Improving questionnaire translations and translation processes

1
Improving questionnaire translations and
translation processes

• Barbara Forsyth, Martha Kudela and Kerry Levin
• Westat
• Deirdre Lawrence and Gordon Willis
• National Cancer Institute

2
Overview

• Tobacco use items
• 5-step translation model
• Review steps
• Survey Language Consultant role
• Implementation
• Lessons learned
• Conclusions

3
Tobacco use items

• Patterns of use
• Smoking prevalence
• Workplace smoking policies
• Nicotine dependence
• Advice about quitting
• Attempts to quit and methods used
• Changes in smoking attitudes

4
5-Step model

• Translation
• Review
• Initial adjudication
• Cognitive interview pretest
• Final adjudication

5
Survey language consultants (SLCs)

• Reviewed initial translations
• Supervised cognitive interview pretests
• Four SLCs
• Mandarin and Cantonese Chinese
• Korean
• Vietnamese

6
Step1 Translation

• Three independent translators
• Instruction Ask the same questions
• Products
• Translated survey items
• Documentation translation issues and approaches
  
• Lesson learned
• Set up more collaborative translation process
  when feasible

7
Step 2 Review

• Formal review process
• Template tool
• Identify problematic items
• Describe problems
• Suggest solutions
• Biweekly meetings progress, questions, and
  interim results
• Lesson learned
• SLC experience with surveys useful but not
  necessary

8
Step 3 Initial adjudication

• Adjudication team
• Adjudicator tasks
• Examine identified problems
• Assess suggested solutions
• Make decisions about changes to translations

9
Step 4 Cognitive interview pretest

• Cognitive interview purposes
• Design
• 9 interviews in Chinese and Korean
• 14 interviews in Vietnamese
• Procedures
• Script
• Cognitive interviewer training
• Incentives and gifts
• Observation and quality monitoring

10
Step 4 Cognitive interview pretest

• Lesson learned Further quality review
• First round smaller sets of interviews
• Time to observe, review and re-train
• Observation with simultaneous interpretation
• Use SLCs as cognitive interviewers
• Analysis
• SLC summaries describe everything that
  happened
• Qualitative analysis
• Input and feedback from SLCs

11
Step 5 Final adjudication

• Few additional revisions
• Some language-specific improvements
• Several more general revisions to address
  issues across all three target languages

12
Conclusions

• 5-Step model and processes effective
• Translations judged accurate
• Successively smaller revisions across steps
• Team approach translation benefited from mix of
  individual expertise
• Additional research
• More quantitative evaluation of translations and
  data quality
• Behavior coding field test