mirror of
https://github.com/jackyzha0/quartz.git
synced 2025-12-24 13:24:05 -06:00
53 lines
1.5 KiB
Markdown
53 lines
1.5 KiB
Markdown
---
|
|
title: "designing-studies"
|
|
aliases:
|
|
tags:
|
|
- info203
|
|
---
|
|
|
|
Need to be more specific than "Do you like my interface". Need to avoid experimenter bias
|
|
|
|
# Terms
|
|
- **Comparison:** What is good
|
|
- "Yark stick"
|
|
- **Baserate:** how often does Y occur
|
|
- requires measurin Y
|
|
- **Correlation:** Do X and Y co-vary
|
|
- Requires measuring X and Y
|
|
- **Causes:** Does X cause Y
|
|
- Requires measuring X and Y and manipulating X
|
|
- **Manipulations:** Indepenedent variables
|
|
- **Measures:** Dependent variables
|
|
- e.g., task completion time, recall, accuracy, emotional response
|
|
- **Precision:** Internal Validity
|
|
- remove confounding factors
|
|
- large sample size
|
|
- **Generlisability:** External Validity
|
|
- does this apply in the real world
|
|
|
|
# Strategies for fairer comparisons
|
|
Two main things differ between prototype and Production: Fidelity (how "polished the design is", Approach (the actual design)
|
|
|
|
- insert your new approach into the production setting
|
|
- recreate the production approach in you new setting
|
|
- scale down so youre just looking at a piece of a larger system
|
|
- when expertise is relevant, train people up
|
|
|
|
## Bad example
|
|
|
|
|
|
manipulation: input style
|
|
measure: words per minute
|
|
external validity: not so much
|
|
|
|
benefits and drawbacks
|
|
- not a fair example: novices vs experts
|
|
- are the results significant
|
|
|
|
## Good example
|
|
manipulation: input style
|
|
measure: words per minute and error rate
|
|
|
|
|
|
|