The Use of Computer-Assisted Identification of ARIMA Time-Series.

N

24

Computer Assisted Instruction; Computer Programs; Data Analysis; Higher Education; Instructional Materials; Mathematics Instruction; Models; Research Methodology; Social Psychology; Statistical Analysis; Teaching Methods

This study was conducted to determine the effects of using various levels of tutorial statistical software for the tentative identification of nonseasonal ARIMA models, a statistical technique proposed by Box and Jenkins for the interpretation of time-series data. The Box-Jenkins approach is an iterative process encompassing several stages of development, with the initial step requiring a tentative identification which relies on two statistics, the autocorrelation function (AUCF) and the partial autocorrelation function (PACF). Both functions are calculated by the statistical software on the time-series data and supplied to the user as a correlogram. A total of 56 time-series data files were evaluated by 30 novice undergraduate students in a repeated measures procedure. Students were randomly assigned to statistical software packages which produced either AUCF and PACF correlograms (a non-tutorial package), the correlograms and access to theoretical ARIMA model correlogram examples, or correlograms and access to a decision-support tutorial. Results indicated that the use of tutorial software significantly improved the identification of ARIMA models over non-tutorial software, and the type of tutorial significantly interacted with the degree of complexity of the time-series data. The report includes 4 references, and 10 tables and figures presenting study data. (LMM)

1983

ED226705