University of Washington HCDE 518 & INDE 545 Empirical Evaluation HCDE 518 & INDE 545 Winter 2012 With credit to Jake Wobbrock, Dave Hendry, Andy Ko, Jennifer Turns, & Mark Zachry

University of Washington HCDE 518 & INDE 545 Agenda  Announcements  Class Activity – Usability Test  Discussion – Remaining Empirical evaluation questions  Break – 5 mins  Lecture & Discussion – Analytical Evaluation  Class Activity – Heuristic Evaluation  Next class

University of Washington HCDE 518 & INDE 545 Announcements  R8, P3 due today  A3 due next Monday  Questions?

University of Washington HCDE 518 & INDE 545 Class Activity: Usability Test  Write a set of 3 usability test tasks for a cell phone, laptop, or anything you have handy  Include description, start state, criteria for success  Pair up with someone in class and have them try your tasks while you try theirs on whatever device you have handy  Remind your partner to think aloud, and take notes about where they struggled or what was difficult, or whether or not they were able to complete the task

University of Washington HCDE 518 & INDE 545 Discussion Questions - Jessica  1. Nielsen’s article discussed that the best results can come from testing no more than 5 users and running as many small test as possible. Can a user study ever have too many users in their study? Give examples and why. 2. In the Dumas and Redish reading they wrote that usability testing “can break down the wall between those who create the product and those who use it.” Are there any instances that usability testing should be skipped during a design of a product? 3. A motto used in usability testing is “test early and test often.” When testing different iterations of a product is it better to test with the same group over the life time of the development of the project or a different group each time? Why? 4. Scenarios are a good way to test out a product. Can a scenario prime a user too much to use a product in a different way they may not have naturally used the product? Is this good or bad? Why? 5. Objective measures vs. subjective measures. Studies have been shown that objective measures do not necessarily reflect subjective measures. If so, why are objective measures still considered “good enough” for validation of a product?

University of Washington HCDE 518 & INDE 545 Discussion Questions - Jessica  6. In the Dumas and Redish reading, they wrote that the goal of a research study is to test if a phenomenon exists while the goal of the usability test is to uncover problems. Are these studies always separable or can a research study be a usability study? 7. Many research projects (like ones at the UW) create technologies or products. Do usability studies need to be conducted before a research study is performed? What are the pros and cons of performing one over the other? Is the research study more important than a usability study (especially in a research University setting)? 8. Usability testing may uncover multiple problems that stem from multiple causes. If a multi-problem instance occurs, how do you pinpoint the root cause of the problem? For example, as the researcher do you step in and question the breakdown that just occur, wait till the end of the usability testing session, or employee a different method? 9. In the Hornbaek reading, they wrote that satisfaction measures should be measured beyond post-use questionnaires. What are some other ways to measure post-use satisfaction other than a questionnaire? 10. Can a users’ attitude towards a product positively or negatively impact usability results? If so, how do you account for these influences?

University of Washington HCDE 518 & INDE 545 Discussion Questions - Sarah  1. According to Nielsen, you only need to test 5 users to discover usability problems in design. Is there an instance where more than 5 users should be tested?  2. Hornbaek argues that both subjective and objective measures of time give a more complete picture of usability. Do you agree or disagree  3. Dumas & Redish give a couple of ideas on how to get participants to pause between tasks. Are there other non-disruptive ways to have participants pause, without giving feedback in between each task?  4. Dumas & Redish have a section on co-discovery. Has anyone been able to use this technique in their testing?

University of Washington HCDE 518 & INDE 545 BREAK – 5 MINUTES

University of Washington HCDE 518 & INDE 545 Reminder - Types of Evaluation  Empirical ( involves users )  Usability testing  Field studies  Click-through studies  Analytic ( design judgment – users not involved )  Heuristic evaluations  Standards enforcement  Cognitive walkthroughs

University of Washington HCDE 518 & INDE 545 Analytical Evaluation  Heuristic Evaluation  Have usability experts go through your prototype to uncover common usability problems  Cognitive Walkthrough  Have experts analyze your prototype in a detailed way to understand how users will understand it  Best for understanding novel use, not expert use 

University of Washington HCDE 518 & INDE 545 Heuristic Evaluation  Developed by Jakob Nielsen  Helps find usability problems in a UI design  Small set (3-5) of evaluators examine UI  independently check for compliance with usability principles (“heuristics”)  different evaluators will find different problems  evaluators only communicate afterwards  findings are then aggregated  Can perform on working UI or on sketches

University of Washington HCDE 518 & INDE 545 Heuristic Evaluation Process  Evaluators go through UI several times  inspect various dialogue elements  compare with list of usability principles  consider other principles/results that come to mind  Usability principles  Nielsen's “heuristics”  supplementary list of category-specific heuristics  competitive analysis & user testing of existing products  Use violations to redesign/fix problems

University of Washington HCDE 518 & INDE 545 Heuristics (Nielsen, 1994) 1.Visibility of system status 2.Match between system and the real world 3.User control and freedom 4.Consistency and standards 5.Error prevention 6.Recognition rather than recall 7.Flexibility and efficiency of use 8.Aesthetic and minimalist design 9.Help users recognize, diagnose, and recover from errors 10.Help and documentation

University of Washington HCDE 518 & INDE 545 Heuristics  H2-1: Visibility of system status  keep users informed about what is going on  example: pay attention to response time  0.1 sec: no special indicators needed  1.0 sec: user tends to lose track of data  10 sec: max. duration if user to stay focused on action  for longer delays, use percent-done progress bars

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-2: Match between system & real world  speak the users’ language  follow real world conventions

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-3: User control & freedom  “exits” for mistaken choices, undo, redo  Don’t force down fixed paths  Use wizards sparingly

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-4: Consistency & standards

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-5: Error prevention  Try to prevent users from making errors in the first place  H2-6: Recognition rather than recall  make objects, actions, options, & directions visible or easily retrievable

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-7: Flexibility and efficiency of use  accelerators for experts (e.g., gestures, keyboard shortcuts)  allow users to tailor frequent actions (e.g., macros)

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-8: Aesthetic & minimalist design  no irrelevant information in dialogues

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-9: Help users recognize, diagnose, & recover from errors  error messages in plain language  precisely indicate the problem  constructively suggest a solution

University of Washington HCDE 518 & INDE 545 Heuristics (cont.)  H2-10: Help and documentation  easy to search  focused on the user’s task  list concrete steps  not too large  use in-context help

University of Washington HCDE 518 & INDE 545 How to Perform Evaluation  At least two passes for each evaluator (3-5 people)  first to get feel for flow and scope of system  second to focus on specific elements  If system is walk-up-and-use or evaluators are domain experts, no assistance needed  otherwise might supply evaluators with scenarios  Each evaluator produces list of problems  explain why with reference to heuristic or other information  be specific & list each problem separately

University of Washington HCDE 518 & INDE 545 Example Errors from Evaluators  Can't copy info from one window to another  violates “Recognition over recall” (H6)  fix: allow copying  Typography uses different fonts in 3 dialog boxes  violates “Consistency and standards” (H4)  slows users down  probably wouldn’t be found by user testing  fix: pick a single format for entire interface

University of Washington HCDE 518 & INDE 545 Severity Rating  Used to allocate resources to fix problems  Estimates of need for more usability efforts  Combination of  frequency  impact  persistence (one time or repeating)  Should be calculated after all evals. are in  Should be done independently by all judges

University of Washington HCDE 518 & INDE 545 Severity Ratings (cont.) 0 - don’t agree that this is a usability problem 1 - cosmetic problem 2 - minor usability problem 3 - major usability problem; important to fix 4 - usability catastrophe; imperative to fix

University of Washington HCDE 518 & INDE 545 Debriefing  Conduct with evaluators, observers, and development team members  Discuss general characteristics of UI  Suggest potential improvements to address major usability problems  Dev. team rates how hard things are to fix  Make it a brainstorming session  little criticism until end of session

University of Washington HCDE 518 & INDE 545 Severity Ratings Example 1. [H4 Consistency] [Severity 3] The interface used the string "Save" on the first screen for saving the user's file, but used the string "Write file" on the second screen. Users may be confused by this different terminology for the same function.

University of Washington HCDE 518 & INDE 545 HE vs. User Testing  HE is much faster  1-2 hours each evaluator vs. days-weeks  HE doesn’t require interpreting user’s actions  User testing is far more accurate (by def.)  takes into account actual users and tasks  HE may miss problems & find “false positives”  Good to alternate between HE & user testing  find different problems  Don't waste participants

University of Washington HCDE 518 & INDE 545 Class Activity: Heuristic Evaluation  Electronic voting machine  Download prototype:   Download form:   Nielsen’s heuristics:   Use form and Nielsen's 1994 heuristics to evaluate the voting interface in groups of 2-3

University of Washington HCDE 518 & INDE 545 Next Class Topics  Wednesday, February 29th  Design Research  Discussants: Ben & Sarah  Monday, March 5 th  Trends in UCD  Upcoming Work  R9, A3

University of Washington HCDE 518 & INDE 545 GROUP PROJECT TIME