Survey Research Insights

Study Notes

Survey responding is a psychological process involving several stages.
The stages include question interpretation, information retrieval, judgment formation, response formatting, and response editing.
These stages must be considered when constructing survey questionnaires.

Respondents must interpret the question's meaning.
Respondents need to determine an appropriate timeframe (e.g., typical day vs. a week).
A survey item should specify numerical values for the question of interest, making better use of a construct like average.
Survey questions should quantify the amount/quantity in numerical terms to make the responses meaningful. (e.g., specify alcohol in ounces, drinks, etc)

Context effects are responses influenced by the surrounding questions, not the question's content.
The order of questions can influence responses (item-order effect).
The placement of political candidates on a ballot can affect how voters perceive their qualifications.
The phrasing of response options can manipulate interpretation. (e.g. How often do you feel really irritated, options of less than a year vs. less than a day)
- Changing the timeframe of the responses alters the interpretation of what constitutes "really irritated".
The middle/center option is often perceived as the standard response (average/normal/typical).

Open-ended items are used to gain detailed information on a topic not well-known.
Open-ended items provide valuable insights and subjective experiences that close-ended questions may miss, but they can be difficult or subjective to quantify.
Closed-ended items are suitable when variables are well-defined.
- they're practical for large-scale assessments due to their efficiency in coding

Rating scales provide an ordered set of responses to make questions effective. - (e.g., strongly agree, agree, neither agree nor disagree, disagree, strongly disagree).
Avoid double-barreled questions (asking multiple things within a single question).
Scales range from 5-7 options are appropriate (or 0-10)
Extreme scores should balance around the neutral point (in bipolar scales)
The modal/neutral point is optional