PQRAOV carries out 1-, 2- and 3-way analyses of variance. The input file may be a DAS file (created by CONCAT) or other ASCII file of appropriate format (see 5.2 below).
The input file must consist of a succession of 1 or more data matrices, each matrix preceded by a title line. Within each matrix, each row (line of the file) contains one observation for each condition, and each column contains all the observations for one condition.
The order of the conditions (columns) across the file corresponds to a hierarchical nesting of the factors. For instance, suppose factor A has 2 levels, factor B has 2 levels and factor C has 3 levels. Then the order of the conditions would be: A1B1C1, A1B1C2, A1B1C3, A1B2C1, A1B2C2, A1B2C3, A2B1C1, A2B1C2, A2B1C3, A2B2C1, A2B2C2, A2B2C3. That is, all levels of the most nested factor (C) are exhausted before the next factor (B) is varied, and all levels of factors B and C are exhausted before A is varied. The maximum number of columns (conditions) in a matrix is 50.
If the input file contains more than one data matrix, each preceded by its own title line, PQRAOV will do a separate analysis of variance on each matrix.
Change the default drive (and directory, if necessary) to the drive (and directory) containing your input file. If you request an output file, it will be created on the same drive (and in the same directory). Run the program by typing
at the terminal in response to the prompt. The program responds by typing its name and version number and the current date.
You will then be asked for the name of your input file. The default input filename extension is DAS.
The program will then ask you for the name of an output file. If you do not want an output file, just press Return. If you enter a filename, all output from the program to the terminal will also be sent to that file. The default output filename extension is AOV. If the output filename you enter is the same as the name of an existing file, you will be asked
OVERWRITE EXISTING FILE? (N/Y):
If you reply Y (yes), the existing file will be deleted and replaced by the output file from PQRAOV; otherwise, you will be asked for a new name for the output file. If you enter PRN: in response to the output filename prompt, all output from the program to the terminal will also be sent to the printer.
The program will now print the title of the current input data matrix and prompt you to specify the format of a line of data. Every line of an input matrix must have the same format - that is, corresponding numeric fields must be the same width and be separated by the same number of characters. You should enter a FORTRAN-type format, enclosed in parentheses and containing only F and X field specifiers. F is used to specify the width of a numeric field and X is used to specify how many characters to skip before the next numeric field. For instance, the format
tells the program that on each line of the current matrix it should skip the first 5 characters, then read 2 numeric fields each 5 characters wide, then skip 5 more characters and read 2 more 5 character numeric fields.
If you just press Return in response to the FORMAT prompt, the default is to use the same format that was used for the previous data matrix. If this is the first data matrix in this file, the default format is (50F5.0) - i.e. every field is 5 characters wide and there is no skipping between fields. This is the format in which data is output by CONCAT, so if your file was creatyed by CONCAT and you want to include every condition in the analysis you should use the default format.
The program will now prompt you to enter the number of lines in the current matrix. The form of the prompt will vary, depending on the title of the matrix. If your input file was created by CONCAT, you would be asked for the number of items per condition or the number of subjects per condition, depending on the kind of data in the matrix (determined from the title). For any other input file you will be asked for the number of cases. Whatever the form of the prompt, your reply is the same: the number of rows in the data matrix being analysed. There is no default.
You will now be prompted for the number of levels of factors A, B and C in that order. A refers to the least nested factor and C to the most nested factor - see 5.2 above. If you are dong a 2-way analysis, enter 1 level for any 1 of the factors. If you are doing a 1-way analysis, enter 1 level for any 2 of the factors. There is no default.
The program assumes that all factors are repeated except those thet you specify here. Enter A and/or B and/or C or just press Return. Non-repeated factors may be entered in any order and in upper or lower case, but must not be separated by spaces or commas.
PQRAOV prints out the mean of each condition (column) in the input matrix, then collapses each factor and pair of factors in turn and prints out means for the collapsed designs. Means are output in the order of columns in the input.
For each factor or interaction of factors and for each error term, the program prints out the source of variance (SV), the sum of squares (SS) and the mean square (MS). For each factor or interaction, the program also prints the Fisher's F value and the corresponding probability. Probabilities are correct to 3 significant figures. When the probability is followed by an E and a negative number, this indicates that the probability is to be multiplied by 10 raised to the power of that negative number.
The program then prints a row of asterisks and proceeds to analysis of the next input matrix, if any, or terminates.