Notes on Bias and Simple Random Sampling (SRS)
Bias types to avoid
- Interviewer bias: phrasing or tone can tilt the respondent's answer.
- Volunteer bias: volunteers are often more extreme; not representative of the whole group.
- Remedy: avoid relying on volunteers; use sampling techniques that include a broader, representative group.
Simple Random Sampling (SRS)
- Abbreviation: SRS (common in papers).
- Goal: every member of the population has an equal chance of selection.
Sampling frame
- Definition: list of all potential participants.
- Example: list of all students from the registrar.
- After obtaining the list, assign numbers 1 through N to the frame.
Random numbers and randomness
- Random number table (historical) vs. technology-based methods (Excel, TI-84, Google Sheets, random.org).
- True random vs pseudorandom:
- True random: no underlying rule; every number is equally likely.
- Pseudorandom: generated by algorithms; not truly random.
- Probability in true random sampling: P(i)=N1 for all i in the sampling frame.
- How to read a random-number table: read left-to-right; decide digits per number based on N (e.g., two digits if N ≤ 99).
Procedure: from numbers to sample
- Use the random numbers to pick individuals from the sampling frame by their assigned numbers.
- Map each selected number to the corresponding participant in the frame.
Example: Governors from 50 states
- Population: 50 governors, numbered 01–50.
- Desired sample: 10 governors.
- Read two digits at a time from the random-number source; each two-digit number selects a governor.
- This yields a simple random sample with no deliberate bias in selection.