60 Atlantic Avenue, Toronto

CAWI: Computer-Assisted Web Interviewing

The Complete Guide to Online Survey Research

What is CAWI?

CAWI emerged in the late 1990s as internet penetration grew, initially supplementing telephone and mail surveys. By the 2010s, CAWI had become the primary mode for many commercial and academic surveys, particularly in developed countries where 80-95% of populations have internet access. The COVID-19 pandemic accelerated CAWI adoption as face-to-face and telephone surveys became impractical.

Key Characteristics

Self-Administration

No interviewer present. Respondents read questions and select answers independently, reducing interviewer effects and social desirability bias for sensitive topics.

Respondent Pacing

Participants complete surveys at their own pace, pausing and resuming as needed. Average completion time: 15-25 minutes optimal, up to 40 minutes acceptable for engaged audiences.

Multi-Device Access

Responsive design allows completion on desktops, tablets, or smartphones. Mobile traffic now represents 50-70% of CAWI responses in most populations.

Rapid Deployment

Launch surveys in hours, collect data in days. Real-time monitoring allows immediate adjustments to quotas, question wording, or targeting.

CAWI Sample Sources

CAWI surveys draw respondents from three primary sources:

Online Panels

Pre-recruited pools of respondents who agreed to periodic surveys.

Examples: Opinion.Tips (PollZapper's federated network), YouGov, Dynata, Prolific

Email Lists

Existing contacts from customer databases, membership lists, or event registrations.

Examples: Customer satisfaction surveys, employee engagement surveys, alumni surveys

River Sampling

Web traffic intercepts via banner ads, social media, or website pop-ups.

Examples: Display ads on news sites, Facebook ads, website exit surveys

When to Use CAWI

CAWI is optimal when speed, cost, and geographic reach are priorities, and when the target population has reliable internet access.

Ideal Use Cases

Large-Scale National Polls

When representative national samples (n=1,000+) are needed quickly:

  • Political tracking polls: Weekly or daily snapshots of public opinion
  • Brand tracking: Monitoring awareness and perception over time
  • Market research: Testing product concepts with quota samples

Cost advantage: CAWI costs $2-$10 per complete vs $15-$50 for CATI and $50-$200 for CAPI.

Surveys Requiring Visual Stimuli

Web surveys excel at displaying images, videos, and interactive content:

  • Ad testing: Show video commercials or print ads
  • Package design: Display product mockups for preference testing
  • Website usability: Show screenshots for feedback
  • Conjoint analysis: Present feature combinations with images
Sensitive Topics

Self-administration reduces social desirability bias for sensitive questions:

  • Health behaviors: Drug use, sexual behavior, mental health
  • Financial information: Income, debt, bankruptcy
  • Illegal activities: Tax evasion, piracy, undocumented work
  • Controversial opinions: Political views in polarized environments

Research shows respondents report 15-30% higher rates of sensitive behaviors in CAWI vs CATI or CAPI.

Customer Feedback & Satisfaction

Businesses with customer email lists benefit from fast, inexpensive CAWI:

  • Post-purchase surveys: Satisfaction and NPS measurement
  • Service evaluation: Support interaction feedback
  • Employee engagement: Anonymous workplace surveys
  • Event feedback: Conference or webinar evaluations

When to Avoid CAWI

Consider Alternatives When:
  • Low internet penetration: Rural or elderly populations may lack access
  • Need for representative samples: Without probability panels, weighting may not correct coverage error
  • Complex or long surveys: Respondents abandon surveys >25 minutes
  • Low literacy populations: Self-administration requires reading comprehension
  • Need for personal interaction: Rapport building or clarification requires CAPI/CATI

Online Panel Management

Building and maintaining a high-quality online panel is central to CAWI success. Panel quality directly determines data quality.

Panel Recruitment Strategies

Gold standard for representative panels. Randomly sample households via RDD, ABS, or voter files, then invite to join panel.

Examples:
  • Pew Research Center: Address-based sampling with mail invitations
  • NORC AmeriSpeak: RDD sampling with $10 cash incentive
  • GfK KnowledgePanel: Provides tablets to non-internet households
Advantages:
  • Known selection probabilities enable weighting
  • Covers non-internet populations (if providing access)
  • Academically rigorous
Challenges:
  • Expensive ($50-$200 per recruited panelist)
  • Low recruitment rates (5-15%)
  • Requires large sample frames

Most common approach. Recruit volunteers via banner ads, social media, email campaigns, and partner websites.

Recruitment Channels:
  • Display advertising: Banner ads on news, content sites
  • Social media: Facebook, Instagram, LinkedIn ads
  • Search advertising: Google Ads targeting "paid surveys"
  • Co-registration: Add checkbox to partner signup forms
  • Referrals: Existing panelists invite friends
Incentive Structures:
  • Points systems: Earn points convertible to gift cards ($0.50-$2 per survey)
  • Sweepstakes: Entry into monthly $500 drawing
  • Charity donations: $1 donated per survey
  • Professional networking: B2B panels offer industry insights

PollZapper Opinion.Tips Network: Federated panel allowing researchers to recruit from pre-profiled respondents across partner sites.

Organization-specific panels recruited from customer or member databases.

Use Cases:
  • Retail: Loyalty program members for product testing
  • Financial services: Customer advisory panels
  • Healthcare: Patient advisory councils
  • Non-profits: Member feedback panels
Advantages:
  • High engagement (existing relationship)
  • Low recruitment cost (existing email list)
  • Domain expertise/experience
Challenges:
  • Limited to existing customers
  • Potential bias (satisfied customers over-represented)
  • Panel fatigue if over-surveyed

Panel Maintenance & Quality Control

Profiling Surveys

Collect demographic and psychographic data for targeting:

  • Age, gender, education, income
  • Geographic location (ZIP/postal code)
  • Household composition
  • Employment status and occupation
  • Political affiliation, media habits
Engagement Tracking

Monitor panelist participation:

  • Response rate: Invitations vs completions
  • Survey frequency: Limit to 1-4 per month
  • Quality scores: Track speeding, straight-lining
  • Attrition monitoring: Identify inactive panelists
Panel Fatigue Prevention

Prevent over-surveying:

  • Frequency caps: Max 4 invitations per month
  • Length limits: Keep surveys <20 minutes
  • Relevant targeting: Only invite qualified respondents
  • Rest periods: 48-72 hours between surveys
Fraud Detection

Identify and remove fraudulent panelists:

  • Duplicate accounts: Same IP, email patterns
  • Bot detection: Suspicious behavior patterns
  • Professional survey takers: High frequency across panels
  • Inconsistent profiling: Age changes, contradictions

Response Rate Optimization

Typical CAWI Response Rates: 5-15% for opt-in panels, 20-40% for customer panels, 40-60% for employee surveys. Every percentage point improvement reduces cost per complete.

Email Invitation Best Practices

Subject Line Optimization

Subject lines are the most important factor in email open rates (15-30% typical).

Approach Example Performance
Personalization "John, we need your opinion" +25% open rate
Time scarcity "Last chance: Survey closes tonight" +20% open rate
Incentive mention "$5 Amazon gift card for 10 minutes" +15% open rate
Generic "Please complete our survey" Baseline
All caps/spam triggers "FREE MONEY!!! Click NOW!!!" Spam filtered
Survey Length & Burden

Completion rate drops sharply after 20 minutes:

  • 5-10 minutes: 70-80% completion rate
  • 10-15 minutes: 60-70% completion rate
  • 15-20 minutes: 50-60% completion rate
  • 20-25 minutes: 40-50% completion rate
  • 25+ minutes: <40% completion rate
Length Reduction Strategies:
  • Remove "nice to have" questions
  • Simplify matrix questions (max 5-7 items)
  • Use skip logic to personalize (PollZapper Virtuoso)
  • Split long surveys into multiple waves
Progress Indicators

Show respondents how much remains:

  • Percent complete: "You're 60% done!"
  • Page numbers: "Page 5 of 8"
  • Visual bars: Filled progress bar
  • Time remaining: "About 5 minutes left"

A/B tests show progress indicators increase completion rates by 8-15%.

Reminder Sequences

Optimal reminder schedule:

  1. Day 0: Initial invitation
  2. Day 3: First reminder (50% of responses come from this)
  3. Day 7: Second reminder
  4. Day 10: Final reminder ("Last chance!")

Reminder messaging:

  • Emphasize scarcity ("Only 3 days left")
  • Social proof ("500 people have already shared their views")
  • Simplify CTA ("Click to start - just 10 minutes")

Mobile Optimization

50-70% of CAWI responses now come from mobile devices. Mobile-unfriendly surveys have 30-50% lower completion rates.

Mobile-Friendly Design
  • Vertical scrolling (no horizontal)
  • Large touch targets (44x44px minimum)
  • Single column layout
  • Thumb-friendly buttons at bottom
  • Appropriate input types (numeric keyboard for numbers)
  • Readable font size (16px minimum)
  • Break long matrices into single questions
Mobile Killers
  • Wide grid/matrix questions requiring zooming
  • Dropdowns with 50+ options
  • Small clickable areas
  • Excessive open-ended questions (typing on phone)
  • Auto-advancing too fast (elderly users)
  • File uploads (difficult on phones)
  • Drag-and-drop ranking

Quality Assurance & Fraud Detection

CAWI's low cost and self-administration make it vulnerable to low-quality responses and fraudulent completions. Robust quality controls are essential.

Common Quality Issues

Speeders

Definition: Respondents completing too quickly to have read questions carefully.

Detection:

  • Set median interview duration during pilot
  • Flag completions <50% of median
  • Track time per question (not just total)

Prevention:

  • Minimum page duration (e.g., 3 seconds per question)
  • Warn respondents: "Please read carefully"
  • Use attention checks (see below)
Straight-Liners

Definition: Selecting same response across all items in a grid/battery.

Detection:

  • Calculate variance across grid questions
  • Flag zero-variance respondents
  • Check for runs (8+ consecutive same answers)

Prevention:

  • Limit grid length (max 5-7 items)
  • Randomize item order within grids
  • Insert reverse-coded items
  • Break long grids into separate questions
Bots & Fraud

Automated bots attempt to complete surveys for incentives.

Detection:

  • reCAPTCHA v3 (invisible bot detection)
  • Device fingerprinting
  • Impossible response patterns
  • IP address analysis (VPNs, click farms)

Prevention:

  • Require unique email verification
  • Block VPN/proxy IP addresses
  • Limit completions per IP (1-2 max)
  • Honeypot questions (hidden fields bots fill)
Professional Survey Takers

Individuals who complete many surveys for income.

Detection:

  • Cross-panel participation tracking
  • High survey frequency (>10 per month)
  • Identical open-ended responses across surveys

Mitigation:

  • Frequency caps (max 4 surveys/month)
  • Quality score thresholds
  • Lower incentives to reduce mercenary respondents

Attention Check Strategies

Types of Attention Checks
Instructional Manipulation

"Please select 'Strongly Agree' for this question."

Fails ~10-15% of inattentive respondents

Logical Consistency

"What year were you born?" then later "How old are you?"

Catches contradictory answers

Impossible Answers

"I have visited the planet Mars" (Yes/No)

Flags respondents clicking randomly

Best Practice: Use 2-3 attention checks per survey. Failing 2+ is typical threshold for exclusion.

CAWI in PollZapper

PollZapper's Professional CAWI Platform

PollZapper provides enterprise CAWI infrastructure including Opinion.Tips federated panel network, mobile-responsive surveys with Virtuoso branching, real-time quota monitoring, anonymous polling architecture, and comprehensive fraud detection.

Opinion.Tips Panel Network

Federated panel architecture connecting researchers with pre-profiled respondents across partner sites.

How It Works:
  1. Define target demographics and quotas
  2. PollZapper sends invitations to qualified panelists
  3. Respondents complete survey in PollZapper
  4. Automatic quota management and de-duplication
Panel Quality:
  • Profiled demographics on 50+ variables
  • Fraud detection and bot filtering
  • Quality scoring based on survey behavior
  • Frequency capping to prevent panel fatigue
Mobile-First Design

Responsive surveys optimized for all devices with automatic mobile adaptations.

Automatic Optimizations:
  • Grid questions collapse to single questions on mobile
  • Large touch targets (44x44px) for all buttons
  • Appropriate input keyboards (numeric, date, email)
  • Vertical scrolling only, no horizontal overflow
  • Readable 16px+ font sizes
Designer Preview:
  • See exactly how surveys render on phones, tablets, desktops
  • Test navigation and timing
  • Identify mobile usability issues before launch
Real-Time Quota Monitoring

Live dashboard shows quota progress and automatically closes cells when filled.

Quota Types:
  • Simple quotas: Age, gender, region
  • Interlocking quotas: Age x Gender x Region
  • Nested quotas: National with regional sub-quotas
  • Dynamic quotas: Adjust mid-field based on completion rates
Auto-Close:
  • Quota cells close automatically when filled
  • Respondents screen out at start (not after completing)
  • Admin alerts when quotas near completion
Anonymous Polling Architecture

Privacy by design: Response data separated from identifying information.

Data Isolation:
  • Survey responses stored without email/name
  • One-way hashing prevents re-identification
  • Optional "blind field operations" (canvassers don't see sponsor)
Compliance:
  • GDPR-ready architecture
  • Configurable data retention (30-365 days)
  • Export for legal/audit requirements
  • Respondent data deletion on request
Fraud Detection Suite

Multi-layer protection against bots, speeders, and low-quality responses.

Detection Methods:
  • reCAPTCHA v3: Invisible bot detection
  • Device fingerprinting: Detect duplicate attempts
  • IP analysis: Flag VPNs, data centers, click farms
  • Speed detection: Flag completions <50% median
  • Pattern analysis: Straight-lining, variance checks
  • Attention checks: Instructional manipulation embedded
Admin Actions:
  • View flagged responses with scores
  • Approve or reject individual completions
  • Export clean data with fraud flags removed
Virtuoso Branching Logic

Advanced skip logic personalizes surveys based on previous answers.

Capabilities:
  • Conditional display (if/then logic)
  • Piping (insert previous answers into questions)
  • Loop iterations (repeat questions for each item)
  • Quota branching (different paths by quota cell)
  • A/B testing (randomize question order/wording)
Benefits:
  • Shorter surveys (skip irrelevant questions)
  • Higher completion rates
  • Better respondent experience
  • More accurate data

CAWI Best Practices

With 50-70% mobile traffic, design for phones and enhance for desktop:

  • Test entire survey on smartphone before launch
  • Avoid grids wider than 3 columns
  • Use dropdowns sparingly (scrolling lists are tedious)
  • Break long questions into multiple pages
  • Use appropriate input types (numeric keyboard, date picker)

PollZapper: Preview mode shows exactly how survey renders on mobile devices.

Email open and response rates vary by day/time:

  • Best days: Tuesday-Thursday (Monday = email overload, Friday = weekend mindset)
  • Best times: 8-10am or 1-3pm local time
  • Avoid weekends: Unless targeting specific populations
  • Account for time zones: Schedule regionally

A/B test different send times to optimize for your specific audience.

Motivate completion throughout the survey:

  • Progress bar: Visual indication of completion
  • Milestone messages: "You're halfway there!"
  • End reminder: "Just 3 more questions"
  • Thank you page: Reinforce incentive delivery

Question and response order affects answers:

  • Randomize question blocks: Reduce primacy/recency bias
  • Randomize response options: Except ordered scales (Strongly Disagree to Strongly Agree)
  • Rotate stimuli: Show ads, concepts in random order
  • A/B test wording: Compare alternative phrasings

PollZapper: Built-in randomization for questions, options, and A/B testing.

Without an interviewer to clarify, questions must be unambiguous:

  • Avoid double-barreled questions: Not "Do you like apples and oranges?"
  • Use simple language: 8th grade reading level maximum
  • Define terms: What does "regularly" mean? Clarify.
  • Avoid negatives: "not disagree" is confusing
  • Be specific: "Last 7 days" not "recently"

CAWI's real-time nature allows mid-field adjustments:

  • Track response rate: If <5% after 24 hours, adjust invitations
  • Monitor completion rate: If <50%, survey may be too long
  • Review drop-off points: High abandonment at specific question? Fix it.
  • Check quota balance: Some demographics over-represented? Adjust targeting.
  • Identify problem questions: High skip rate? Reword or make optional.

PollZapper: Real-time dashboard with response rate, completion rate, quota progress, and question-level analytics.

Establish data quality criteria and apply systematically:

  • Remove speeders: <50% median duration
  • Remove straight-liners: Zero variance on grids
  • Remove attention check failures: 2+ failures
  • Remove duplicate IPs: More than 2 responses per IP
  • Document exclusions: Report n before and after cleaning

Typical cleaning removes 10-20% of responses in opt-in panels, 5-10% in customer panels.

Opt-in panels require weighting to approximate representative samples:

  • Identify benchmarks: Use census data for demographics
  • Weight key variables: Age, gender, education, region
  • Use raking: Iteratively adjust weights to match multiple benchmarks
  • Trim weights: Cap at 3-5x to avoid excessive influence
  • Report design effect: Weighting increases margin of error

PollZapper: Built-in weighting calculator with raking algorithm and automatic design effect calculation.

Ready to Launch Professional CAWI Research?

PollZapper's CAWI platform includes Opinion.Tips panel access, mobile-responsive surveys, Virtuoso branching, real-time quota monitoring, fraud detection, and anonymous polling architecture. From 100-respondent customer surveys to 5,000-respondent national polls.

Join the Waitlist Explore More Resources

©2024 - PollZapper.com. All Rights Reserved.