Menu
Data Scrubbing
Data Scrubbing
b³’s bad responder identification (AKA Data Scrubbing) solution utilizes multiple techniques stemming from the basics of human psychology. The Basic premise being that humans are habitual and will do certain things repeatedly. Our scrubbing offer uses this trait coupled with out of the box thinking to identify bad responders.
What is Data Scrubbing?
scrub
verb (used with object), scrubbed, scrub-bing
• to rub hard with a brush, cloth, etc., or against a rough surface in washing.
• to subject to friction; rub.
verb (used with object), scrubbed, scrub-bing
• to cleanse something by hard rubbing.
• to cleanse one’s hands and arms as a preparation to performing or assisting in surgery(often followed by up).
Identification and removal of bad responders
Responders who take the survey
• Without paying attention
• Clicking through just to get to the end
• Lying to participate
• Participating just to get incentives
Uniqueness
Evaluates responses to questions
• Scrubbing looks at responses to survey questions
• Observes how responders type on their keyboard when answering verbatim questions
• Checks for illogical responses, e.g., 22 years old with a high school education with income over $200,000
• Observe patterns in responses
Identifies duplicates without PII
• Scrubbing uses survey responses along with interview duration and machine characteristics for identification of duplicates
• No PII is used in the process because
• Responders can game the system by joining panel from multiple sources or be part of multiple panels
• Surveys where sample comes from multiple panel providers, this method of duplicate identification is extremely valuable
• Unlike machine based duplicate response identification a.k.a. RelevantID
Why is scrubbing important?
Garbage In Garbage Out
• Uncleaned data sets impact results
• Bad responders can swing results wildely
Improves client confidence
• Helps in instilling greater confidence in clients on data
• Improves brand perception with regards to quality product
• Competitive differentiator in market
Data Scrubbing Techniques
• Straightliners/Flatliner
• Bad Verbatims
• Speeders
• Inattentive
• Duplicates
• Bots
• Geolocation
• Illogical answers
• Maxdiff/Conjoint Satisficing
• Video/Audio usage
• Questionnaire Review
• Verbatim -Contextual
Competitor differentiation
Scrubbing process
Scrubbing is done parallel to fielding, to ensure there is enough time during
fielding to collect replacement sample.
Round 1: Processed once 60% of data has been collected.
Round 2: Processed at 90-95% data collection mark.
Round 3 or more: Processed at field closure or based on client request.
Turnaround times
95% of projects are turned around within 8-10 hours
• Projects received before 3pm eastern are turned around the same day, albeit, later in the evening
• Projects received after 3pm eastern and turned around before 9am eastern the following day
Expedited requests
• Urgent requests can be turned around faster
• Projects received after 3pm can still be turned around the same day
What impacts scrubbing cost?
Number of completes
• Under 5,000 / Over 5,000
Number of files to be scrubbed
• Each data file is processed individually
• Multiple files can constitute multiple rounds
Number of rounds
• Each scrubbing project cost includes up to 2 rounds, additional rounds are charged
Number of non-English languages
• Verbatim responses in non-English languages are translated via Google
• Each non-English language has an additional charge
• If no verbatims in the survey, no additional charge for non-English
What’s needed to get started?
Questionnaire
Word or Excel
Data file
• SPSS is the preferred format
• Excel files lead to longer turnaround time and higher costing
• Multiple files can constitute multiple rounds
Email request to scrubbing@b3intelligence.com
Case Study
Financial Services Survey across 38 countries
Round 1: Country X went through first round of scrubbing with 320 completes. All regular checks were performed, and output was within standard removal rates barring two Issue 1 – IP Address: 96% of all records came in from Country Y
One specific ISP was used to take the survey over 300 times, with 70% of completes coming from 8 main IP blocks Issue 2 – Verbatim: None of the suspect records had any responses to verbatim questions
b³ Volume & Capacity
b³ processes 420-440 scrubbing projects per month
• An average project goes through 2 rounds
• 750 to 800 files are scrubbed per month
• Team of 12 analysts
b³ can processes 550-600 scrubbing projects per month.
• At current capacity
• Scalable to a higher level due to automation and staffing
Contact
Toronto – Corporate Headquarters
100 Sheppard Ave East. Suite 503
Toronto, Ontario M2N 6N5
+1 (416) 549-8000
+1 (888) 224-6198
info@b3intelligence.com
Copyright © 2023 b3Intelligence, All Rights Reserved | Disclaimer | Privacy Policy