Reinforcing students\u27 correlation comprehension

Graphical presentation of data provides a fast way of extracting information from data and is the primary method for illustrating complicated data structure, and thus facilitates the process of data examination. The graphical representation of data provides valuable clues to what and how further statistical work ought to be carried out because technically correct statistical graphics disclose significant elements of the information they represent and also serve as an alternative means of effective communication. Even though students who know how to draw statistical graphs and charts often fail to present data in proper graphical format and are unable to convey their thoughts about their findings from their graphical displays of data to their audience because they have difficulty gaining insights and seeing hidden relationships of data. This hinders the development of students\u27 graphical understanding of scatterplots, a statistical topic taught in undergraduate study. In this paper we present a critical review of current research in this area from three different perspectives: pedagogy, statistics and cognitive psychology. Specifically, Curcio (1987) outlined a classification model of reading graphical data from a mathematics educator\u27s perspective; Cook and Weisberg (1997) advocated a model of statistical graphing based on the statistician\u27s viewpoint; while Carpenter and Shah (1998) developed their model of graphical comprehension and interpretation from a cognitive science position. Arising from our synthesis of this research we propose a cognitive model of correlation comprehension which helps statistics lecturers to structure undergraduate students\u27 thought so as to reinforce their graphing capabilities when teaching how to solve correlation problems