for the quartiles [6 marks]
A café owner wants to know how long customers wait for their drinks. At one Monday lunchtime she times the first 12 customers to order. The waiting times, in minutes, are
3, 4, 4, 5, 5, 6, 6, 7, 7, 8, 9, 21.
(a) Give one reason why this sample may give a biased estimate of the usual waiting time.
(b) Use the interquartile range rule to show that the waiting time of 21 minutes is an outlier.
Worked solution: worked example
Answer: (a) the sample is a convenience sample from one lunchtime, so it may not represent the usual waiting time (b) , , upper boundary ; since , the time of 21 minutes is an outlier.
Step 1 - Find the weakness in the sample.
The first 12 customers on one lunchtime were chosen because they were easy to reach, not by chance. Lunchtime on a Monday may be busier, or quieter, than the rest of the week, so these waiting times may not represent the usual waiting time. R1 - A reason that says why a sample from one lunchtime may not represent the usual waiting time
Why? Explaining bias means saying which group is left out or over-represented, and why it would differ from the whole population.
Step 2 - Find the quartiles with the calculator.
Enter the 12 times in a list and run the one-variable statistics. The calculator gives and . A1 - A1 -
Why? The syllabus says to use technology for quartiles. Several methods for quartiles exist, so the calculator and a hand method can differ, but for this data set they agree.
Step 3 - Find the upper boundary.
The interquartile range is . The upper boundary for outliers is M1 - Use with their interquartile range A1 - Upper boundary
Why? An outlier is a value more than from the nearest quartile. For a large value, the test is against the upper boundary.
Step 4 - Compare and conclude.
The waiting time 21 is greater than 12, so it lies more than above the upper quartile. It is an outlier. (The lower boundary is , so no time is an outlier at the low end.) R1 - Compare 21 with 12 and conclude that 21 is an outlier
Why? "Show that" needs the comparison written out: the value, the boundary and the conclusion. Stating the boundary alone does not show anything.
Step 5 - Decide what to do with it.
An outlier is not automatically a mistake. If the owner finds that the 21 minutes was a genuine long wait, it is a valid part of the sample and stays. If it was recorded wrongly, for example an order entered twice, it should be corrected or removed.
Why? The syllabus asks you to be aware that in context some outliers are valid and some are errors. The rule only flags a value for a closer look.
R1 A reason that says why a sample from one lunchtime may not represent the usual waiting time
A1
A1
M1 Use with their interquartile range
A1 Upper boundary
R1 Compare 21 with 12 and conclude that 21 is an outlier
Now you try: enter your answers to be marked
Worked solution: practice question 1
State the type of bias that may arise.
Worked solution: practice question 2
State the likely effect on the sample mean as an estimate of the population mean.
Worked solution: practice question 3
Your score so far: 0 out of 4 marks
Calculator steps: One-variable statistics (mean, median, standard deviation)
Common mistakes
- Treating every outlier as an error. The rule flags a value as unusual, not as wrong - some outliers are a genuine, valid part of the data and should stay in unless there's a clear reason to exclude them.
- Confusing discrete with continuous. Data you count - number of pets, goals scored - is discrete. Only data that comes from measuring belongs on a continuous scale.
- Blurring "small" with "biased". A small sample gives an imprecise estimate; a biased sample gives a wrong one regardless of size. Be clear about which criticism actually applies to a scenario.
Using your GDC
The exact key presses for the TI-84 Plus CE, Casio fx-CG50, Casio fx-CG100 and TI-Nspire CX, with what the screen shows. Pick your calculator.
One-variable statistics (mean, median, standard deviation)
Used in: Data types and collection
How to find summary statistics on a TI-84 Plus CE
1-Var Stats (from a list) | |
Press stat and choose 1:Edit. If L1 already has numbers in it, press ▲ to put the cursor on the name L1, then press clear and enter to empty the list. |
|
Type the data into L1, one value at a time, pressing enter after each: 4, 5, 5, 6, 7, 8, 9, 10. |
L1 4 5 5 6 7 8 9 10 |
Press stat, then ▶ to move to CALC, and choose 1:1-Var Stats. With the Stat Wizards on, set Xlist to L1 (2nd 1), leave FreqList empty, move down to Calculate and press enter. If the Stat Wizards are off, 1-Var Stats is pasted on the home screen: press 2nd 1 and then enter. |
1-Var Stats Xlist:L1 FreqList: Calculate |
The results appear. Press ▼ to scroll down to minX, Q1, Med, Q3 and maxX. x̄ is the mean, σx is the standard deviation to use, and Sx is not needed. Mean and standard deviation , each to 3 significant figures. |
x̄=6.75 Σx=54 Σx²=396 Sx=2.121320 σx=1.984313 n=8 minX=4 Q1=5 Med=6.5 Q3=8.5 maxX=10 |
Data with frequencies | |
For a frequency table, type the values in L1 and the frequencies in L2. In 1-Var Stats set Xlist to L1 and FreqList to L2 (2nd 2). For the table in the box below, L1 is 1, 2, 3, 4, 5 and L2 is 3, 5, 8, 4, 2. |
1-Var Stats Xlist:L1 FreqList:L2 Calculate |
Use σx for the standard deviation and ignore Sx. The calculator's Q1 and Q3 are the medians of the lower and upper halves of the data.
The full guide: One-variable statistics (mean, median, standard deviation)
See the full list of GDC guides for every calculator task.
Syllabus content covered
| Code | Syllabus content |
|---|---|
| SL4.1 | Concepts of population, sample, random sample, discrete and continuous data. Reliability of data sources and bias in sampling. Interpretation of outliers, defined as a data item more than 1.5 the interquartile range (IQR) from the nearest quartile. Sampling techniques and their effectiveness: simple random, convenience, systematic, quota and stratified sampling methods. |
Official IB source: Mathematics: applications and interpretation subject brief (PDF, ibo.org).
Quick questions
What is the formula for a systematic sampling interval?
. Pick a random start within the first interval, then take every th member after that.
What counts as an outlier in IB AI SL statistics?
A data item more than the interquartile range (IQR) from the nearest quartile - so below or above .
How do you find a stratified sample size for each group?
stratum size - the same sampling fraction applied to every group so each is represented proportionally.
What is the difference between discrete and continuous data?
Discrete data can only take separate, countable values (e.g. number of goals scored); continuous data can take any value in a range and usually comes from measuring (e.g. height, time).
How is systematic sampling different from simple random sampling?
In systematic sampling only the starting point is chosen at random, and the rest of the sample follows at a fixed spacing down the list. In simple random sampling every member is chosen by chance.
Does a small sample always mean the results are biased?
No. Bias comes from how the members are chosen, not from how many there are. A small sample may be less reliable because it varies more, but a large sample chosen badly can still be biased.
What if a stratified sample size is not a whole number?
Round each value to the nearest whole number, then check that the group sizes add up to the total sample size. If they do not, adjust the largest group by one.
How do I explain bias in an exam answer?
Name the group that is left out or over-represented, then say why that group might answer differently from the whole population. A statement such as "people without internet access cannot take part, so their views are missing" explains the bias clearly.
Why is quota sampling not a random method?
The interviewer chooses who to ask until each quota is filled, so not every member of the population has an equal chance of being picked. Only the group sizes are controlled.
Is the time taken to run a race discrete or continuous?
It is continuous, because time is measured and can take any value in a range, such as seconds. Data from counting, such as the number of runners, are discrete.
Ready to practise properly?
Practice questions on every section, marked instantly like the real exam.