AlphaStat is a program that performs a variety of statistical functions on a user-created database. It is designed to perform descriptive, bivariate, and multivariate analysis on databases both small and large. While I have tried to make the package useful for anyone who might need to add a little number crunching power to their Amiga, it was really put together with social science research in mind. AlphaStat will probably be most useful to students, social workers, and others who have a need for computer analysis of large bodies of data. And while AlphaStat doesn't have the features of the higher end statistical packages (conical rotations, advanced factor analysis, etc.) it can also serve as a limited aid to more serious researchers.
The end user assumes all risks inherent in the use of this program. The author cannot be held responsible for any of the consequences resulting from the failure of this software to perform correctly or for any other consequence that may directly or indirectly cause damage (eg- a bad grade, professional embarrassment, lost data, world war, the election of Lyndon LaRouche to public office, the return of the Bee Gees, Rocky V, etc.)
HOW TO REGISTER YOUR PROGRAM...
This program is Shareware. This means that you get it for free to try out. If you don't like it, just file it away or delete it. If you like it and you use it, you should send $10 to the author to register your program. When and if enough interest (read: "registered users") develops, I have every intention of expanding this program considerably. If you have a favorite statistical test or function and would like to be able to run it on AlphaStat, please do not hesitate to send me the formula (along with all the necessary explainations of what the variables are and how they are derived) with your registration. There is a good chance I will be able to include it in future upgrades.
Send your $10 to--
Matt Ritchie
ACU Box 6825
Abilene, TX 79699
If you also include your name, address, and/or Plink and Compuserve handles, I will notify you of the next upgrade. It would also help me if you would tell me how you plan to use AlphaStat. Future upgrades may not be made available to the general public (but only through mail to registered users.)
All users should feel free to report AlphaStat bugs by contacting me through modem or mail. I would also appreciate any CONSTRUCTIVE criticism anyone might have to offer.
AN OVERVIEW
AlphaStat projects are done in three stages. First, the user creates an ASCII text file known as a "source" file. You can create an ASCII file using "ed" or your favorite word processor. This file consists of a list of all the variables and indexes (or indices for the syntactically pure) to be in his database. It also contains a list of the contents of each record in the database. Second, the user compiles this source file. If everything goes well-- two additional files will be created by AlphaStat. Now the user can move to the third stage-- analyzing the data.
AlphaStat is menu/mouse driven. If you are unfamiliar with the use of these devices, please see your INTRODUCTION TO THE AMIGA. Most of the keyboard input to AlphaStat is accomplished through input statements (as opposed to string gadgets)-- if you ever make a mistake in an input statement, simply press RETURN at the next prompt; AlphaStat will automatically abort the procedure.
One additional caveat- AlphaStat was designed to be as flexible as possible. This flexibility is provided at a price, however. If you ask it to perform some sort of mathematical function (such as division by zero) that is statistically impossible or awkward on your database, you may find it spits out garbage or returns an error. As a user, you have the responsibility of understanding what you're asking it to do.
ABOUT- Select this item to get information on the current version of AlphaStat.
QUIT- Select this item to exit AlphaStat and return to the CLI or Workbench.
COMPILE- This command will present you with a requester asking which file it is that you wish to compile. Select the string gadget in the top of the window and type the name of the directory containing the source file. Every source file should have the appendix ".src" attached to the end of its name.
The "credibility.src" file is a good example of what a source file should look like. If you get a chance, load it into your favorite word processor or "ed" and take a look at it.
Each source file should be a plain, ASCII text file which conforms to the following format---
Line 1: Number of Variables. This is the number of variables each record will contain. "Variable" is used here in the sense of research variables. Each record will have a single value to represent each variable. Comparisons in AlphaStat are made between variables. Thus if one variable was IQ-- each record (in this case, a record representing a single person) would have a value representing a person's IQ. Some variables may have to be codified to work in AlphaStat. For example, if you wish to compare the IQ of men vs. women you might create a variable for sex and codify all records of men as 1 and all records of women as 2.
Line 2: Number of Indices. An index is either a sum or an average of a series of variables. These are extremely useful if you need to sum up a large numbers of variables into one value. If you do not need indices, simply put a zero on line 2.
Line 3: Number of Records. The total N size of your database.
Line 4: Variable Names. Each name should be a single word no longer than ten letters which describes the variable to which it corresponds. Separate each name with a single comma.
Line 5: Variable Types. Each variable should have a corresponding "A" or "N" associated with it. "A" represents a variable containing alphabetical characters. AlphaStat is incapable of analyzing alphabetical characters, but you may include an alphabetical field if you need room for comments or identification codes. "N" stands for "numerical." You should use this as the code for all the variables you wish for AlphaStat to process. As in line 4, separate each variable type designation with a comma.
Line 6: Variable Comments. This can be any note -- up to 20 characters long expressing a comment on a particular variable. It might be an explaination of the significance of each value assigned to the variable, or it might be a brief explaination of the variable itself.
Line 7: Index Names. (If you have no indices, you should omit lines 7-9.) These follow the same format as the variable names.
Line 8: Index Types. All index types should read "M" or "T". Use M to average the scores of the variables in the index. Use T to total the scores of the variables in the index.
Line 9: Index Comments. This is identical to the variable comments line, except it contains comments on the indexes.
Line 10: Index Contents. Each index may account for up to 20 variables. This line must contain twenty numbers-- each separated by a comma-- for each index. The values in each of these twenty numbers contain information on which variables to include in the index. To include a variable, simply place its "number" in one of the twenty "slots". A variable's number is the place in which it is listed on the Variable name list. The first name listed is "1", the second is "2", etc. In the "credibility.src" file-- note that variables 3,4,and 5 are used. These numbers correspond to the variable names Competence, Character, and Charisma. If you do not fill all the slots for an index, fill the unused slots with zeros. For example, both of the variable contents listings in "credibility.dat" contain a 3, a 4, a 5, and 17 zeros. Note that the index contents information should ALL BE ON THE SAME LINE. Thus the length of an index contents line should always be 20 x the number of variables.
Lines 11 +: Records. These lines contain your records. Each line corresponds with exactly one record. Enter each variable for each record in correspondence with the order in which the variables were listed in the line for variable names. Separate each value with a comma. Alphabetical values should not exceed 20 characters and should not contain commas themselves. DO NOT enter index values. AlphaStat will calculate the index values for each record and include them in the database.
After you enter the last record, it is a good idea to go back and "double-check" your work. Do the number of items in lines 3-6 correspond with the number of variables in your file? Are the lines for indices consistent as well? Does the index contents line have items equal to exactly twenty times the number of indices? Does each record line have the correct number of variables? Do the number of record lines correspond with the number given in line 3? This stage represents the hardest work you will have to do with AlphaStat. I am hoping that future versions will include a built-in data editor that will make it easier to create files. If you have any questions about how a .src file should be put together, look at the example included. If that is insufficient, contact me by mail or modem.
Once you have selected the file that you wish to compile, simply click the "compile" button and AlphaStat will create the necessary files for processing the data. The first file will contain the same name as the source file, but will end with an ".inf". This file contains information on each variable's name and the numbers of records and indices. The other file will have the suffix ".dat." This file is the database that has been created from your source file. If you ever need to change the contents of either of these files, you will need to recompile the source file after correcting the necessary items.
OPEN- This item will bring up a requester similar to the requester for compiling files. This time, however, you should select the ".dat" file which you wish to load into AlphaStat. Once the file has been loaded, you may use any of AlphaStat's statistical functions on any of the numerical variables in the file.
VIEW VARIABLES- Select this for a list of all the variables that are in the file that is currently in AlphaStat's memory. You are given an option for outputting to either the screen or printer. Select the "button" beside the option you wish to choose and click OK. AlphaStat will provide you with the name, type, and comment for each variable in the file.
VIEW CONTENTS- This menu option prints out a list of the contents of each record in the file. As with all AlphaStat functions, you are first given the option for outputting to printer or screen.
EDIT FILE- if you have a workbench disk in df0 with a copy of ed, this option will load the specified file into ed. If you name a file or path name that is invalid, ed will load an empty document. This option is extremely useful for editing and re-editing source files. Once you quit ed, control will be returned to AlphaStat. Strike another one up for Amiga multitasking!
FREQUENCY- This option gives you the chance to group your data into varying class intervals. The first input statement asks you for the variable on which you wish to perform the frequency tabulation. Enter a name of a valid, numerical variable. The second statement asks you for the number of class intervals. The final two prompts ask for a LOWER and UPPER limit for the class intervals. These specify the lowest boundary for the lowest class interval and the highest boundary for the highest class interval. AlphaStat automatically calculates the size for each class interval.
(NOTE: While AlphaStat does not directly support discrete data frequency descriptions, it is possible to do the equivalent by entering the following values: the desired number of class intervals, the high value PLUS ONE, and the low value. This will tabulate discrete data with the LOW number in each interval representing the correct discrete value.)
CENTRAL TENDENCY- This function prints a table which gives the mean, median, standard deviation, and standard error for each variable in the file. Very handy!
CHI SQUARE- Enter the variable on which you wish to perform a chi-square. Then you will be asked for all the values (within the variable) you wish to tabulate. Each value desired should be separated by a comma. Make certain that each of these values is included in a record at least once or AlphaStat may return a "division by zero" error. AlphaStat will print a table containing all the critical information it used to calculate the chi-square value. Currently, AlphaStat does not support multiple chi-squares. (NOTE: Yes, I know that this should not technically be grouped under descriptive data analysis, but it seemed the best place to fit it since it went with measurements of central tendency and frequency tabulations.)
(Please note that significance (p) level calculations are not yet implemented on any of AlphaStat's functions. I hope to add them at a later date. Until then, you will have to look these values up in your friendly, neighborhood Stats text.)
T-TESTS- This item brings up an input requester asking you to enter the dependent variable for your t-test. This can be any valid, numeric variable. The next item asks you to enter the independent variable. The third and fourth items ask you to input two values for the independent variable. Each of these values will be used to define the two groups that are being compared. AlphaStat performs a simple student's t on these groups and prints out all relevant information. Note: if the independent variable contains records with values other than those provided in the third and fourth input statements, not all of the records will be included in the t-test.
Z-MEAN- This item works identically to the t-test but instead prints out a z value for the two groups in question.
CORRELATION- (Pearson's r) Simply input the two variables you wish to correlate. AlphaStat will print out a table describing the mean, sum of squares, sum of dd and r value.
REGRESSION- Selecting this item brings up a request for an X and Y variable. It will print out a formula that can be used to calculate the approximate value for any score in a variable Y, given a score for X.
ANOVA- This item is very similar to the t-test. It asks for a dependent and an independent variable. But instead of asking for two values, the third input prompt asks for a LIST of all the values you wish to test in the independent variable. Input as many values as you like, each separated by a comma. AlphaStat will calculate an F value and print a table with the relevant information.
FACTOR- AlphaStat is capable of performing one, simple factor analysis known as "elementary linkage." When this item is selected, AlphaStat generates a table which contains the correlations of each of the variables with each of the other variables. When you are asked for input, enter the name of each of the variables you wish to factor (up to 15). After a few seconds (or minutes, if you have a large database), AlphaStat will print out a correlation matrix of all the variables listed in the input statement. It is up to you to interpret the groupings of each factor.
Statistics are confusing things, filled with all kinds of complicated terminology and jargon. If you need help using AlphaStat, I am willing to provide technical support provided several things--
1. You are a registered user.
2. You exhaust other sources of information first. Please consult a stats textbook for the information you need before contacting me. I am willing to give you help with the PROGRAM, but I cannot afford to be turned into a stats tutor.
3. You contact me by mail or modem. Please, no phone calls.