Data Quality Management
Divergent Insights employs a three-tiered approach to quality control, focusing on panel quality, data quality, and anti-fraud measures to uphold our high standards and ensure the integrity of our online research platform.
1. Panel Quality Management:
Ensuring the reliability of survey results is paramount, and we manage panel quality through meticulous screening processes. Our suppliers conduct screening exercises, and eligible respondents undergo validation checks for email addresses, and contact details. We employ various methods, such as phone calls, LinkedIn profile checks, and company website directory checks, to verify panellists’ authenticity.
- We update members, attribute information annually through attribute acquisition surveys to maintain accurate survey results.
- Duplication of panellists is proactively managed using cookies, IP, Operating system, and email addresses etc…in total 45+ attributes are checked to prevent response bias.
- To avoid response bias, we diversify our surveys by including members with a broad range of attributes, including gender, age, and place of residence
2. Data Quality:
Divergent Insights ensures unbiased and representative survey results through a robust data quality management process. Our commitment to maintaining high-quality standards involves technology-based quality control and rigorous data checking performed by our skilled research operators. With our Text Analysis technology, we enhance the screening process for respondent inputs in open-ended questions. The technology effectively identifies and filters out gibberish responses and nonsensical answers.
- Incorporating Pre-screener Red Herring Questions – These serve as a strategic layer aimed at identifying and filtering out bots and VPN traffic.
- Managing survey logic to prevent contradictory responses.
- Standardizing workflow to ensure service quality uniformity.
- Utilizing advanced techniques such as cookies and email address analysis, we effectively identify and manage instances of duplicated respondents.
- Managing survey logic to detect fraudulent respondents.
- Real-time confirmation of response status and data collection
- Cleaning raw data
- Eliminating inconsistent responses, speedsters, straight liners, duplicate response, and obviously fraudulent open-ended responses
- Systematically detecting and excluding fraudulent data through aggregation
3. Anti-Fraud Measures
We employ rigorous measures to detect inconsistencies in survey response data, including straight-lining, bot respondents, and speedsters with extremely short response times and the one who takes longer time. Respondents meeting fraud standards are excluded from receiving further surveys.
Periodic quality check surveys are conducted to identify untruthful or fraudulent responses.
DFPT (Digital Fingerprinting): Digital Fingerprinting, also known as Device Fingerprinting, Device ID, or Machine ID, highlighted as a crucial component of our quality strategy. This technology allows us to uniquely identify devices used by respondents, enabling us to detect and mitigate fraudulent activities effectively.
Data Validation Process
BEHAVIOUR AT PRE-SCREENING
An initial respondent data quality assessment evaluates attention levels and verifies demographic information.
BEHAVIOUR AT TIME OF SURVEY
Respondents displaying inconsistent or fraudulent behavior will be categorized as such.
BEHAVIOUR DURING SURVEY
If the length of interview (LOI) significantly differs from the creator’s specified time, the respondent may be subject to categorization.
STATUS ON SURVEY COMPLETION
If a survey is concluded without activating any security checks, the respondent’s answers can be deemed reliable.
Utilizing IP tracking, completion rates, and termination rates contributes to forming a more comprehensive understanding of respondent quality.