TI-89STATISTICS
Linear Regression, Normal CDF and z-Scores on the TI-89
If you need to know how to do linear regression on the TI-89 Titanium, or how to get a normal cdf value for a probability question, the commands exist but they are not on the keypad. This article shows where they live, how to type them, and what the results mean, from the regression line to the z-score.
Where statistics lives on the TI-89 Titanium
On the handheld, statistics runs in two places. The Stats/List Editor app, opened from APPS, holds the lists, the regression models and the distributions in its Calc and Distr menus, and asks for the arguments in dialog boxes. The other place is the home line: every command has a typed form that you can pick from CATALOG and finish with your own arguments. Calc89 works on the home line, so this article uses the typed form. It is shorter to write down, easy to check, and you can reuse it with the ENTRY history.
Lists use curly braces: 2nd ( gives { and 2nd ) gives }. Separate the values with commas.
Linear regression step by step
A linear regression needs two lists of the same length, the x values and the y values. Calc89 has both linear forms: LinRegMx fits y = m·x + b, and LinRegBx fits y = a + b·x. The model is the same line; only the names of the coefficients differ, so use the one your textbook uses.
- Press CATALOG and pick LinRegMx.
- Type the x list, a comma, then the y list.
- Add a colon and the result you want to see, for example
stat.RegEqnfor the equation. - Press ENTER.
The regression stores its results in variables that start with stat.. stat.RegEqn is the fitted equation, and the colon lets you run the regression and read one result on the same line:
| Input | Result |
|---|---|
| LinRegMx {1,2,3,4},{2,4,5,9}:stat.RegEqn | 2.2·x−0.5 |
| LinRegBx {1,2,3,4},{2,4,5,9}:stat.RegEqn | 2.2·x−0.5 |
| TwoVar {1,2,3,4},{2,4,5,9}:stat.r | 0.964763821238 |
Both forms return the same line for the same data, as the first two rows show. The third row answers a common follow-up question, how to find the correlation coefficient: TwoVar computes the two-variable statistics for the same lists, and stat.r reads r. A value close to 1 means the points lie close to a rising line.
On the handheld the regression dialog shows the coefficients and r² on one screen. If you want to graph the line, store the equation in a Y= function and graph it together with a scatter plot of the data.
Other regression models
The same pattern works for curved data. Pick the model in CATALOG, give the two lists and read the result. Calc89 has every regression from linear to logistic; four of them with checked examples:
| Input | Result |
|---|---|
| QuadReg {0,1,2,3},{1,2,5,10}:stat.RegEqn | 1.0·x2+1.0 |
| ExpReg {0,1,2,3},{3,6,12,24}:stat.RegEqn | 3.0·2.0x |
| LnReg {1,2,3,4},{0,0.69,1.1,1.39}:stat.b | 1.00270607399 |
| Logistic {0,1,2,3,4,5},{1,2,4,6,7,7.5}:stat.c | 7.78954768529 |
The last two rows read a single coefficient instead of the whole equation: stat.b for LnReg and stat.c, the upper limit, for Logistic. Results of regressions are decimals, because the fit is numeric.
Mean, median and standard deviation
For a single list, mean( and median( give one number each. For the standard deviation of a sample use stDevSamp(; OneVar computes all one-variable statistics at once, and stat.sx reads the sample standard deviation. Notice that the CAS keeps the mean and the standard deviation exact when the data are exact. Press ◆ ENTER for a decimal.
| Input | Result |
|---|---|
| mean({1,2,3,4}) | 5/2 |
| median({3,1,2}) | 2 |
| stDevSamp({1,2,3,4}) | √15/3 |
| OneVar {1,2,3,4}:stat.sx | 1.29099444874 |
Normal cdf: the area under the bell curve
The TI-89 normal cdf command is normCdf(. The syntax is normCdf(lower, upper, μ, σ). Leave out μ and σ and it uses the standard normal distribution with mean 0 and standard deviation 1. The result is the probability that a value falls between the two bounds.
| Input | Result |
|---|---|
| normCdf(-1,1) | 0.682689492137 |
| normPdf(0) | 0.398942280401 |
The first row is the familiar 68 percent of the empirical rule: about 0.6827 of the area lies within one standard deviation of the mean. For a left tail, use a very large negative number as the lower bound; for a right tail, a very large positive number as the upper bound. normPdf( is the height of the curve, not a probability, so you rarely need it for exam-style questions.
On the handheld the same calculation sits in the Stats/List Editor app under Distr, Normal Cdf, where a dialog asks for the lower value, the upper value, μ and σ.
z-scores and invNorm(
There is no zScore command on the TI-89 Titanium. A z-score is a one-line calculation: subtract the mean and divide by the standard deviation, z = (x − μ)/σ. Type it on the home line with your numbers in parentheses, for example (x − μ)/σ with the values of your problem, and press ◆ ENTER for a decimal. Once you have z, normCdf( with the bounds in z units gives the probability.
The reverse question, which z value cuts off a given area, is invNorm(. It takes the area to the left and returns the z value. With μ and σ as extra arguments it returns the value in the units of your data.
| Input | Result |
|---|---|
| invNorm(0.975) | 1.95996398454 |
invNorm(0.975) is the critical value z* for a 95 percent confidence interval, because 2.5 percent of the area lies to the right of it.
In Calc89
Calc89 uses the same CATALOG, the same command names and the same argument order, so the lines above work as written. The camera problem scanner reads a printed data table and puts it into the list editor, and the AI keystroke tutor answers questions like “how do I find r for these points” with numbered steps and tappable key chips for each key. Calc89 is free to download, with Pro features as an in-app purchase.
A note on exams, as of October 2026: phone and tablet apps are not permitted on standardized exams. The SAT (since August 2025) and the ACT prohibit CAS calculators, including the TI-89 Titanium. The AP Calculus and AP Physics exams permit the physical TI-89 Titanium. For AP Statistics, check the current calculator policy of the College Board.
Frequently asked questions
Is there a zScore command on the TI-89 Titanium?
No. Compute z = (x − μ)/σ on the home line. For the z value that cuts off a given area, use invNorm(.
How do I find the correlation coefficient r on the TI-89?
Run a two-variable statistic or a linear regression. In Calc89, TwoVar with :stat.r returns r directly.
Why does normCdf( give a different answer than my table?
Check the order of the arguments: lower bound, upper bound, then μ and σ. Without μ and σ the standard normal distribution is used.
Related pages
- /commands/linregmx/
- /commands/twovar/
- /commands/onevar/
- /commands/normcdf/
- /commands/invnorm/
- /guides/ti-89-keystrokes-basics/
More from the blog
- TI-89 Titanium vs TI-Nspire CAS: Which CAS Do You Need for Engineering?Keys, display, documents, CAS power and exam rules: how the TI-89 Titanium and the TI-Nspire CAS differ, and which one fits engineering.
- TI-89 vs TI-84 for Calculus: What a CAS AddsExact symbolic answers versus numeric ones, side by side: what the TI-89 Titanium adds over the TI-84 and when each one makes sense.
- Taylor Series, Sums and Sequences on the TI-89taylor( for Taylor polynomials, Σ( for sigma notation, sum( and seq( for lists, plus partial sums, products and infinite series on the TI-89 Titanium.