Sampling & Errors
What is the difference between a population & a sample in research?
Population = everyone
Sample = subset of population
What type of scale asks respondents to rate their agreement on a 1–5 or 1–7 scale with statements?
Likert
What is one advantage and one disadvantage of online surveys?
Advantage: Fast, inexpensive, easy to distribute.
Disadvantage: Risk of low response rate, lack of clarification.
Why is it important to avoid double-barreled questions in a survey?
It confuses respondents because it asks about two things in one.
e.g., “Do you find our service fast and courteous?”
In the 2016 U.S. election, what was one sampling-related reason polls failed to predict the correct outcome?
Didn’t weight by education — underrepresented non-college voters (a key Trump demo).
Line graphs are best for which data?
continuous
Name 2 non-probability sampling methods. When may you use each of these?
Convenience sampling — easiest to access (e.g., students in hallway).
Snowball sampling — used for hard-to-reach groups (e.g., referrals in niche communities).
Describe the main advantage and main disadvantage of a semantic differential scale.
Advantage: Quick and effective for measuring attitudes.
Disadvantage: Can lack nuance; respondents may interpret adjectives differently.
Describe how mall intercept surveys differ from mail surveys in terms of bias and control.
Mall Intercept: Face-to-face, more interviewer bias, real-time clarification.
Mail: No interviewer bias, less control, delayed responses.
Give an example of a poorly worded leading question, and rewrite it to be neutral.
Poor: “Don’t you think our product is amazing?”
Better: “How satisfied are you with our product?”
What is nonresponse bias and how could it affect survey data in a healthcare study?
People who don’t respond may differ systematically from those who do (e.g., sicker patients may skip a survey).
Give examples of when mean, median, and mode are best used.
Mean: Interval/ratio data (constant sum)
Median: Ordinal (rank orders)
Mode: Nominal (zip code, gender)
A company used an outdated customer list and included non-customers in their survey. What type of sampling error is this, and how could it affect results?
Frame error.
It leads to biased results by surveying people outside the target population.
You want to avoid a neutral response and force people to choose a side. What scale design decision supports this, and what are the pros and cons?
Forced-choice, even scale.
Pros: No neutral fence-sitting, encourages decision.
Cons: May frustrate respondents who genuinely feel neutral.
Independent: we can manipulate to see its effect on the dependent variable
When giving closed-ended responses on a questionnaire, what 2 criteria must be met?
Mutually exclusive & exhaustive
A soda company ran taste tests to find out if people liked the new formula, but didn’t ask if they’d prefer it over the original. What error is this and why is it critical?
Surrogate information error.
They didn’t measure the decision-making behavior they actually needed to predict.
Give examples for each: positively skewed, negatively skewed, normal curve.
positive: # kids
negative: age of death
normal: height