Measurements Flashcards
What should you consider with regards to scale selection with regards to the subjects?
- Meaningful to subjects (ie will children be able to meaningfully use it?)
- Uncomplicated to use
- Magnitude of scale
Give an example of scales that have been adapted to their subjects.
1 - emoji scale for teens
2 - “yummy yucky” scale for children
What scientific considerations should you take in making scales?
- Must be unbiased
- Relevancy
- Sensitive to differences (length/categories)
(Ie 9 point scale has more sensitivity to differences than 5 point scale)
Is scaling something from “dry” to “juicy” biased? If so, how would you change it? Why would the wording of something make it biased?
- Yes
- Low juicy to high juicy
- Words can make pre conceived ideas/can describe negative/positive attributes making it more subjective
What are the “end effects” on a scale?
The highest and lowest point on a scale (0 and 10)
Why do people gravitate towards the middle numbers on a scale
People save the ends more extreme samples (highest intensity) and they avoid using it in case the next sample is more intense. Another reason would be if they are unsure and want ton”place a safe bet”
How do you avoid the “end effects”
Scale with indented lines - provides physiological comfort so that they can use the full sale.
used by trained panels
Which kind if data is normally distributed?
Parametric data
Which type of data is distribution free?
Non-parametric
Which type of data is normally distributed?
Parametric
Why is normal distribution important?
You can use more powerful statistical techniques
Why is it important to know what kind of data you are working with?
Dictates the type of statistical analysis you can use
What does a normally distributed graph with very little standard deviation look like?
Slim thin peak (look at slides)
What does it mean if there is less standard deviation?
Less variation/spread in data (more consistent)
What does a normally distributed graph with high standard deviation look like?
Wide, low peak, spread (look at slides)
What does it mean if there is a high standard deviation in data?
A lot of variability
What is nominal data?
- Simplest form of data
- NO NATURAL ORDERING
- Has no numerical value.
What do you use nominal data for?
Labeling, coding, classifying items or responses
How would you analyze nominal data?
Non-parametric analysis
- Frequency counts
- mode
- Chi-squared
What are the strengths of Nominal data
- easy to use by subjects
- short test time
- simple computations for getting results
What is a limitation of nominal data?
Easy to Mis-classify responses (ie different people interpret color differently)
What is ordinal data?
Numbers that represent RANK (distances between numbers not necessarily the same — don’t know this from data)
Ie - low to high
What type of scales are used for ordinal data
Ranking and rating scales (look at slide)
What kind of analysis is used for ordinal data?
non parametric (Friedman analysis for ranked data)
