Surveys 2 Flashcards
Avoid these types of questions
- Double-barreled questions
- Ambiguous questions
- Recall-dependent questions
- Leading questions
- Non-neutral questions
- Social desirability question
Avoid these traits in surveys
- No overlapping answers (Options should be mutually exclusive and collectively exhaustive)
- Avoid typos
- Avoid formatting issues
- Avoid lengthy surveys
Length of questions
- Simple, short questions are preferable to long ones.
- As a rule of thumb, a question or a statement in a survey should not exceed 20 words.
- Better split questionnaire in multiple pages
- Better click next than having to scroll down
Test Ride / Pretest/Experts
Ask small sub-sample of population to respond to your survey and either ask them about the survey or analyze their responses.
Check whether your respondents understand the questions as you would expect them to.
Ask experts (like your coach and co-reader)
NOTE: Make sure the participants of your pre-test are not participating in the main study.
- ask small SUB-SAMPLE of the POPULATION
- check if they UNDERSTAND
- ask EXPERTS
Administering Survey
How will you contact respondents?
• Using online platforms for data collection, such as AMT / Prolific
• Using student samples/FB
• Company surveys: ask ‘support’ from management
A reminder can increase your response rate by up to 50%
Amazon Mechinal Turk (AMT)
• Very fast (up to 500 responses in a few hours)
• Relatively reliable (respondents are experienced)
Option to filter based on quality and experience
• Cost included
Rule of Thumb is approx. $0.10 per minute per respondent
Prolific Academic
• Fast (up to 200 responses in a few hours)
• More reliable (respondents are experienced)
Option to filter based on quality and experience
• Cost included
Rule of Thumb is approx. $0.11 per minute per respondent
Sequence of elements
- Introduction
- Experimental Task (choice, stimuli etc.)
- Start with DV
- Include manipulation checks
- Main IVs of your conceptual model (if these are not the manipulated variables in the experiment)
- Control variables
- Demographics
Order of questions
DV -> IV -> Controls
A good introduction
- Proper introduction that clearly discloses the identity of the researcher and conveys the purpose of the survey is necessary.
- Assurance of confidentiality of the information provided by them will allow for less biased answers.
Reliability
- How accurate is the measurement of our constructs?
- Cronbach’s Alpha (>0.7)
- Inter-rater reliability
Validity
Are we measuring the right thing?
- Face validity
- Convergent validity
- Discriminant validity
- External validity
Factor Analysis
• Check reliability of multi-item constructs
• Check if the questions we think reflect on 1 construct do
indeed reflect on that construct (convergent validity)
• Check if the questions we think reflect on 1 construct do NOT reflect on other constructs (discriminant validity)
Moderation effects
- If you include a moderator (x2), you need to include both the interaction term (x1*x2) as well as the main effect of the moderator (x2).
- You may consider floodlight / spotlight analysis
Possible threats to surveys
- Response biases
2. Common method bias (CMB)