The World of Computer Software/The World of Computer Software.iso/estat21.zip/DEMO

Text File | 1992-08-05 | 11KB | 482 lines

; Demo file for EASISTAT
; All lines beginning with a ';' will be ignored

; An 'e' at the start of a line is short for 'echo' - the line will be
; copied to the output but there will be no other effect

E                         EASISTAT - DEMONSTRATION
E
E Welcome to this demonstration of EASISTAT. EASISTAT is a cheap and
E cheerful statistics package, and this demonstration is intended to show
E you most of its features so you can see what is available.
E
E This demonstration will pause intermittently to allow you to read what
E is displayed on the screen. When it is paused you have the option of
E stopping the demonstration (by typing 'E'), taking temporary control (by
E typing 'K') or carrying on with the rest of the demonstration (by
E pressing any other key). If this is your first experience of EASISTAT
E then it is probably best just to go straight through the demonstration.
E If you do want to finish it then after you have typed 'E' you can enter
E 'Q' for quit and then confirm it by entering 'Y' for yes and you will
E leave the program.
E
PAUSE
E                          A WORD ABOUT SHAREWARE
E
E EASISTAT is distributed as a Shareware program. That means that you are
E legally allowed to copy it and give copies to other people who you think
E might like it - in fact you are encouraged to do so. However please note
E the following points:
PAUSE
E
E 1. The copyright of the program remains with ARC Scientific limited.
E
E 2. You must not charge more than a nominal fee for the program if you
E pass it on to someone else, nor may you distribute the disc for any
E promotional purposes or for your personal gain.
E
E 3. Please acknowledge the program in any publications of work for which
E it was used.
E
E 4. If you pass it on you must do so in unmodified form all the related
E files must all be passed on as well.
PAUSE
E
E Once you have evaluated the program we hope you will pay for it (current
E price 40 pounds) and if you do so you will receive an up-to-date
E version of the program together with a full manual, and you will be
E supporting the concept of shareware. For details please read the file
E READ.ME or contact:
E
E   ARC Scientific Limited,
E   257 Woodstock Road,
E   Oxford OX2 7AE.
E
E   Tel. [phone]
E
E
E
E ....Coming next..... The Demonstration!
PAUSE
E
E
E Right, that's that over with. Let's show you some of EASISTAT'S
E commands. The first one is one to show us what commands are available
E and is called HELP, or H for short (you can always use upper or lower
E case letters, but maybe upper case is clearer for you to see what is
E happening). Once this pause is finished this DEMO file will type:
E HELP
PAUSE



HELP
PAUSE
E
E
E So, that's how to remind ourselves of the commands available. Now let's
E read in some data. If we want see what files are available on disc we
E can use the SYSTEM command like this:
E SYSTEM DIR
PAUSE
SYSTEM DIR

E Since EASISTAT is being run from a command file, it does not pause to
E allow you to inspect the list of files. We already know we're going to
E use the file called EXAMPLE.DAT. The command is:
E DATA EXAMPLE.DAT
PAUSE



DATA EXAMPLE.DAT
E
E Let's see what we've read in. To do this we can use the LIST command,
E but we don't want to see all 100 lines of the data so we'll use the LIST
E IF option to look at just the first 5 lines, like this:
E LIST IF ROW<=5
PAUSE



LIST IF ROW<=5
E
E The data is just arranged in vertical columns under one line of titles.
E There are too many titles to see all of them across the screen, and we
E also need to know the numbers of the columns they refer to. To do that we
E use the TITLE command like this:
E TITLE
PAUSE



TITLE
E
E Now we can find out about one of the columns. Let's get the mean etc. of
E column 15, the score on the GHQ (a measure of psychological distress).
E We use the BASICS command like this:
E BASICS C15
PAUSE



BASICS C15
E
E Just like the LIST command we can use BASICS with IF to look at only
E some of our sample. How about the first 50 rows:
E BASICS GHQ IF ROW<=50
PAUSE



BASICS GHQ IF ROW<=50
E
E Or how about all the men (column 3 codes for sex):
E BASICS GHQ IF C3=1
PAUSE



BASICS GHQ IF C3=1
E
E Not all the commands can use IF, so we need another way to be able to
E look at only some of the data. To do this we use the NARROW command.
E Another way to look at only the men is as follows:
E NARROW SEX=1
E BASICS GHQ
PAUSE



NARROW SEX=1
BASICS GHQ
E
E In this case though, the data set will remain narrowed down permanently
E until we issue the WIDEN command. We'll do that now:
E WIDEN
PAUSE



WIDEN
E
E Now we'll do our first statistical test. Maybe we could see if the men
E have higher GHQ scores than the women. Let's see if there are more men
E scoring 12 or above than women using a chi-squared test on a contingency
E table. We'll want 2 rows and 2 columns, and we'll need to define
E conditions to be included in each row and each column. The whole thing
E will look like this:
E CHISQ
E 2
E 2
E GHQ<12
E GHQ>=12
E SEX=1
E SEX=2
PAUSE



E Here we go then...
CHISQ
2
2
GHQ<12
GHQ>=12
SEX=1
SEX=2
E
E Or we could use Wilcoxon's rank sum test to see if the scores are
E generally higher for men or for women. That would look like this:
E WILC C15
E SEX=1
E SEX=2
PAUSE



WILC C15
SEX=1
SEX=2
E
E If we think the data is reasonably distributed, we might prefer to use a
E T test to make the same comparison, like this:
E TTEST C15
E C3=1
E C3=2
PAUSE



TTEST C15
C3=1
C3=2
E
E We can do a paired T test on two different columns to see if one is in
E general higher than the other by entering:
E TTEST P C16 C17  (where the P stands for paired)
PAUSE



TTEST P C16 C17
E
E We can see how well correlated two columns are by using Kendall's rank
E correlation coefficient:
E KENDALL HDA HDD
PAUSE



KENDALL HDA HDD
E
E The equivalent test to be used with normally distributed variables will
E also give us the least squares linear regression of one variable on the
E other:
E REGRESS C15 C16
PAUSE



REGRESS C15 C16
E
E Or we can do multiple linear regression to determine how well one
E variable is predicted by a number of others:
E MULT C24
E 4
E C19
E C20
E C21
E C22
PAUSE



MULT C24
4
C19
C20
C21
C22
PAUSE



E
E If we can't measure a dependent variable directly, maybe we can deduce
E it's existence by doing principal component analysis:
E COMP
E 4
E C19
E C20
E C21
E C22
E 0.05
PAUSE



COMP
4
C19
C20
C21
C22
0.05
PAUSE



E
E What we've also done is to make new columns which are the scores on all
E the principal components which contribute more than 0.05 to the overall
E variance. We've already been shown how well they correlate with each of
E the variables from which they were formed, but now it is also possible
E to study them in other ways. The first one went into C25. Maybe we could
E see how well it correlates with C24:
E REGRESS C24 C25
PAUSE



REGRESS C24 C25
E
E We can make new columns anyhow we want. We'll make a new column 16 which
E will be 8 digits wide with 3 decimal places, and call it Frank:
E NEW C16
E FORMAT C16 8 3
E LABEL C16 Frank
E TITLE
PAUSE



NEW C16
FORMAT C16 8 3
LABEL C16 Frank
TITLE
E
E There it is.
PAUSE



E
E At the moment there's nothing in it:
E BASICS Frank
BASICS Frank



E
E Let's make it the difference between C15 and C17:
E DERIVE C16
E C15-C17
E BASICS C16
PAUSE



DERIVE C16
C15-C17
BASICS C16
E
E Or about the natural log of C15? That would give us an error if
E C15 is ever zero, so we'll make any values less than or equal to zero
E to be 0.0001 instead first:
E DERIVE C16
E LN(C15*(C15>0)+0.0001*(C15<=0))
PAUSE



DERIVE C16
LN(C15*(C15>0)+0.0001*(C15<=0))
E
E Looks complicated, but it shows the kind of thing that can be done. Of
E course we could have done it more simply using the narrow and widen
E commands:
E NARROW C15<=0
E DERIVE C16
E LN(0.0001)
E WIDEN
E NARROW C15>0
E DERIVE C16
E LN(C15)
E WIDEN
PAUSE



NARROW C15<=0
DERIVE C16
LN(0.0001)
WIDEN
NARROW C15>0
DERIVE C16
LN(C15)
WIDEN
E
E There are two special functions to make life simple. The first derives a
E column from another by subtracting the mean and dividing by the
E standard deviation:
E DERIVE C16 ZED C15
E BASICS C16
PAUSE



DERIVE C16 ZED C15
BASICS C16
E
E And the other takes the rank of the value in its column:
E DERIVE C16 RANK C15
E BASICS C16
PAUSE



DERIVE C16 RANK C15
BASICS C16
E
E We can just do sums:
E ARITH
E 5 +5*38
E 2.6/7+(3.5-.6)pow(0.5)
E
PAUSE



ARITH
5 +5*38
2.6/7+(3.5-.6)pow(0.5)

; The above blank line takes us out of the ARITHMETIC command

E
E Or we can use the probability look-up functions, e.g. to find the
E significance of a Chi-squared of 13.6 with 2 degrees of freedom:
E ARITH
E 13.6PX2
E
PAUSE



ARITH
13.6PX2

PAUSE



E
E Now let's have a look at the special data editor supplied which is
E called up with the EDIT command. You can use the cursor keys to move
E around the data and alter it by typing in new values. New columns are
E added at the end by pushing Ctrl-RIGHT, and new rows are added with the
E ENTER key. When you have finished playing with the editor push the ESC
E key.
E EDIT
PAUSE



EDIT
E
E
E
E Well, that gives you a rough idea of some of what's available. You can
E write output to files or printer, and you can input commands from the
E keyboard or from files like this one. There's a one-way analysis of
E variance test that we haven't mentioned, and a few other odds and ends
E around the place.
E
E If you're going to use EASISTAT you are STRONGLY advised to print off
E this file (called DEMO) and study it carefully. If you don't have a
E manual then you should unpack the archive EASIDOC.EXE
E and study the documentation files it contains. How to do this is
E described in the file titled READ.ME. This is a fairly powerful
E package and you will probabably save yourself a lot of time and
E effort if you do this rather than diving straight in.
E
E Naturally we hope that you will decide to pay for the program and
E obtain a proper manual if you have not already done so.
PAUSE
E
E
E This demonstration of EASISTAT's features is now finished, and you
E will be left in control of the program. You may wish to run a second
E demonstration which shows how EASISTAT can work with EASIGRAF, a
E graph-drawing program. If you wish to do this and have received the
E EASIGRAF programs on a separate disk then you should insert this disk
E in the drive now and then enter:
E INPUT GDEMO
E
E Otherwise you can enter any other EASISTAT commands you like. When
E you wish to leave the program enter QUIT followed by YES.
E
E Good luck!     - ARC Scientific 1992