Revised Simulation Of Non Response

Simulation of Nonresponse In order to arrive with a data set that has nonresponse observations with the assumption that the nonresponse observations are missing at random (MAR), the generation of random numbers was introduced. These random numbers were then matched to each observation of the second visit data. The second visit observations were then sorted in ascending order by their assigned random number. In the program, the declaration of the matrices to be used was done. Files to be uploaded in the program were first converted to comma separated values (CSV) format. Two separate files were created for the nonresponse variables in saving the files with nonresponse observations. In each file, there are five columns that were made. The imputation class categories, set number also known as the trials, the actual data from the second visit nonresponse variable, the imputed data and the flagging of imputed values, 1 for the actual observation and 0 for the imputed one were in the first, second, third, fourth and fifth column respectively. Flagging of values are important to distinguish the imputed observations from the actual values. The process of simulating nonresponse begins with the indication of the nonresponse rate. The number of nonresponse observations was computed by multiplying the total number of observations by the nonresponse rate that was indicated. Random numbers were then generated to be matched on the observations of the second visit nonresponse variable. The same random numbers were implied for the two nonresponse variables for easier setting of nonresponse observations and to imply the assumption that the nonresponse observations were partial nonresponse. The data were sorted in ascending order by the random number generated earlier. In getting the observations to be set to nonresponse, the first set of observations from the second visit sorted data were taken. Since the program does not allow for missing values, the selected second visit observations were substituted by a negative value indicating that the values are set to nonresponse.