F F ä DATA ANALYSIS WITH YOUR ATARI ¢¢Determine if variations in experimental data occurred by chance alone by using t-test analysis for dependent samples. This BASIC program works on all 8-bit Atari computers with 48K, Atari compatible printer and disk drive.¢¢By Michael Dalto¢¢Data can be analyzed from any experiment based on two sets of observations that are paired either because subjects are matched or because the same subject is measured twice. In this type of experimental design the paired measurements from control and test groups of the same experiment differ only with respect to the parameter being examined. Any variation due to individual differences cancel making the results more meaningful.¢¢ BACKGROUND ¢This program determines the validity of the null hypothesis which states that there is no difference between the mean values from two sets of paired measurements. The decision concerning the acceptance or rejection of the null hypothesis is made by comparing a test magnitude calculated from the experimental data with a test distribution. The test distribution employed in this program is a two-tailed t-distribution. Since t-distributions are used when the sample size is 30 or less up to 30 paired measurements may be analyzed by this program. The value of the t-distribution which is to be compared against a calculated test value is affected by the confidence level which the operator selects. This is the probability of rejecting the null hypothesis when it is true. For scientific purposes a probability of error of 5% or less should be used.¢¢ GETTING STARTED ¢When the program is run it initially presents the operator with A probability selection menu. The probability of error chosen will be used by the program until reset. The operator is then presented with a main menu from which several options may be selected. The "ANALYSIS OF DATA" option will alternately enable or disable the printing of intermediate values used in the calculation of the test value. The "END PROGRAM" option will terminate the program. The "PROBABILITY RESET" option returns the program to the probability selection menu. The "RECALL DATA" and "STORE DATA" options are similar in their operation. When either option is selected a file directory is displayed and a line from the program will be listed followed by a "READY" prompt. At this point the operator has been returned to Atari basic. The listed program line may now be edited to the file name being recalled or stored. After editing press the return key and type "CONT" to proceed with the program. The contents of a recalled file will be printed before returning the operator to the main menu. Once a file has been recalled the program will be reset to whatever probability of error was stored in the file. The probability reset option should be employed to change this value if needed. After storing a file the program will return to the main menu. The "RETURN TO CONTINUE" option will allow the operator to input the next data set.¢¢ INPUT OF DATA ¢The program allows for the entering of a report title. If a previous data set was analyzed the title displayed may be edited. The number of paired measurements to be analyzed must be entered before the the input of data. For each paired measurement a number is assigned. After entering data the program allows for the correction of any erroneous entry. The paired measurement number is used to refer to the particular measurement to be corrected. If a non-numeric character is inputted the data entry correction routine is aborted and the operator is returned to the main menu.¢¢ REPORT GENERATION ¢The program generates a report which can be easily pasted into a laboratory note book. The first section of the report consists of a title and a list of the paired measurements entered into the program. If the data analysis option was selected the next section of the report will consist of all the intermediate values calculated for the generation of the test value. The last section of the report consists of a statement indicating the null hypothesis tested. This is followed by a comparison of the generated test value and the table value with a concluding statement indicating if a difference exists between the pairs of measurements. If the observed variations in data values occurred by chance no difference exists between measurements.¢
