Lecture 1: Notes on Measurement Scales: Ratio Scale and Data
Key ideas and framing
The speaker challenges a statement about time-zero framing: "That's not really true. Right? The earth existed years and years before, zero starts." This highlights a common confusion about where zero lies in measuring time or quantities.
Emphasizes the concept of scales in measurement, contrasting time-origin notions with the idea of a ratio scale.
The transcript notes: "On the other hand, like, a ratio scale is one that" (incomplete), leading into a formal definition of ratio scale.
Data are described as the collection of variables being measured.
Practical takeaway: the speaker encourages students not to be afraid to ask for help.
The passage ends with a closing prompt: "I'm gonna go ahead and end there. If somebody else, you have any" (invitation for questions).
Ratio scale: definition and core properties
Definition (inferred from the transcript plus standard theory): a ratio scale is a measurement scale that has a true zero, equal intervals between adjacent values, and meaningful ratios between values.
Key properties:
True zero point: 0 indicates the complete absence of the quantity.
Equal intervals: the difference between adjacent values is consistent across the scale.
Meaningful ratios: for any values (x) and (y) with (x>0) and (y>0), the ratio (\frac{x}{y}) is meaningful.
Arithmetic operations: addition, subtraction, multiplication, and division are meaningful on ratio-scale data.
Mathematical representation:
\text{Ratio scale properties: } 0 \text{ is the true absence of the quantity, and } \frac{x}{y} \text{ is meaningful for } x,y>0.Examples of ratio-scale data:
Height (e.g., cm),
Weight (e.g., kg),
Age (e.g., years),
Duration (e.g., seconds),
Income, etc.
Contrast with other scales:
Interval scale: equal intervals but no true zero (e.g., Celsius, Fahrenheit). Ratios like (30/0) are not meaningful.
Ordinal/Nominal scales: qualitative or ordered categories without meaningful numerical differences or ratios.
Data and variables
Data are described as the collection of variables being measured.
Each variable may be measured on a different scale; the ratio-scale status of a variable dictates what statistical operations are appropriate.
Implication: choosing the correct scale guides which descriptive and inferential statistics can be valid (e.g., means and standard deviations are meaningful for ratio-scale data).
Practical takeaways and guidance
Advice from the speaker: "please don't be afraid to ask for help" when dealing with measurement concepts.
Practical checks for data to determine scale type:
Is there a true zero that indicates absence?
Are intervals between measurements equal?
Are ratios between measurements meaningful?
Based on scale type, select appropriate statistical methods and interpretations.
Connections to previous lectures / foundations
This content fits with foundational concepts of measurement levels: nominal, ordinal, interval, and ratio.
Understanding the scale type informs appropriate data analysis choices (e.g., which averages are appropriate, which transformations are valid, etc.).
Examples and hypothetical scenarios
Example scenario: height in cm is ratio-scale because 0 cm means no height, and a person who is 180 cm tall is twice as tall as someone 90 cm tall.
Hypothetical nuance: temperature in Celsius is interval-scale (equal intervals) but does not have a true zero that implies absence of temperature; hence, the ratio 60°C/30°C is not a meaningful physical comparison.
Time-to-event measures can be tricky: if zero is a fixed origin (e.g., time since birth), ratios may be meaningful; if zero is arbitrary, interpretation of ratios may be misleading.
Ethical, philosophical, or practical implications
Misclassifying the scale can lead to invalid conclusions and misused statistics.
Careful attention to data scale prevents inappropriate inferences and promotes rigor in data interpretation.
Encourages proactive help-seeking when uncertainty about scale or methods arises.
Quick reference formulas and notes (LaTeX)
Ratio scale property:
0 \text{ is the true absence of the quantity, and } \frac{x}{y} \text{ is meaningful for } x,y>0.Arithmetic mean (for context, valid on ratio-scale data):
\bar{x} = \frac{1}{n}\sum{i=1}^{n} xi.Geometric mean (for positive values, often used with ratio data):
G = \left(\prod{i=1}^{n} xi\right)^{1/n}, \quad x_i > 0.Distinction reminder:
Ratio-scale data support both additive and multiplicative operations (e.g., differences and ratios).
Interval-scale data support additive operations, but ratios may be misleading due to the lack of a true zero.
Closing note
The fragment ends with an invitation for questions, reinforcing the importance of seeking help when scale concepts are unclear.