UPDATE transfers raw subject data from the temporary DTP file (created on floppy disk by the DMASTR "Unload" command) to the user's permanent DAT file. It then calculates the item and subject means and errors from the data for each incorporated subject and stores them in the DAT file (see section 4 - structure of Data Files - fotr details of how the means are stored). UPDATE can also be used to print reports of subject statistics, item means and subject means.
UPDATE cannot both analyse new subject data and print out subject statistics in the same run. To do both, you must run the program twice.
If you are adding new subject data to your DAT file, you must first create your DTP file using the DMASTR Unload ("U") command. If your DAT file is on a floppy disk, unload your DTP onto that same disk. If your DAT file is on the hard disk, unload onto any scratch disk, preferably the one labelled "PC1 UNLOAD DISK".
Place the disk containing your DTP and/or DAT file in drive A. Even if you have no DTP file and your DAT file is on the hard disk, you must still place a disk in drive A, because the program checks there to see whether your files are resent. Change the default drive to C and the default directory to your own directory (this is where your DAT file should be if it is on the hard disk; if you do not already have a directory, ask a more experienced user to create one for you). Then run the program by typing
in response to the prompt. The program will respond by typing its name, version number and the current date.
You will then be asked to enter the name of your DAT file. The default filename extension is DAT.
(for the case where no DTP file is present, see section 8.5)
If a DTP file with the filename you specified is present on the disk in drive A, UPDATE will append that data to your DAT file, and delete the temporary DTP file. It also closes and re-opens the DAT file, in order to save it so that data will not be lost in the event of the program aborting for any reason. The program then prints out the number of subjects already incorporated in the DAT file and the number of subjects whose data was copied from the DTP file. UPDATE then commences the analysis at the first unanalysed block of subject data in the file. (This will normally be the first block copied from the DTP file, unless there was unanalysed data left over after a previous run of UPDATE for some reason.)
The program prints out for the subject being analysed the subject number and the overall percentage of incorrect responses for that subject. This percentage is based on all non-zero RTs in the subject's data, not just the items in onditions.
You will then be asked whether you want this subject's data incorporated into the subject and item means. If you answer "N" (no), that block of data is flagged as having been rejected, and analysis proceeds to the next unanalysed block of subject data. If you answer "Y" (yes) and the program finds no exceptional conditions in the data (see 8.4 below), the program will print out statistics and condition means and errors for that subject, store the subject means, flag the data as having been included in means, and analysis moves on to the next unanalysed subject's data.
When all new subject data has been analysed and incorporated or rejected, the program calculates and stores item means over all incorporated subjects. It then copies the DAT file with its new subject data and means to a new file with the filename extension DBK. This is your data backup file. You will need it if some terrible accident befalls your DAT file. It will be created on the current default dirve and in the current default directory; this is why you should be in your own directory when you run UPDATE. If a DBK file already exists for this DAT file, it will be deleted when the new one is created. The program then moves on to reporting of item and subject means (see 8.6 below).
After you request that a block of subject data be incorporated, UPDATE prints out warning or error messages if any of the conditions listed below is found. If any of messages 8.4.3, 8.4.4 or 8.4.8 is generated, the program gives you another chance to reject that block of subject data, if you should so desire.
8.4.1 DATA BLOCK ALL ZEROES
A non-DTP file has been copied into the DAT file by accident, or no items were presented to the subject for some reason. Remove this block from the DAT file (using the Transfer command of FILE) and run UPDATE again.
8.4.2 *** SUBJECT/CONDITION OVERFLOW ***
The space reserved for subject means is limited. Once the number of subjects incorporated multiplied by the number of conditions in the experiment reaches 512, no more subject means can be stored, so no more subjects can be incorporated. The remaining subject data is in the DAT file but it has not been analysed.
8.4.3 *ABSOLUTE CUTOFF* - ITEM nnn ZEROED
The reaction time for this item was less than the absolute lower cutoff for this file (specified when running DATMAK). This item is excluded from subject and item means, but the original reaction time is stored unchanged in the data block. Thus it is possible, if you so desire, to run DATMAK to change the absolute lower cutoff so that this item will be included.
8.4.4 ## WARNING ## DATA MISSING ? ZERO RT ITEM nnn ?
A reaction time of zero was found for an item in a condition. For some reason, DMASTR did not collect a reaction time for this item.
8.4.5 * WARNING LIMIT EXCEEDED *
More than 10 zero RTs were found. Only the first 10 have been listed.
8.4.6 DATA UNANALYSABLE - LESS THAN 2 CORRECT RESPONSES FOUND
A standard deviation cannot be calculated over less than 2 reaction times. The subject has been automatically rejected by UPDATE.
8.4.7 CONDITION nnn - NO CORRECT RESPONSES
The subject mean for this condition will be set to zero.
8.4.8 CONDITION nnn ERROR RATE EXCEEDS 50%
If no DTP file exists UPDATE will print out the comments stored in Block 1 of the DAT file and the number of subjects incorporated into the means.
This situation would normally arise when the DAT file has been set up for re- analysis by DATMAK. If no subjects have been incorporated into the DAT file means, the program will offer you the choice of four analysis options. They are:
N - NORMAL (CORRECT RT)
E - ERRORS ONLY
A - ALL RESPONSES
M - MEAN SUBSTITUTION
The default option is "N". In this analysis, only correct responses will be included in means. Under option "E", correct responses are treated as incorrect and incorrect responses as correct; thus only incorrect responses are included in means. Under option "A", all responses are treated as correct and included in means. Under option "M", all incorrect responses are replaced by the overall mean for correct responses between the absolute lower and upper cutoffs.
When a file is set up for reanalysis using a non-default option, it is recommended that some filename extension other than DAT be used to make this clear at some future date.
Note that mean substitution and the normal analysis treat errors due to very long RTs (stored in the DAT file as -4000) as errors, whereas the error and all response analyses treat these as equipment failures.
When an option has been chosen, the program will then ask:
IS FULL SUBJECT PRINTOUT REQUIRED? (Y/N):
If you answer yes, analysis will proceed exactly as for the case when a DTP
file is present (see 8.3 and 8.4 above) except that no DBK file is created. If you answer no, all subjects except those previously rejected will be automatically incorporated without asking the user, and no warning messages will be printed. For this reason, subject printout should only be omitted for reanalysis of data that has already been checked for problems of this kind. In either case, the program will then calculate and store item means and then proceed to the reporting of item and subject means (see 8.6 below).
The program will then ask whether you want subject statistics. If you answer "N" (no), the program will search the DAT file for unanalysed data. If any is found it will be analysed just as in 8.3 and 8.4 above, except that no DBK file is created. It will then calculate and store item means and then proceed straight to the reporting of item and subject means (see 8.6 below). If you answer "Y" (yes), the program will list for each subject (including rejected subjects) the subject number, rejection flag, percentage of incorrect responses, mean of all correct RTs between lower and upper absolute cutoffs in all conditions, standard deviation of same, ratio of s.d. to mean, number of times the standard deviation cutoffs were applied in calculating the item and subject means, number of zero RTs and the number of RTs less than the absolute lower cutoff.
If, while reporting subject statistics, UPDATE encounters unanalysed subject data, it will print a message and abort. You must re-run UPDATE and do not request subject statistics, so that the program can analyse this data. It cannot do both stats and analysis in the same run.
After subject analysis has been completed the program asks whether the user requires an item summary (report of item means). If so, a summary of mean RT and number of errors for each item in all conditions is printed at the terminal. The program then asks whether the user requires a subject summary (report of subject means by condition). If so, a summary of mean RT and number of errors for each condition for each subject is given.
To get a printout of all the reports sent to the terminal by UPDATE, just type control-P before running the program and again following program termination (assuming a printer is connected).
If you are connected to a network printer, this option will usually not work. In this case, you can use a batch file to create a disk file output, which can then be printed. An example is the file "upd.bat":
echo %1 > $cmd
echo y >> $cmd
echo y >> $cmd
echo y >> $cmd
update < $cmd > %1.upd
In this batch program, %1 represents the name of the DAT file to be updated. The program constructs a temporary file called "cmd" which contains three "Yes" commands. This file is then input to UPDATE, which generates a disk file with the same filename as the DAT file, but with .upd as the extension. The final command uses the program V which views the output file. Any program would do.
All recently created DAT files (late 1986 onwards) have the scaling factor for the item means set to a value of 1. If you update an old DAT file where the scaling factor has some other value, UPDATE will automatically change
the scaling factor to 1 and recalculate all the item means. You will not be able to request subject stats on that run. The next time you run UPDATE on that file it will behave just like any other file.
Normal exit occurs after the subject summary if it is requested or after the question if it is not.
It is not, in general, a good practice to terminate programs with control-C unless you know just what they are doing at the time. If you respond to a prompt from UPDATE with a control-C, you should be safe enough because the program will be doing nothing while it waits for your reply. If, on the other hand, you type control-C while the program is busily pursuing its own mysterious purposes, you may stop it halfway through creating the DBK file or deleting the DTP file, and you'll probably end up in a mess.