Spoofing technology and the potential for skewed election polls and survey results

It seems like there have been more inaccuracies in survey results in recent years, most notably in American election polls. What is going on?

The United States is currently undergoing a boom in the construction of data centers. Data centers, among other things, enable spoofing technology such as virtual private servers with residential IP addresses, VOIP phone numbers, and virtual workstations. Would non-Americans living outside of the United States be throwing off survey data via increased use of this technology?

Here is what Google Gemini, itself a beneficiary of increased data center capacity, had to say (blockquote below the line, unindented).

TL;DR: It’s probably not much of a factor. Also, getting good survey data is difficult.

---

It’s unlikely that non-Americans using US-based residential VPNs, VOIP numbers, or virtual workstations are the primary driver of the inaccuracies seen in recent American survey data, such as election polling.

While these tools can be used to spoof a US identity and are a genuine concern for survey integrity, the consensus among polling experts is that the main sources of error are due to traditional methodological challenges in an increasingly difficult data collection environment.

Major Sources of Survey Error

The bulk of survey inaccuracies are attributed to well-known statistical issues, not necessarily foreign fraud:

• Nonresponse Bias: This is perhaps the biggest issue. The people who refuse to participate in surveys often differ systematically from those who do participate. For instance, people with lower trust in institutions or who are harder to reach (like younger, highly mobile, or lower-income individuals) may be underrepresented, skewing the results. This is especially pronounced in political polling.
• Sampling Error and Weighting: Even with a representative sample frame, the way data is weighted to match population demographics (age, education, race, etc.) can be flawed. If pollsters fail to accurately estimate or weight by a critical variable—like education level (which has become a strong predictor of political alignment)—the final result will be off.
• Measurement Error: The questions themselves may be biased, confusing, or poorly worded, leading respondents to provide inaccurate information, either intentionally (e.g., social desirability bias) or unintentionally.
• The Shift to Online Polling: As landline phone participation declines, surveys increasingly rely on online opt-in panels. These panels can suffer from selection bias because the participants are self-selected and may not be representative of the general population, even with screening questions.

The Role of Spoofing Technology

The technology you mentioned—residential VPNs, VOIP, and virtual workstations—is designed to make non-US users appear as legitimate US residents, making it harder to detect them.

• Residential Proxies/VPNs: These route traffic through real, legitimate home internet connections, giving the appearance of a genuine US resident. They are significantly harder to detect than traditional data center VPNs.
• VOIP Numbers and Virtual Workstations: These tools provide a convincing US digital footprint (US phone number, US IP address) that can bypass basic geographical filtering in online surveys.

Why This Is More of a Fraud/Panel Quality Issue

While this activity can inject non-Americans into US surveys, it is generally considered a challenge in panel quality control rather than the primary cause of large-scale polling errors.

• Survey Controls: High-quality survey firms and panel providers use sophisticated fraud detection software that goes beyond simple IP address checks. They analyze behavioral patterns (speed of response, consistency of answers), device fingerprinting, and account history to flag and remove suspicious respondents, including those using residential proxies.
• Motivation: The primary motivation for mass foreign spoofing in surveys is often financial fraud (collecting small payments for survey completion) or ad fraud, not necessarily a targeted, large-scale political disinformation campaign aimed at skewing results. While the latter is possible, the sheer volume of fraudulent respondents needed to swing a major election poll by several points is substantial and highly susceptible to detection.
• Limited Scale: While the sophisticated tools make it difficult to detect, the number of fraudulent responses that successfully slip past multiple layers of quality control in a large-scale, professional survey is typically very small (often near 0-5% of a well-managed panel). This is usually not enough to account for the large 5-7 point misses seen in some major election polls.

In summary, the sophisticated tools can certainly contribute to data contamination, making the pollster’s job harder, but the historical and ongoing problems of nonresponse and flawed weighting are still considered the dominant explanations for major polling failures.

---

Contact Charadata