MDM4U Mathematics of Data Management: the course map (24 ideas in the order they are taught), the three ideas everything leans on, and how to study it

MDM4U (Mathematics of Data Management) is the Ontario Grade 12 university-preparation course that teaches how to count outcomes, model chance, collect data honestly and read what it says: permutations and combinations, probability and its distributions, sampling and study design, and the statistics of centre, spread, correlation and the margin of error. Our map has 24 ideas in teaching order; three of them, combinations, the binomial distribution and standard deviation, carry most of what follows. Here is the map and how to study it.

What MDM4U is, and who takes it

The code reads as every Ontario code does: MDM is Mathematics of Data Management, 4 is Grade 12, U is university preparation. The Ministry's course description sets the prerequisite as the Grade 11 functions course, either MCR3U (university preparation) or MCF3M (university/college preparation), and that text is Crown copyright, so this page uses our own idea titles and our own one-line descriptions of the strands.

Who takes it: students heading for business, economics, psychology and the other social sciences, life sciences, nursing and the humanities, where a first-year statistics or research-methods course is waiting. Many take it as the "third math" beside MHF4U and MCV4U, often expecting a course about spreadsheets, and are surprised by the first strand, which is counting. It is one of the six Grade 12 courses this blog maps.

Most schools end the course with a culminating investigation, which the Ministry names as part of it: you pose a question, gather or find data, analyse it with the course's tools and present what you found. It is the course's own study method if you let it be, and the week plan below treats it that way.

For a reader in the United States, the nearest course is AP Statistics, which shares the probability, distribution, sampling and estimation ideas, goes further into inference, and has no counting strand of this size.

The course map: 24 ideas in the order they are taught

This is the pack's teaching order, grouped by strand with our one-line description of each strand. Some schools start with the data strands and end with counting; the method on this page works in either order.

Counting and probability (ideas 1 to 6): how many ways, and how likely.

1. The fundamental counting principle
2. Permutations, with and without repeated items
3. Combinations and choosing committees
4. Theoretical probability of equally likely outcomes
5. Conditional probability and two-way tables
6. Probability using counting techniques

Probability distributions (ideas 7 to 14): the shapes chance takes.

7. Discrete random variables and expected value
8. The discrete uniform distribution
9. The binomial distribution
10. The geometric distribution
11. The hypergeometric distribution
12. The normal distribution and areas under the curve
13. z-scores and standardizing
14. The normal approximation to the binomial

Collecting and organizing data (ideas 15 to 18): where numbers come from.

15. Types of data and variables
16. Sampling methods
17. Experiments and observational studies
18. Frequency tables, histograms and box plots

Statistical analysis (ideas 19 to 24): what the numbers say.

19. Measures of central tendency
20. Range, variance and standard deviation
21. Quartiles and the interquartile range
22. Scatter plots and the correlation coefficient
23. The least-squares line of best fit
24. Estimating a proportion and the margin of error

Chapter 3 of the free playbook is the method for turning this list into the checklist you keep for the semester: one line per idea, marked each time you test yourself on it.

What each strand unlocks

The order is a chain inside each strand, and knowing the chain tells you where a gap will show up later.

In the counting strand, the counting principle (1) gives permutations (2), which give combinations (3) once you divide out the orderings; counting probability (6) is combinations over combinations; and the coefficients in the binomial (9) and hypergeometric (11) distributions are combinations again. A student who cannot decide whether order matters is stuck from idea 3 onward.

In the distributions strand, expected value (7) is the idea every distribution's mean rests on; the binomial (9) is the centre, with the geometric (10) and hypergeometric (11) as its siblings and the normal approximation (14) as its limit. The normal distribution (12) and z-scores (13) are the bridge to the analysis strand.

In the data strand, the types of data (15) decide which sampling method (16) and which display (18) make sense, and study design (17) decides what you are allowed to conclude. This strand is where the course stops being arithmetic and starts being judgement.

In the analysis strand, centre (19) and spread (20) give z-scores a second time, correlation (22) and the regression slope (23) both divide by standard deviations, and the margin of error (24) rests on the binomial from idea 9 and the normal from idea 12. The last idea in the course uses the first half of it.

The three ideas everything leans on

By the map's dependencies, not by any statistic, three ideas carry the most weight. Each comes with the one thing students get wrong and a two-line check.

Combinations (idea 3). Counting probability (6), the binomial coefficient (9) and the hypergeometric (11) are all built from C(n, r). The wrong line: using a permutation when order does not matter, which overcounts by r!. The check: choosing 2 people from 3 by hand gives 3 pairs, and C(3, 2) = 3 while 3 × 2 = 6 counts each pair twice.

The binomial distribution (idea 9). The geometric and hypergeometric are its siblings, the normal approximation is its limit, and the margin of error for a proportion is its spread in disguise. The wrong line: forgetting the coefficient, writing pᵏ(1 − p)ⁿ⁻ᵏ for the probability of k successes without the C(n, k) that counts the orders. The check: for two tosses, one head, the formula without the coefficient gives 1/4, and listing HT and TH gives 2/4.

Standard deviation (idea 20). z-scores, the normal table, the correlation coefficient and the regression slope all divide by it. The wrong line: mixing the population formula, which divides by n, with the sample formula, which divides by n − 1, and reporting one without saying which. The check: a data set of two values, 2 and 4, has a population standard deviation of 1 and a sample standard deviation of √2, and the question's wording says which one it wants. The Feynman walk-through is the way to explain to yourself why the sample version divides by n − 1 until it stops being a rule to memorise.

Worked example: one problem from each strand, with the common wrong line

Each answer is re-derived here, so you can check your own against it line by line.

Counting. How many arrangements are there of the letters of MATH, and how many committees of 2 can be chosen from 6 people?

• Arrangements: four distinct letters, so 4! = 4 × 3 × 2 × 1 = 24.
• Committees: C(6, 2) = (6 × 5) ÷ 2 = 15.

The wrong line is 6 × 5 = 30 for the committees, which counts "Ana then Ben" and "Ben then Ana" as two committees. A committee has no order, so divide by 2.

Distributions. The probability of exactly 2 heads in 4 fair tosses, and the expected value of a game that pays 10 dollars with probability 1/6 and costs 1 dollar otherwise.

• Two heads: C(4, 2) × (1/2)² × (1/2)² = 6 × 1/16 = 6/16 = 0.375.
• Expected value: E = 10 × (1/6) + (−1) × (5/6) = 10/6 − 5/6 = 5/6 ≈ 0.83, so about 83 cents a game, in your favour.

The wrong lines: 2/4 for the heads, which is not a probability of a count at all; and 10/6 ≈ 1.67 for the game, forgetting that losing costs a dollar.

Data. A survey of the first 50 people leaving a gym at 6 pm, asked about their exercise habits. What kind of sample is this, and what is the fix?

It is a convenience sample: whoever was easy to reach. It is also biased by place and time, since people leaving a gym at 6 pm exercise more than the population you probably care about. The fix is a sampling frame, a list of the population you want to describe, and a random draw from it. No arithmetic here, and this is what the data strand tests: the reasoning, written in a sentence.

Analysis. The data set 4, 8, 6, 2.

• Mean: (4 + 8 + 6 + 2) ÷ 4 = 20 ÷ 4 = 5.
• Deviations from the mean: −1, 3, 1, −3.
• Squared deviations: 1, 9, 1, 9; their sum is 20.
• Population standard deviation: √(20 ÷ 4) = √5 ≈ 2.24.
• Sample standard deviation: √(20 ÷ 3) ≈ 2.58.

Both are right; the question's wording decides which. "These four values are the whole set" asks for the population version; "a sample of four from a larger group" asks for the sample version, and that is the version most statistics software gives by default. The wrong line is giving one without saying which, and the mark scheme usually asks for the word.

And a z-score: a mark of 70 in a class with mean 60 and standard deviation 8 has z = (70 − 60) ÷ 8 = 10 ÷ 8 = 1.25. The wrong line is (60 − 70) ÷ 8 = −1.25, subtracting the other way; a mark above the mean has a positive z.

Six answers to check: 24; 15; 0.375; 0.83; 2.24 and 2.58; 1.25.

How to study MDM4U

The strands are different kinds of knowledge, so the method changes by strand.

Counting and distributions are procedures. Read one worked example slowly, cover it, redo it from the question alone, then do a sibling with different numbers, and write every wrong line in an error log with the right line beside it. Sweller and Cooper (1985) found that students who studied worked examples before attempting problems learned algebra procedures faster and made fewer errors than students who only solved problems. How to study math has the full five-step cycle.

The data strand is vocabulary and judgement. Make cards that ask "what kind of sample is this, and what is wrong with it", with a one-line scenario on the front and the kind and the fault on the back. Say the answer before you turn the card. Self-testing is the method with the strongest evidence, and it fits this strand exactly.

Formulas as derivations you can rebuild. The binomial mean np is expected value applied to n coin flips; the variance of a sample divides by n − 1 for a reason you can say in two sentences. A formula you can rebuild under pressure is one you cannot misremember.

Mixed sets across strands from mid-semester. The exam will not say "this is a hypergeometric". Rohrer and Taylor (2007) found that students who practised shuffled problem types did better on the test a week later than students who practised them in blocks, because the shuffled students had practised deciding which method to use. Mixed sets across strands shows how to shuffle an exercise set by hand.

The three explainers beside the analysis strand. The statistics by hand are on mean, median and standard deviation by hand; the normal table and z-scores on the normal distribution and z-scores; and what a margin of error and a confidence interval mean on confidence intervals and the margin of error.

A week-by-week plan for a semester

An 18-week semester in the pack's order. Each week has one recall task (closed book, five minutes, before anything else) and one mixed set (problems from this week and every earlier week, shuffled). Schools order the strands differently: rearrange the rows, keep the method. The culminating project runs beside the weeks from week 10.

Weeks	Ideas	The recall task	The mixed set
1–2	1–3: the counting principle, permutations, combinations	Decide order-or-not for five one-line scenarios, cold	Counting problems, permutations and combinations shuffled
3–4	4–6: probability, conditional probability, probability by counting	Build a two-way table from a sentence	Probability problems mixed with weeks 1–2
5–6	7–9: expected value, the uniform, the binomial	Write the binomial formula and say what each factor counts	Expected value and binomial problems, shuffled
7–9	10–14: geometric, hypergeometric, the normal, z-scores, the normal approximation	Name which distribution five scenarios are, cold	Distribution problems that do not name the distribution
10–12	15–18: data types, sampling, study design, displays; the project's question and data	"What kind of sample, and what is wrong with it" for three scenarios	Judgement questions mixed with distribution problems
13–16	19–24: centre, spread, quartiles, correlation, regression, the margin of error; the project's analysis	Standard deviation of four values by hand, both conventions	Analysis problems mixed with weeks 7–9
17	Mixed review by strand; the project's final draft	Blank-page recall of each strand's ideas	Sets that mix all four strands
18	Exam week	The three load-bearing ideas, once more	The one-week exam plan

Exam review in one week: seven days out, sort the 24 ideas with one closed-book attempt each into Can, Shaky and Cannot. The Cannot pile first by the worked-example cycle, the Shaky pile by recall and one problem each, a timed practice run two days out, a light review the night before, and sleep. The three load-bearing ideas go into the week whether or not they sorted as Can; they turn up inside other strands' questions.

What carries into first year

Statistics courses start where idea 12 leaves off: the normal distribution, z-scores and estimation become sampling distributions, confidence intervals and hypothesis tests, and a student who owns ideas 12 to 14 and 19 to 24 recognises the first month. Research-methods courses in psychology and the social sciences are ideas 15 to 17 at length: sampling, study design and what you may conclude. Business and economics courses use ideas 19 to 24 as everyday tools.

The one jump is the tool. In Grade 12 the arithmetic is by hand and by calculator; in first year it is software. The ideas are identical; the new thing is reading an output table, and a student who computed a standard deviation by hand in week 13 knows what every number in that table is. For the AP route, the comparison with AP Statistics is in the FAQ below.

MDM4U on one screen

• Four content strands: counting and probability; probability distributions; collecting and organizing data; statistical analysis; plus the culminating investigation.
• Three load-bearing ideas: combinations; the binomial distribution; standard deviation.
• The method: procedures by the worked-example cycle with an error log; the data strand by judgement cards; formulas by derivation; mixed sets across strands from mid-semester.
• The plan: every idea on at least two separate days; the 18-week table above; the one-week triage before the exam; the project as the course's own study method.
• What it leads to: first-year statistics, research methods, business and economics, which use the normal distribution, sampling and the measures of centre and spread from the first week.