SCSRU Newsletter Vol. 22, Issue #2 – Fall 2024

Wednesday, December 11, 2024
by Brandon Yong

Balancing Reach and Quality: The Bot Problem in Social Media Survey Recruitment

As discussed in our previous newsletter, differentiating between genuine online survey participants and "bots" has become increasingly challenging. Bots, which are computer programs designed to automate tasks, have grown more sophisticated in recent years.  With the rise of AI tools becoming more accessible, they now pose an even greater threat to the integrity of survey data. Additionally, imposter participants—individuals who intentionally fake their identities or exaggerate their experiences—have become more prevalent. Both bots and imposter participants are considered fraudulent responses, which significantly compromises data quality and skews results.

To combat this growing issue, the SCSRU has implemented several measures to detect and block bots. These include the use of CAPTCHAs, mandatory attention-checking questions, and the analysis of metadata such as duplicate IP addresses, location, and survey completion times. However, as bots become more sophisticated, they are increasingly able to bypass these measures. They avoid detection by using dynamic IP addresses, not following predictable patterns in completion times, and even generating AI-written responses for open-ended questions, making them harder to spot with traditional methods.

Given these challenges, the SCSRU recommends taking additional steps to identify and prevent bot or imposter participation. One useful strategy is to introduce questions that catch contradictions—such as asking similar questions in different ways to ensure consistent answers. In addition, there are software tools available that can detect AI-generated text, though they require a sufficient amount of text to work effectively. For high-stakes research, one approach is to verify respondent identities through follow-up phone calls or video chats, especially if survey incentives are offered as part of recruitment.  However, caution is advised, as some individuals may use filters or other techniques to impersonate others during these verification processes.

The best way to minimize bot interference is to avoid advertising surveys on social media, as bots frequently target these platforms. Even without incentives, bots will often target surveys across online communities. If social media advertising is necessary, it’s advisable to use closed groups whenever possible, though they may not be foolproof. If you need additional guidance or best practices to prevent and detect bots and fraudulent responses, please don’t hesitate to reach out to our team for expert advice.

Back to top


Partnership with IAP & AQuE

The SCSRU is partnering with Institutional Analysis & Planning (IAP) and the Academic Quality Enhancement Office (AQuE) to administer an Academic Program Review survey for eligible programs. The survey will be open to all alumni of the programs under review (all graduates within the last 7 years). UW’s program reviews are generally scheduled to take place every seven years for most programs, with the objective of helping each program achieve and maintain the highest possible standards of academic excellence by systematically reflecting on its strengths and weaknesses and determining what actions would further enhance the quality of the program. The survey will launch in winter of 2025 and will be conducted on an annual basis, with different programs opting in on a rotating schedule. This cycle of program reviews will include the following programs:

  • Global Business and Digital Arts (BGBDA)
  • Digital Experience Innovation (MDEI)
  • Kinesiology
  • Earth Science
  • Liberal Studies
  • French
  • Architectural Engineering
  • Environmental Sciences
  • Accounting
  • Theological Studies

Back to top


Interested in Affordable Research?

Share survey research costs and collect high-quality data. The Statistical Consulting and Survey Research Unit (SCSRU) conducts an annual survey of the Waterloo Region called the Waterloo Region Matters Survey where interested researchers and organizations can collect data for a fraction of the cost of an independent telephone or web study. This unique opportunity is of particular benefit to researchers and organizations with limited budgets who want to test research concepts such as public opinions on specific issues, measure awareness and usage of services or products, or refine survey questions for other research. 

Key Details:

Study population: Adult residents (18 and older) within the Waterloo Region, including Kitchener, Waterloo, Cambridge and the four townships.

Sample size: 500 completed surveys, representative by region/municipality and age.

Methodology: Random-digit dialed (RDD) telephone (using both landline and cell phone telephone numbers), recruit to web address-based sampling (ABS) and web surveys. RDD sampling of households within the Waterloo Region population and random selection within the household are used in initial recruitment by the Statistical Consulting and Survey Research Unit. This approach, along with ABS which uses postal addresses to reach households based on demographic data in different postal code areas, more accurately represents the Waterloo Region compared to non-probability sampling methods, such as recruiting members from web panel firms or visitors to an organization website. The sample used for the web component of the Waterloo Region Matters Survey was originally recruited using RDD. Having a random selection of respondents and data that are representative of the population are important aspects of data relevance and utility. The addition of ABS recruitment was added in 2024 to better target young respondents, who are difficult to recruit.  This combination of sampling methods ensures that households meeting the diverse characteristics of the region can accurately be targeted and recruited to the study.

Next iteration of data collection: Spring 2025

Deadline to express interest: January 24, 2025

Cost: For more information about the Waterloo Region Matters survey, please visit the Waterloo Region Matters web page, contact Reem Tawfik or call 519 888-4567 ext. 35071.

Can't wait until the next Waterloo Region Matters survey? Consider using the Waterloo Region Matters Panel. The SCSRU maintains a small probability-based web survey panel that is fielded in conjunction with the Waterloo Region Matters Survey. For more immediate projects that do not require large sample sizes of Waterloo Region residents (e.g. n=200 respondents or fewer), a web survey can be fielded at any time.

Back to top


Annoucements

We've moved!

Our office has moved from Hagey Boulevard back onto main campus. The new call centre is in Mathematics 3 and our offices are in the Mathematics and Computing building.

Dr. Geoffrey Fong named a Canadian Medical Hall of Fame inductee

Congratulation Dr. Fong for being named a Canadian Medical Hall of Fame inductee! Dr. Fong is the founder and Chief Principal Investigator of the International Tobacco Control Policy Evaluation Project, which the SCSRU has helped administer for several years. Dr. Fong will join the hall of fame at an induction ceremony in June 2025. Well done!

Personnel changes

Project Manager Tony Ly recently accepted a position in the Office of the Vice-President, Academic and Provost as Associate Director, Strategic Initiatives. The SCSRU wishes him success in his new position and welcomes Reem Tawfik as one of our new project managers.

Back to top


Upcoming Events

The SCSRU has several free workshops planned for students, faculty, and staff this upcoming winter 2025 term. Be the first to know when workshop registration opens by signing up for our workshop mailing list.

Here's a peak at our Winter 2025 workshop lineup!

Unleash the Power in Your Data

January 22, 2025

Data often come in the form of tables or databases, making it challenging to understand beyond their sizes and variable types. This workshop will introduce you to graphical and numerical techniques to uncover the structure of your data

Unlock the Power of Linear Regression

February 19, 2025

Linear regression models, also known as linear models, are essential tools in various fields due to their simplicity and ease of interpretation. This workshop offers a comprehensive introduction to both simple and multiple linear regression

Explore Your Count Data

March 19, 2025

Count data, which consists of non-negative integers, is commonly collected and analyzed in fields like epidemiology, genomics, and clinical research. This workshop will introduce you to Poisson regression, a powerful tool for modeling count data over time.

Back to top