You may download the calculator instruction book by searching for "TI84 guidebook", for instance. In May 2015, an English version was available.

If your calculator is older, borrow a cable and update it. If your calculator is not a "Plus Silver" either borrow on or learn the keystrokes. Some keystrokes below are given for the older calculators.

Newer calculators have a "Stat Wizard" that gives a nice interface to input values. Make sure that the Stat Wizard is ON, along with other newer features. MODE/(scroll to 2nd window)
MATHPRINT (not CLASSIC)
STAT DIAGNOTICS ON
STAT WIZARDS: ON

Restore a list that no longer shows up on the Stat menu...
Stat/Edit
move the cursor onto the name, such as L3, where you wish to insert a list.
2nd-INS for insert; now you should see "Name=" at the bottom of the screen.
from here, 2nd-L1 works, or you can select a list name through 2nd-LIST/NAMES/ and scroll to the name.

Enter Data
STAT/EDIT
For our course in general, use L1 for the x-values. If there are y-values, use L2.
I suggest you clear specified lists for each new problem: from the STAT LIST editor, cursor up onto a list name, and then press CLEAR (not DEL) and ENTER.
Then type the data.

Perform a statistical analysis on one variable
Enter the data into L1 (or another list).
STAT/CALC/1-Var Stats to paste this command into the home screen.
For each STAT CALC menu item, if neither List nor FreqList is specified, then the default list names are L1 and L2. If you do not specify freqlist, then the default is 1 occurrence of each list element.
(older: STAT/CALC/1-Var Stats to paste this command into the home screen. Then 2nd-L1 ,(type a comma) 2nd-L2 Enter)
You can re-access these values by repeating the steps above or with VARS/5:Statistics and selecting the values you wish.
Remember that Sx is the sample standard deviation; this is what we will almost always use. On occasion we might need sx, which is the population standard deviation. Remember that we use population statistics when we are certain that we have absolutely all of the data. 2, which is the sum of the squares of the data.
For IQR, you will calculate Q3-Q1 (the calculator doesn't show it.)

Access values of calculated statistics (like the mean), so that the exact value can be used,
enter data and calculate the appropriate statistics. (If you edit a list or change the type of analysis, all stat variables are cleared and will need to be re-calculated.)
VARS/5:Statistics, and select the appropriate menu.
For instance, here are some common statistics and the menu in which each is found.

XY	Σ	PTS
number of data points mean standard deviation min max	sum of x's sum of squared x's	1st quartile median 3rd quartile

Statistical plot types
in order of appearance in the set-up:

scatter	xyLine (not used in Intro to Statistics)	histogram
"modified" box plot (the one we will use)	(boxplot) (not used in Intro to Statistics)	normal probability plot

Create a histogram
Enter the data into L1 (or another list).
Deselect all graphs and statistical plots. Make sure the graph mode is set to function.
Define the plot.
    2ND-[STAT PLOT] (above y=).
    cursor up to Plot1 (or whichever)     ENTER.
    cursor to "On"     ENTER
    cursor to "Type ="     cursor to the rightmost on the top row.     ENTER.
    cursor to "Data List:"     2ND-LIST     select list     ENTER
ZOOM/ZOOMSTAT after graphing.

Other one-variable plots are similar.
boxplot: use the one which shows possible outliers.
normal probability plot (last plot selection): select Y for the DataAxis setting.

Normal Distribution Areas
This actually worked better on the old calculators. Graphing does not always work.
De-select all plots, including StatPlots.
Define the window settings for graphing. Remember that 4 standard deviations on either side of the mean shows practically all of a normal distribution; the y-values are between 0 and 1.
2nd-[DRAW]     DRAW     1:ClrDraw     ENTER. this clears any previous drawings. You MUST do this between each drawing (or you'll get wacky results.) Because these are "drawings" and not "graphs", the Graph button is useless; there's nothing in the y= menu or defined in a StatPlot to graph.
2nd-[DIST]     DRAW     1:ShadeNorm(     give the values for the lower value and upper value, then close with ) if this is for a standard normal. If, however, it is not a standard normal, then include the mean first, then the standard deviation; then close with ). Separate these values with commas.

Example: We want Prob{-0.5 < Z < 0.5} for a standard normal. Good window settings are Xmin=-4, Xmax=4, Xscl=1, Ymin=-0.2, Ymax=0.8, Yscl=0.2,Xres=1. 2nd-[DIST]     DRAW     1:ShadeNorm( -.5,.5)

Example: We want Prob{55 < X < 62} for a normal distribution with mean 60 and standard deviation 7. 4 std devs from the mean yields 60 - 4*7 = 32, 60 + 4*7 = 88. Good window settings are Xmin=30, Xmax=90, Xscl=10, Ymin=-0.1, Ymax=0.1, Yscl=0.2,Xres=1. 2nd-[DIST]     DRAW     1:ShadeNorm(55,62,60,7)

TIP: If you have to fiddle with the window settings, 2nd-QUIT from the drawing, change the window settings, then 2nd-ENTRY     ENTER.

Creating a scatterplot without regression line

Enter the data into two lists.

Deselect all graphs and statistical plots. Make sure the graph mode is set to function.

Either do the following prior to graphing or use ZOOM/ZOOMSTAT after graphing. ZOOM/ZOOMSTAT adjusts the window settings to include all values.
Perform a "two variable statistics" to obtain the settings for the graphing window. Use the WINDOW button to change these settings.
    Use the min and the max of the x-data to set xmin and xmax on the graphing window.
    Use the min and the max of the y-data to set ymin and ymax on the graphing window.
Xlist and Ylist must be the same length.

Define the plot.
    2ND-[STAT PLOT] (above y=).
    cursor up to Plot1 (or whichever)     ENTER.
    cursor to "Type ="     (use the first one)
    cursor to "Xlist:"     2ND-LIST     select list     ENTER
    cursor to "Ylist:"     2ND-LIST     select list     ENTER
    GRAPH    

optional resizing: ZOOM     9:ZOOM/ZOOMSTAT; ENTER    

Displaying the correlation coefficient
Do this only once. It will remain set unless you change batteries or Reset the calculator.
    MODE/(scroll to 2nd window)     STAT DIAGNOTICS ON
    Alternatively, 2nd-CATALOG     press "D"     down arrow to "DiagnosticOn"     ENTER     ENTER
Now, when you perform a regression, the correlation coefficient will also be shown.

Linear Regression
Enter the data into two lists.
Clear Y1 if you wish to save the regression equation.
    Y=     CLEAR
Perform the regression
    STAT     cursor to CALC     "4:LinReg(ax+b)" or "8:LinReg(a+bx)" (do not choose "9:LnReg")
    2ND-LIST     select list     ENTER     , (comma)     select list    
    optional: to store equation
        , (comma)     VARS     cursor to Y-VARS     "1:Function"     cursor to the function desired     ENTER
    ENTER
Remember to turn off the plot of the regression line in the Y= menu prior to showing any other statistical plots.

Creating a scatterplot with regression line
make the scatterplot in PLOT1 as above. If regression seems appropriate, continue.
Perform the regression, but augment the commands stated above for a regression as follows.
    STAT     cursor to CALC     "8:LinReg(a+bx)"
    2ND-LIST     select list     ENTER     , (comma)     select list    
        , (comma)     VARS     cursor to Y-VARS     "1:Function"     ENTER to select Y1
    ENTER
Now GRAPH.

Find a confidence interval for 1-sample or 2-sample mean(s)
For means, either we have raw data or we have summary statistics. If we have data, enter it into a list or 2 lists, as is appropriate.
    STAT     TESTS     cursor down to the appropriate line; use 8 for a single mean, 0 for independent means.     ENTER.
(7 and 9 assume you know σ(s). You could choose that and enter the sample std dev value(s) for &sigma to approximate what we would do by hand if we use a z-distribution for large samples.)
If you have data for 1 mean or 2 means, choose "Data", select the list(s). If you have summary statistics, choose "Stats" and enter the values.
Enter the confidence level. For 2 means, choose "No" for Pooled (for our text.) "Calculate" and ENTER.
You will see the confidence interval. For a t-test, you will see df.

Perform a hypothesis test for 1-sample or 2-sample mean(s)
Either we have raw data or we have summary statistics. If we have data, enter it into a list or 2 lists, as is appropriate.
    STAT     TESTS     cursor down to the appropriate line; use 2 for a single mean, 4 for independent means with a t-distribution.     ENTER.
(1 and 3 assume you know σ(s). You could choose that and enter the sample std dev value(s) for &sigma to approximate what we would do by hand if we use a z-distribution for large samples.)
If you have data for 1 mean or 2 means, choose "Data", select the list(s). If you have summary statistics, choose "Stats" and enter the values.
Enter the hypothesized value. Select the appropriate alternative hypothesis. For 2 means, choose "No" for Pooled (for our text.) "Calculate" and ENTER.
You will see the value of the test statistic and the p-value. For a t-test, you will see df.
If you wish to see the shaded area beyond the test statistic value for the alternative hypothesis that you chose, STAT/TESTS/ etc. as before but select DRAW instead of calculate. This also shows the test statistic value and a rounded p-value.

Find a confidence interval for 1-sample or 2-sample proportion(s)
For means, either we have raw data or we have summary statistics. If we have data, enter it into a list or 2 lists, as is appropriate.
    STAT     TESTS     cursor down to the appropriate line; use 8 for a single mean, 0 for independent means; Cursor down to A for a single proportion, B for two proportions.     ENTER.
The "x" value(s) are the number of "successes". For instance, if our sample indicates that 25 out of 200 surveyed love chocolate, then the sample proportion, p-hat, is .00125, but "x" is 25. If you are given n and p-hat, then x is the product of n and p-hat.
Enter the confidence level. "Calculate" and ENTER.
You will see the confidence interval.

Perform a hypothesis test for 1-sample or 2-sample proportion(s)
Either we have raw data or we have summary statistics. If we have data, enter it into a list or 2 lists, as is appropriate.
    STAT     TESTS     cursor down to the appropriate line; use 5 for a single proportion, 6 for two proportions.     ENTER.
Enter the hypothesized value for a single proportion. Select the appropriate alternative hypothesis.
The "x" value(s) are the number of "successes". For instance, if our sample indicates that 25 out of 200 surveyed love chocolate, then the sample proportion, p-hat, is .00125, but "x" is 25. If you are given n and p-hat, then x is the product of n and p-hat.
"Calculate" and ENTER.
You will see the value of the test statistic and the p-value.
If you wish to see the shaded area beyond the test statistic value for the alternative hypothesis that you chose, STAT/TESTS/ etc. as before but select DRAW instead of calculate. This also shows the test statistic value and a rounded p-value.

Entering data into a matrix
We do this before performing a chi-square test.
    2nd-MATRIX     EDIT     choose the matrix name, such as [A].     ENTER    
type the number of rows;     ENTER     type the number of columns     ENTER. enter all of the values in the contingency table.

Perform a chi-square test
Enter the counts (absolute frequencies) of the table into a matrix, as above; it is easiest to use [A].
    STAT     cursor to TESTS     cursor down to "C: χ²-Test..."     ENTER.
(If you used matrix [A], just) cursor down to "Calculate" and ENTER. Expected values will be in Matrix [B].
See the p-value and the df.
If you wish to see the shaded area past the test statistic value, STAT/TESTS/"C: χ²-Test...", DRAW. This also shows the test statistic value and a rounded p-value.