With the Texas Instruments TI-86, you can calculate the most commonly-used descriptive statistics for one variable (mean, standard deviation, sum of values, sum of the values-squared) and for regression/correlation between two variables (regression coefficient and intercept, correlation coefficient). There is an upgrade (set of programs) avaialable from TI which is supposed to give the TI-86 the capabilities of the TI-83 for statistics. I have not yet had the opportunity to try it out.
Data are entered and stored in a list
(which allows checking that you have entered correctly & keeps
the numbers for recalculation)
For all statistical
functions: Press 2nd
then the + key [STAT] (Referred to as "2nd +[STAT]"
in notes below) to see the statistics menu (above the function
keys)
There are three situations in which you
enter data - one variable (actual data), one variable with frequencies
(grouped or ungrouped), two variable. In the TI-86, you always
have to indicate frequencies, even if they are all 1's.
Always clear old data before entering new data, or your
results will be messed up:
A. One variable
Press 2nd + [STAT], press F2[EDIT] This brings up the list editing screen
Values of the variable go in the first column (xStat): enter the number (you see it at the bottom of the screen), press down-arrow, enter the next, etc.
To change a value already typed in: Use the up & down arrows to highlight it, enter the new number & press the down-arrow or ENTER to put the new numbber in.You must also enter a list of values in the list "fStat" - these will all be 1's. The list must be the same length as your list of values.
Press EXIT to return to the home screen.
B. One-variable with varying frequencies
Press 2nd + [STAT], press F2[EDIT] This brings up the list editing screen
Values of the variable go in the first column (xStat): enter the number (you see it at the bottom of the screen), press down-arrow, enter the next, etc.
To change a value already typed in: Use the up & down arrows to highlight it, enter the new number & press the down-arrow or ENTER to put the new numbber in.The frequencies (number of times repeated) for the values are entered in the list fStat - there must be a frequency for each value.
Press EXIT to return to the home screen.
C. Two-variable
Press 2nd + [STAT], press F2[EDIT] This brings up the list editing screen.
Values go in pairs : x values in first column (headed xStat) y-values in second column (yStat) (easiest to enter all x-values down the xStat column, then y-values down the yStat column, next to corresponding x-values)You must also enter a list of values in the list "fStat" - these will all be 1's. the list must be the same length as your lists of values.
You cannot calculate with the data table showing - you must return to the home screen (by pressing "Exit" if you are looking at the data.
A. One-variable (either case)
[From home screen]
Press 2nd+ [STAT], press F1[CALC] press F1[OneVa] press ENTER
You see a screen with mean( x-bar), sum of x's sum of x-squares ( not "sum of squares" SS(x)) sample standard deviation (Sx) population standard deviation [not much use] and sample size (n) - scrolling down will show minimum, maximum, 1st & 3d quartiles and median [5-number description]B. Two-variable
Regression/correlation
[From home screen]
Press 2nd+ [STAT], Press F1 [CALC] Press F3[LinR] (note not F4 which is LnR) , press ENTER
You see screen giving a (intercept), b (coefficient) and correlation coefficient (also shows form of equation: y = ax + b, so you can see which number is which) scrolling down shows n (number of pairs).
Note:to save the regression equation for graphing, you add to the regression command. After LinR (before pressing ENTER) type y1 (using the 2nd ALPHA [alpha] 0[y] to get lower-case y) & then press ENTER - this defines y1 (on the graphing screen) to be the regression equation. You can, of course, use any other function name in place of y1.Combined statistics
Press 2nd+ [STAT], press F1 [CALC], press F2[TwoVa], press ENTER
You to see a screen giving mean of data in x-list (x-bar), sum of x-list data ( sum of x's) sum of x-squared ( not "sum of squares" SS(x)) sample standard deviation for x-list (Sx), then same data for y-list (called y on screen) and the sum of the xy's .
The TI-86 will produce boxplots, modified box plots, or histograms from one-variable data, and scatter plots from 2-variable data. Up to three plots can be defined at one time.
To define a plot
Press 2nd+ [STAT] press F3 [PLOT] , press F1, F2 or F3 [PLOT1, PLOT2 or PLOT3]
Highlight "On" to turn the plot on or "Off" to turn th eplot off and press ENTER (This determines whether this plot will apear when you give the "Graph" command)
Use the cursor key to move down to "Type" and select the type by pressing the appropriate function key. The one variable types are "Box" for basic boxplot, "MBox" for modified boxplot (dots for outliers) "Hist" for histogram. The two variable types are "Scat" for scatterplot (the more useful) and "xyline" for a scatterplot with the points connected in the order they appear in the list - not generally useful unless the x-varaible represents time or order and the list is in order.
Cursor down. the options will depend on the type of plot. Generally "XList Name" will be xStat, "YList Name" will be yStat, "Freq" will be 1 (unless you have a frequency list where the entries are not 1's - then "Freq" will be "fStat" "Mark" indicates the symbol to be used to show points on the graph (for scatterplot)
Press EXIT press EXIT (to return to the home screen)To see your plot, press GRAPH, press F5[GRAPH]
You can get two side-by-side boxplots by the following trick:
Enter the values for one set of data in xStat, the values for the other in yStat
Turn on Plot1, select MBox and leave evrything as it comes (XList is xStat, Frequency 1)
Turn on Plot 2, select MBox, for "XList name type "yStat" (using the upper & lower case Alphabetic key), leave frequency as 1
EXIT, EXIT to home screen and press GRAPH F5[Graph] and the two boxplots will appear together.If you define (and turn on) a scatterplot and calculate the regression line with the same data (using the command that saves the line to y1) the line will appear on the plot when you give the "graph" command.
Errors or questions? Please let me know: cpeltier@saintmarys.edu
Last update 8/24/2000