Comparative analysis of iconic representation of select word processing functions with command string counterparts as a function of experience

TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Polytechnic Institute and State University


The primary motivation for this research was to study the effect of presentation format on acquisition time, recall, and preference. While the main focus was on the relative difference between iconic and textual representations, the independent variables include: presentation format (icon, text and mixed), experience (low, medium, and high), command set (basic, enhanced) and session (primary and secondary recall) . Dependent measures include: acquisition time, response times (primary and secondary recall), and response errors (primary and secondary recall).

To assess the presence of functional differences between the independent variables, the ANOVA procedure was applied to each set of dependent measure data. Where appropriate, significant effects were subsequently investigated via application the Newman-Keuls post hoc test procedure. Finally, subject preference measures were correlated with dependent measure data to determine how they were related. The findings associated with each analysis are detailed below.

A significant effect was obtained for the main effect of format using acquisition time as the dependent variable. Thus, indicating that acquisition time was influenced by the presentation format of select word processing commands. No other acquisition effects were found to be significant.

An ANOVA utilizing response time as the dependent variable supported the hypothesis that presentation format affects average primary response time (p < .01). Additionally, the interaction of the main effects of format and setblock achieved significance (p < .01).

Format, setblock and the interaction of the two main effects produce significant (p < .01) average differences in terms of response errors. Therefore, the hypothesis that format and setblock significantly affect the number of response errors was supported.

Comparison of the ANOVA summary tables for both response time and errors for primary and secondary tasks indicate that in general, the effects that held significance in the primary recall tasks also achieved (or perhaps more correctly maintained) significance in the secondary recall task. Specifically, using response time as the dependent variable, significance was achieved for the main effects of format (p < .01), and setblock (p < .01), and for the interaction of experience and setblock (p < .05).

Finally, using response errors as the dependent variable, the format by setblock interaction was determined to be significant (p < .01). The results support the hypothesis that the number of secondary recall percent correct responses is affected by format and setblock.

Subject preference data was utilized to determine the correlation between preference and: average acquisition time, average primary recall response time, average primary recall response errors, average secondary recall response time, and average recall response errors. The correlations achieved indicate that preference is not a particularly useful indicator of performance.

In an attempt to summarize the results and findings of the study, the following recommendations/observations are set forth:

• With the exception of novices, word processing experience does not dramatically influence the time required to learn or recall stimuli. Similarly, experience does not affect the number of response errors made.

• Word processing systems which require subjects to learn and recall exact command sets benefit from enhanced text or enhanced text with graphics.

• In applications where a precise verbal response is required, or where differences between functions is subtle, use of an accepted term in conjunction with a graphic is the preferred method.

• In general, preference appears to be a poor predictor of performance. Therefore, preference should not be the only basis for deciding on command set formats. However, when subject preference is the primary driver, enhanced text commands should be used. The next preferred approach is the mixed enhanced command set.