Comparison of Interaction Models

Understanding Norman, GOMS, and KLM through practical examples and performance prediction.

1. Norman’s Model (The Execution–Evaluation Cycle)

Norman’s Model (Chapter 3: Interaction) breaks user interaction into two major phases, further divided into seven stages.

The Seven Stages of Action: Light Switch Example

  1. Goal: “I need more light.”
  2. Intention: “I will turn on the light switch.”
  3. Specify: “I will stand up, go to the switch, and prepare to press it.”
  4. Execute: Performs the specified action (the physical press).
  5. Perceive: The user sees the light turns on and notes, “The room is bright now.”
  6. Interpret: “The room is bright, so this means the light is on now.”
  7. Evaluate: “My goal was to get light, and now there is light! Mission complete.”

Key Limitation: It does not take into account the expertise of the user.

Purpose: The purpose of this model is to find out the Gulf of Execution (gap between the user's goal and system’s action) or the Gulf of Evaluation (gap between the user's goals and system presentation) in any UI so it can be improved.


2. GOMS Model

Expertise Assumption: Specify, Perceive, Interpret, and Evaluate are not explicitly modeled in GOMS because it assumes the user is an expert.

Example: The “Who Paid” Problem Hierarchy

GOAL: ADD-EXPENSE
.   [select GOAL: USE-QUICK-ADD-METHOD
.           CLICK-ADD-BUTTON-ON-NAV-BAR
.           GOAL: ENTER-EXPENSE-DETAILS
.                   TYPE-FRIEND-NAME-OR-GROUP
.                   TYPE-DESCRIPTION
.                   ENTER-AMOUNT
.           CLICK-SAVE-BUTTON
.   
.           GOAL: USE-GROUP-SPECIFIC-METHOD
.                   SELECT-GROUP-FROM-DASHBOARD
.                   CLICK-ADD-EXPENSE-IN-GROUP
.                   GOAL: ENTER-EXPENSE-DETAILS
.                           TYPE-DESCRIPTION
.                           ENTER-AMOUNT
.                   CLICK-SAVE-BUTTON]
      

The goal hierarchy in GOMS is useful for identifying task complexity and Short-Term Memory (STM) load.


3. KLM Model (Keystroke-Level Model)

This is a physical/device model used to predict user performance in detail. It is a low-level version of GOMS focused on smaller tasks like closing a window or using find-and-replace, rather than complex tasks like drawing (Dix, 436).

KLM Phases (Dix 12.5.1)

  • Acquisition: The mental representation of the task in the user's mind.
  • Execution: Using system facilities to perform the task. KLM focuses strictly on this part.

KLM Operators & Standard Values (Dix 438)

Standard Execution Times for KLM Operators
Operator Description Estimated Time (s)
K Keystroke 0.2s
P Pointing with mouse 1.1s
H Homing (hand movement between devices) 0.4s
M Mental preparation 1.35s
B Button click/release (full click = 0.2s) 0.1s

Task Comparison: Deleting a File

Method 1: Dragging to Trash (Mouse Only)
Step Description Operator Time (s)
1Mentally prepare to deleteM1.35
2Point to the fileP1.10
3Press and hold left buttonB[down]0.10
4Drag file to Trash iconP1.10
5Release mouse buttonB[up]0.10
Total Execution Time3.75s

Method 2: Keyboard Shortcut (Ctrl+D)
Step Description Operator Time (s)
1Mentally prepare to deleteM1.35
2Point to the fileP1.10
3Click to select the fileB[click]0.20
4Move hand to keyboardH0.40
5Press Ctrl+D (Two keys)2K0.40
Total Execution Time3.45s

Key Takeaways for Analysis

  • The "Homing" Factor: Method 2 includes a 0.4s penalty (H) for moving the hand, yet it remains faster.
  • Mental Operators (M): Placed before cognitive "chunks." In highly expert users, M might be reduced or removed.
  • Efficiency vs. Distance: As per Fitts' Law, if the Trash icon is far, the Pointing (P) time in Method 1 increases, making the Keyboard method even superior.