Presentation is loading. Please wait.

Presentation is loading. Please wait.

Department of Industrial Management Engineering 1.Introduction ○Usability evaluation primarily summative ○Informal intuitive evaluations by designers even.

Similar presentations


Presentation on theme: "Department of Industrial Management Engineering 1.Introduction ○Usability evaluation primarily summative ○Informal intuitive evaluations by designers even."— Presentation transcript:

1 Department of Industrial Management Engineering 1.Introduction ○Usability evaluation primarily summative ○Informal intuitive evaluations by designers even users unable to provide sufficiently accurate feedback ○Why formative evaluation seldom practiced? Egocentric intuition fallacy ○Illusion about knowing the determinants of one’s own behavior ○Underestimate variability in performance and preference Behavioral Research

2 Department of Industrial Management Engineering 2.Goals Behavioral Research New Technology Existing Technology New Domain Guidelines & Standards Expand the applicability Existing Domain Evaluation and Comparison More general understanding of the determinants

3 Department of Industrial Management Engineering 4.Research Design 1.Bias ○Because of the fundamental nature of human perception and thinking, human judgment not by the evidence at hand but by personal interpretation of events ○Experimenter bias – subtle and insidious; it often works by simple, unintentional, unconscious shifts in attention or interpretations ○To avoid bias  double-blind procedures; always have a comparison system Behavioral Research

4 Department of Industrial Management Engineering 4.Research Design 2.Statistical Significance ○Not so important ○More interested in knowing how much difference there probably is to guide cost-benefit strategy ○Science decisions can be postponed without enough evidence while engineering decisions can be made on whatever evidence is at hand Behavioral Research

5 Department of Industrial Management Engineering 4.Research Design 3.Generality and Transfer ○Participants who are similar in background, occupation, activity, age, and variability ○Systems that are easy for first half hour with novices, but not very effective thereafter  usability testing was done only in “first-use” settings ○Robustness in variation  generality Behavioral Research

6 Department of Industrial Management Engineering 4.Research Design 6.Invention and Specification Oriented Method ○task analysis – what task to design and how it should be performed ○Thoughtful observation and analysis of the whole task in its naturalistic context – ethnographic method Human-problem Oriented Invention 1.Identify a goal 2.Find what causes the difficulty 3.If possible, invent a way to help people around the difficulty 4.Incorporate it in a system Behavioral Research

7 Department of Industrial Management Engineering 4.Research Design 6.Invention and Specification Oriented Method Failure Analysis ○ Find out where people go wrong or slowly Individual Difference Analysis Time Profiling ○ Measuring the amount of time spent on all isolatable components ○ Components occupying large amount of time are obvious candidates for new method ○ Variability of time Behavioral Research

8 Department of Industrial Management Engineering 4.Research Design 7.Design-Oriented Research Method ○Full-scale Evaluation Studies ○Simply compare the overall effectiveness of two or more systems – fairly standard and obvious ○The biggest problems in the cost and in the selection of the sample of participants ○Already experienced users of each of the different kinds of systems ○What to measure – a set of benchmark tests ○Study strengths and deficiencies of systems already long in use Behavioral Research

9 Department of Industrial Management Engineering 4.Research Design 7.Design-Oriented Research Method ○Formative Evaluation ○The best strategy for good design is to try various options, test them, redesign, and iterate ○Effective and economical ○2-3 iterations, each requiring less than a dozen hours of human testing and an equivalent amount of reprogramming Behavioral Research

10 Department of Industrial Management Engineering 5.Measurement and Analysis 1.What to Measure and How Many ○Power ○The ability of the experiment to detect differences of a certain size, or to make measurements at a certain level of precision ○Determined by the accuracy of the measurements, the amount of variability in the performance, and the size of sample ○What Measurements? ○Completion time, number of errors, subjective satisfaction ○Individual differences (cognitive ability measures such as spatial memory, logical reasoning) ○Measures during the marketing or use of a system Behavioral Research

11 Department of Industrial Management Engineering 5.Measurement and Analysis 2.Data Quality ○To prevent inadvertent bias, data recording and analysis should be fully automated ○Take excessive caution 3.Reliability ○Random as well as bias errors ○Standard approach in Psychology – to have observations made by two different people working independently for a sizable random or representative subsample of the occasions ○Increase reliability – revise the way the judgments are made, more explicit rules, train more thoroughly, more practice and feedback Behavioral Research

12 Department of Industrial Management Engineering 5.Measurement and Analysis 3.Reliability ○Statistical Analysis ○Summarizing Results Accurately ○Descriptive statistics – central tendency measures ○Robust statistical methods ○Trimming – don’t use a certain proportion ○Measure of variability ○Probability distribution ○Confidence interval Behavioral Research

13 Department of Industrial Management Engineering 5.Measurement and Analysis 3.Reliability ○Estimating Effect Size: The Most Important Step ○Magnitude of the effect of a difference ○Critical ratio, odds ratio ○Instead of significance test, effect sizes, CI, and odds ratios are the proper tools Behavioral Research

14 Department of Industrial Management Engineering 5.Measurement and Analysis 3.Reliability ○Exploration ○Exploratory data analysis -- plotting of data – scaled differently ○Regression, Structural Analysis (factor analysis, multi- dimensional scaling, cluster analysis) ○Modeling Behavioral Research


Download ppt "Department of Industrial Management Engineering 1.Introduction ○Usability evaluation primarily summative ○Informal intuitive evaluations by designers even."

Similar presentations


Ads by Google