Human Factors Experiment upon Learning SQL: An Evaluation of the Roles of Conceptual Model and Computer Interface

Since SQL has been accepted as an ANSI standard, how to help SQL users to learn and use the language becomes an important issue for both vendors and training providers. At the same time, with the growing concern fur making DBMS easier to learn and use, the vendors have also used tabular interfaces in their product, such as QBE-like language. Tabular interfaces use two dimensional table-like qualities of relational systems as an aid in query writing. This trend has led to an interesting research question: "When user training is conducted, will this tabular-like interface help users to learn and use SQL?" Assimilation Theory suggests that meaningful learning can occur only if all three conditions, Reception, Availability and Assimilation, are met (Davis and Bostrom 1993). This theory provides a basis for understanding why a given type of interface may be more effective than another in training. It also suggests that in order to achieve meaningful learning, the cues must be provided to help learners to retrieve appropriate existing concepts. In an end-user computing environment, it is important that a designer or trainer of the DBMS query language can formulate ati appropriate coticeptual model (appropriate in the sense of being accurate, consistent, and complete) to ensure that ilidividuals learn to use the language effectively. Each operation of SQL takes either one or more relations (tables) as its operand(s) alid produces a new relation (table) as its result Reisner (1971) developed a syntactic model of query writing froin English questions. This model characterizes a query as consisting of a syntactic form, or template, together with the lexical items inserted into it. Both the template and the lexical items are created by a human process which transforms an English sentence into the relevant query components. To be able to identify characteristics that are important for conceptual models provided in a query language manual with different interfaces upon learning and using SQL in a query writing task, the present experiment attempts to test two hypotheses: (1) Will a table model better prepare users to deal with problems that may occur in query writing than a syntactic model? (2) Will it help users learn the language when a given model is consistent with the form of interface? This research adopted a framework based on Bostrom's end-user training model (Bostrom, Olfman and Sein 1990) but focusing only on the influences of the target system and training approaches on training outcomes. It defines the target A conducted to analyze the effect of two types of interfaces (a command based interface.with tmditional linear syntax and a system based on two interface types and training approaches based on two conceptual models. A laboratory experiment was table-like interface) and two conceptual models through training (syntactic model and table model). User performance wasmeasured through hands-on use of a relational DBMS.The preliminary results of this research suggest that using different conceptual models may not lead to a difference in userperformance, but using conceptual models in conjunction with the consistent form of interface ·can lead to a difference inuser performance. The table manipulation description with support of tabular interface yields user learning and using SQL. REFERENCESDavis, S. A., and Bostrom, R. R "Training End Users: An Experimental Investigation of the Roles of the ComputerInterface and Training Methods." MLS Quarterly, Volume 17, Number 1, March 1993, pp. 61-85.=r