Scribba
Statistics surprises people who did well in math. The arithmetic is usually easy, and software or a calculator does most of it anyway. What's hard is choosing the right method, checking that it applies, and saying what the result means in plain words. A correct number with the wrong interpretation may earn only partial credit.
That changes what good notes look like. Stats notes need fewer long calculations and more careful wording, clear notation, small sketches of distributions, and a way to decide which procedure fits a problem.
Keep the notation straight from day one
Statistics uses similar symbols for different things, and mixing them up costs points. The biggest one is the difference between a population and a sample:
- Population parameters are usually Greek letters: μ for the mean, σ for the standard deviation. A proportion is the usual exception: p.
- Sample statistics are usually Roman letters: x̄ for the mean, s for the standard deviation, and p̂ for a proportion.
Start a notation page in the first week and add every new symbol with what it means and an example. Write it once, carefully, and refer back to it. In Scribba, give that page a named bookmark in the page sidebar so it's one tap away from any lecture.
The marks that matter are small: the bar over x, the hat over p, a subscript 0 on a null value. Scribba's Zoom Window lets you write big in a magnified strip, and your writing lands small and neat on the page, so x̄ doesn't turn into a smudged x.
Sketch the distribution every time
Probability questions are much easier with a picture. Before you look anything up in a table or type it into a calculator, draw the curve:
- Sketch the shape, usually a bell curve for normal and t distributions, or a skewed curve for chi-square.
- Mark the mean (the center, on a symmetric curve) and the value you're asking about.
- Shade the area you want: left tail, right tail, between two values, or both tails.
- Estimate before you calculate. If you shaded less than half the curve, your probability should be less than 0.5.
That last step catches a lot of errors, like reading the wrong side of a table. For a two-tailed test, shading both tails reminds you to split or double the probability.
In Scribba, hold the Apple Pencil still at the end of a stroke to snap a straight line, which keeps the horizontal axis and the vertical line at a cutoff value neat. For histograms, box plots and scatterplots you draw by hand, graph paper keeps bars and scales even. See graph paper on iPad.
Write every hypothesis test in the same layout
Hypothesis tests come up all semester: one-sample t, two-sample t, proportions, chi-square, regression slopes. The details change, but the structure doesn't. Use the same layout for every one:
- State the hypotheses in symbols and in words. H₀ is the "no effect" claim; Hₐ is what you're looking for evidence of.
- Name the test and the significance level.
- Check the conditions: random sampling or assignment, independence, and whatever your course requires for the sampling distribution to be approximately normal.
- Compute the test statistic and the p-value.
- Make a decision: reject H₀ or fail to reject H₀. Never "accept" it.
- Conclude in context, in a full sentence about the real situation in the problem.
Step 3 and step 6 are where most points are lost. Students skip the conditions and write a conclusion that says "reject the null" without saying what that means for the drug, the survey or the factory.
Do the same with confidence intervals: state the parameter, check conditions, compute, then interpret in context.
Learn the wording, not just the idea
Stats exams often grade the exact phrasing of interpretations. Some to write down carefully and memorize:
- A confidence interval: we are 95% confident that the true population mean lies between the two values. Not "there's a 95% chance the mean is in this interval."
- A p-value: the probability of getting results at least as extreme as the ones observed, assuming the null hypothesis is true.
- A slope in regression: for each one-unit increase in x, the predicted y changes by the slope, in the units of the problem.
- Correlation vs. causation: an observational study can show association, but only a randomized experiment supports a cause-and-effect conclusion.
Keep these on their own page as a template, with blanks where the context goes. Then practice filling them in for real problems.
These definitions are perfect for flashcards. In Scribba, a Vocabulary page has TERM and DEFINITION / EXAMPLE columns. Write each term and its exact wording in a row, and Scribba reads your handwriting and turns each row into a flashcard. Review and fix any misread words first. This doesn't need Apple Intelligence. See flashcards from handwritten notes.
Build a decision page for choosing a test
The hardest exam question in stats is often "which procedure?" Build one page that sorts procedures by a few questions: Is the variable categorical or quantitative? One sample, two independent samples, or paired data? Are you estimating or testing? Each answer leads to a method.
Make this page yourself as the course goes, instead of copying one from the textbook. Building it is how you learn it. If your exam allows a formula sheet, this page is the backbone of it. More in how to make an exam cheat sheet.
Annotate software output
Many stats courses use software like R, Excel or a graphing calculator, and exams may show printed output you have to read. When your professor shares output as a PDF, import it into Scribba and write on it: circle the test statistic, label the p-value, and write what each number means in the margin. Keep this notebook in a tab beside your lecture notes; Scribba keeps up to six notebooks open as tabs.
Practice with recall
Drag study tape over the conclusion of a worked example and write your own interpretation before you peek. Do the same on your decision page: cover the method and name it from the questions alone. See study tape on iPad.
Frequently asked questions
Can Scribba calculate statistics or run tests?
No. Scribba doesn't recognize equations or do calculations. Use your calculator or software for the numbers and Scribba for notes, sketches and studying.
What paper is best for statistics notes?
Ruled or dot grid paper for lecture notes, graph paper for drawing histograms and scatterplots by hand, and problem set paper for homework.
How do I remember which test to use?
Build a decision page that sorts tests by data type and design, then practice with mixed problems where you have to choose the method before solving.