Survey Data Cleaning With AI: Pattern Detection That Speeds Up the Tedious Work

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-research-data-cleaning-from-survey-creators

1. A respondent selects 'agree' for every single matrix question and completes the survey in 22 seconds. This pattern is an example of:

   1. Duplicate response patterns suggesting bot activity
   2. Impossible value combination requiring exclusion
   3. A legitimate fast response pattern
   4. Straightlining behavior indicating possible satisficing

2. Which task is AI specifically well-suited to perform in survey data cleaning?

   1. Determining whether a borderline response should be included in final analysis
   2. Flagging responses where age=12 and marital status='married 5+ years'
   3. Deciding which cleaning rules align with research goals
   4. Making final inclusion/exclusion calls for the dataset

3. A respondent writes 'asdfghjkl' as their answer to an open-ended question. What should AI flag this as?

   1. Gibberish or off-topic response
   2. Speed flag violation
   3. Missing data point
   4. Straightlining pattern

4. What is a fundamental limitation of AI in survey data cleaning?

   1. AI cannot process survey response data efficiently
   2. AI cannot detect straightlining patterns at scale
   3. AI cannot distinguish between high and low severity issues
   4. AI cannot identify systematic bias that falls outside predefined cleaning rules

5. What does the lesson recommend researchers do after AI flags problematic survey responses?

   1. Automatically exclude all flagged responses from analysis
   2. Run sensitivity analyses with and without flagged respondents
   3. Delete flagged responses from the database immediately
   4. Use flagged responses only for descriptive statistics

6. Which of the following represents an 'impossible value combination' that AI should flag?

   1. Respondent reports age 8 and grade in school='12th grade'
   2. Respondent reports being unmarried and spouse name='John Smith'
   3. Respondent reports age 25 and income of $150,000
   4. Respondent reports being employed full-time and student full-time

7. The lesson distinguishes between what AI can do and what researchers must do. What is the researcher's responsibility that AI cannot replace?

   1. Identifying patterns in large datasets
   2. Flagging responses completed in under 30 seconds
   3. Making the final inclusion/exclusion decision for borderline cases
   4. Detecting duplicate response patterns

8. A respondent's completion time is 20% of the median completion time. Based on the lesson, this would trigger what type of flag?

   1. Gibberish flag
   2. Speed flag
   3. Duplicate pattern flag
   4. Straightlining flag

9. The lesson warns that AI cleaning rules may miss certain problems. Which problem is specifically mentioned as potentially invisible to AI?

   1. Straightlining patterns
   2. Systematic bias not captured by cleaning rules
   3. Responses with missing data
   4. Gibberish in open-ended responses

10. Why might two nearly identical survey responses from different IP addresses still be flagged as duplicates?

    1. They violate the straightlining rule
    2. They suggest possible bot or fraudulent activity
    3. They contain impossible value combinations
    4. They have missing data in open-ended fields

11. The lesson emphasizes that cleaning rules should be reported in which section of research?

    1. Methods section
    2. Abstract
    3. Results section
    4. Appendix only

12. A respondent's answers show a consistent pattern that matches 4 other respondents exactly, including identical answers to open-ended questions. This would most likely be flagged as:

    1. Straightlining
    2. Duplicate response pattern
    3. Gibberish
    4. Speed flag violation

13. The lesson notes that AI cannot substitute for human-coded validity flags on what type of cases?

    1. Speed violations
    2. Borderline cases requiring judgment
    3. Straightlining responses
    4. Impossible value combinations

14. What distinguishes straightlining from other quality issues in survey responses?

    1. It is detected by analyzing completion time alone
    2. It requires comparing responses to impossible values
    3. It involves giving minimal effort across many similar questions
    4. It only applies to open-ended questions

15. A survey shows respondents completing it in times ranging from 45 seconds to 25 minutes. If the median time is 12 minutes, which completion time would definitely trigger a speed flag?

    1. 14 minutes (117% of median)
    2. 6 minutes (50% of median)
    3. 8 minutes (67% of median)
    4. 4 minutes (33% of median)