How priming bias can influence survey results
Editor’s note: Seb Martin is U.K. managing director at Bryter, an insights agency based in London and New York. He has over 20 years of experience in the market research industry and has written and published numerous reports and papers on related to the tech sector. Find Martin on LinkedIn.
For market researchers, few risks are as pervasive and underestimated as priming bias. Subtle shifts in how questions are ordered or framed can meaningfully alter outcomes, often without researchers or clients realizing.
In a recent Bryter study (registration required) exploring attitudes toward smartphones for children, we deliberately designed part of the survey to test one specific manifestation of the priming effect. The results provide a real-world illustration of how easily quantitative findings can be skewed, and what researchers can do to mitigate this.
What is the priming effect, and why does it matter?
The priming effect refers to the psychological phenomenon whereby exposure to one stimulus influences how individuals respond to subsequent stimuli.
In quantitative research, priming most commonly appears as:
- Question order bias: Earlier questions influence responses to later ones.
- Context effects: The thematic framing of a survey shapes how respondents interpret subsequent questions.
- Emotional activation: Exposure to positively or negatively framed content shifts respondents' mind-sets.
Priming in surveys can have a number of negative consequences including inflated or suppressed purchase intent, artificially polarized attitudes, overstatement about behavior change or artificial inflation of the importance of a subject that consumers really aren’t thinking that hard about.
Critically, priming often doesn’t feel like bias. Responses remain internally consistent but are anchored in a temporarily activated mind-set rather than stable underlying beliefs.
Priming is not limited to quantitative research and can occur across all forms of market research and study design. Even the location of where a study is held can influence expectations and outcomes from the research. If you were conducting research on wine, hosting the taste test in a fancy restaurant could yield very different results than if that same test was conducted in a church hall.
For the purposes of this article, I’ll be focusing on one aspect of priming and how it can influence quantitative survey results.
Research context: smartphones, parents and conflicting attitudes
The Bryter study surveyed 1,823 adults of parenting age across the U.K., Germany and Italy to understand attitudes toward smartphone ownership and usage among children.
The broader findings revealed a clear tension: A total of 63% of parents believe smartphones should be banned for children under 11. However, 61% believe the benefits of smartphones outweigh the negatives. And smartphone rather than feature phone ownership dominants across all age groups, including for young children under the age of 11.
These tensions, and apparently contradictory consumer behaviors, created an ideal context to explore how the priming effect might influence stated purchase intentions.
Experimental design: Isolating a priming effect, the questions order bias
To test the impact of question order bias we introduced a simple experimental split, with respondents following one of two survey paths:
- Primed group
- This group answered attitudinal questions about smartphones first. Questions asked about perceived benefits but also concerns related to smartphone usage covering addiction, impact on sleep, mental health, etc.
- Respondents were then asked purchase intent questions regarding next choice of phone for themselves and also their children. Would they be buying a smartphone or a feature phone?
- Unprimed group
- This group answered purchase intention questions first.
- The attitudinal questions followed later in the survey.
This design isolated one key variable: Does thinking about the impact of smartphone usage change what type of phone parents say they will buy their kids next?
Results: A clear and measurable priming impact
The impact was immediate and significant. Let’s look at the purchase intent for next device for children under 11:
- Primed respondents
- Fifty-nine percent intend to buy a smartphone.
- Forty-one percent intend to buy a feature phone.
- Unprimed respondents
- Seventy-one percent intend to buy a smartphone.
- Twenty-nine percent intend to buy a feature phone.
What this shows
Exposure to attitudinal questions about risks associated with smartphone use increased feature phone purchase intent by 12 percentage points. The unprimed responses aligned more closely with current device ownership statistics.
In other words, the act of asking about concerns reshaped future intentions. Parents genuinely hold concerns about smartphones, but these concerns are often latent rather than decision driving. When prompted, they became more salient. When not prompted, decisions revert to pragmatic norms and market reality.
Implications for market research practice
This case illustrates how easily well-intentioned survey design can introduce systematic bias and how a single instance of question order change can shift results by double digit percentages. For insight teams advising on product strategy, innovation or forecasting, that margin of error is material.
For practitioners, the lessons are clear:
1. Question order is not neutral. Seemingly logical survey flows can:
- Inflate rationalization effects.
- Overstate the role of beliefs in decision-making.
- Distort predictive validity.
2. Attitudinal questioning can act as an intervention. In this study, attitudinal questions didn’t just measure opinion they changed it, in the moment. This is particularly risky when exploring:
- Social responsibility topics.
- Health and well-being.
- Ethical consumption.
- Parenting decisions.
3. Behavioral questions should be isolated where possible. If the objective is to predict behavior:
- Ask behavioral or choice questions first.
- Try to capture more instinctive, System 1 responses.
- Avoid contaminating them with reflective or moral frames.
The solution is not to eliminate attitudes from research, but to be mindful of the subject matter and how survey design is structured in order to minimize the potentially misleading influence of the priming effect.