Introduction to Research Methods
Download
Report
Transcript Introduction to Research Methods
Questionnaire Design
Jenny Driscoll
2013
Surveys: uses
Best suited for:
large-scale research (large numbers)
quantitative data – can be counted or measured
description, knowledge, comparison, attitude,
behaviour
need for standardisation
[large budget]
competent respondents
Rough and ready rule for statistical analysis is minimum
of 30 counts in each subgroup
Can adapt for smaller-scale research BUT
Care and ingenuity required
Questionnaires and research process
Identify research aims, objectives and research
questions
Identify population and sample
Decide on means of response (postal, email,
online, face-to-face)
Design questionnaire
Run a pilot survey
Carry out main survey
Analyse the data
Report findings and dissemination
Surveys : Advantages
Can access large numbers of participants
Can cover wide geographical spread
Low cost per respondent
Low pressure on participants
Data can be easily compared
Pre-coded answers allow ease of analysis
No interviewer effect?
Surveys: Disadvantages
Often low response rate (introduces bias)
Pre-coded questions may introduce bias
No opportunity to clarify or explore
answers
No way to assess truthfulness of answers
Unsuitable for some groups
Eg visually impaired, young children,
poor literacy
Terms
Population: all members of the group you are
researching eg children in year 11
Sample: subset of population selected to
participate in research eg children in year 11 in
one London borough
Respondents: subset of sample that actually
participate
Try to ensure sample is representative of
population eg do children in sample schools
match population statistics for socio-economic
groups, ethnicity, religion, gender etc?
Sampling
Random (gold standard: randomly selected
subset of population - large numbers
required)
Systematic (every nth case)
Stratified (proportions with known variations
selected from within population)
Quota (ensure eg balance of genders, boost
ethnic minorities: implications for bias)
Cluster (eg school)
Sampling, response rate and bias
How does your sample differ from the
population you wish to study?
What is your response rate?
Are you able to ascertain how
respondents differ from non-respondents?
To what extent might your sampling
strategy and response rate have
introduced bias into your survey?
Administration of questionnaires
Postal
Email
Face to face
Online eg SurveyMonkey
Consider how administration mode affects
Response rate
Anonymity
Bias
How could you maximise response rate?
Structure of questionnaire
Preamble
Often included in main body, not as a separate
sheet. Check kcl ethics requirements for contents:
anonymity, confidentiality, use of data, right to
withdraw etc
Questionnaire schedule
Concluding statement
Thanks; how to return questionnaire; may include
invitation to participate in next stage by leaving
contact details
Questionnaire content
Content informed by
Place of questionnaire in research design
Scholarly research relating to your topic area
Account of the way you have used the research
literature will figure in write-up
Make sure you include appropriate basics eg
Age
Socio-economic status
Gender
Questionnaire Design
Keep it short!
Plan overall structure in advance
Use logical sequence
Group questions in topics
Use transitional phrases between topics
Start with easy questions
Start with relevant questions
Put sensitive questions at end
Ensure clear routing instructions
Check for ordering – might an answer be
influenced by earlier questions?
Questionnaires: Question Writing
Keep questions as short and simple as possible BUT
Meaning must be crystal clear
Use simple language – not jargon, slang or
acronyms
Think about memory issues
Missing categories – consider including ‘other’/
‘don’t know’/‘not applicable’
See http://surveynet.ac.uk/sqb/ for sample questions,
questionnaires and variables
Question writing: things to avoid
ambiguous terms
leading questions
double-barrelled questions
double negatives
sensitive questions (as far as possible)
suggestions of social desirability
Types of questions
Closed (are you male/female?)
Partially closed
Inventories (list of options)(tick all that apply)
Ratings (how much did you enjoy the art
gallery…on scale of 1-5, where 1 is ‘not at all’
and 5 is ‘enjoyed it very much’)
Rankings (list in order of importance to you)
Likert scales (attitudes) (my teacher makes
Maths fun: all of the time/most of the
time/sometimes/not much of the time /never)
Open (how do you think we could improve
your experience of the health clinic?)
Open and closed questions (from
Oppenheim)
Open
Greater response
freedom
Opportunity to probe
Can test ideas or
awareness
Time-consuming
Coding more problematic
More effort for
participants
Closed
Requires little time
No extended writing
Low costs
Easy to process
Group comparisons easy
Can test specific hypothesis
No spontaneous responses
Bias in answer categories
Sometimes too crude
May irritate respondents
Questionnaires: Piloting
Always pilot questionnaire on small
group
Test for:
Consistent interpretation of question
Understanding of instructions
“other” categories to include in main list
Successful routing
Time to complete questionnaire
Cognitive interview
Checklist
Explanation of purpose
Sponsor
Voluntary completion
Confidentiality and Data Protection Act
Return address
No omissions
No duplication
Thanks to respondents
Surveying children
See Scott (2000) in Christiansen and James
Research with Children
Until recently researchers have tended to
take proxy data from parents/teachers
Concerns about cognitive abilities of children
Accreditation of adults with superior knowledge
and experience: ‘adults know what is best for
children’
Practical and ethical issues
But validity?
Research with children under 11
Under 7, hypothetical questions unlikely to be
understood by children
Once children are able to process and
respond to standard questions they are also
adept at selecting what they choose to reveal
Young children may be very suggestible
Under 11, visual stimuli particularly useful
Questions may need to be broken down eg
do you agree or disagree? (Agree) Do you
agree a lot, quite a lot, or only a little?
Surveying children 11+
Standardized questionnaires can be used at
11+ provided researcher takes into account:
Literacy levels may vary
Confidentiality issues
Context: adolescents may not accord same
meaning to questions as adults
Boredom and flippancy
Careful pre-testing with cognitive interviews
Use focus groups to develop response
options
Reliability of children
Children are used to adults expecting a
response – include ‘don’t know’ options?
Children may be more prone to the
primacy effect
But there is no reason to suppose
children are inherently unreliable
compared to adults
Reading
Bryman, A. (2012, 4th ed) Social Research Methods Oxford: Oxford
University Press, ch 10
Christensen, P. and James, A. (eds) (2000) Research with Children:
Perspectives and Practices London: Falmer Press
De Vaus, D.A. (2002, 5th ed) Surveys in Social Research Routledge
Fink, A. (2006, 3rd ed) How to Conduct Surveys Sage
Foddy, W. (1994) Constructing Questions for Interviews and Questionnaires
Cambridge: Cambridge University Press
Gilbert, N. (2001, 2nd ed) Researching Social Life London: Sage
Gorard, S. (2001) Quantitative Methods in Educational Research London:
Continuum
Grove, R. (2009, 2nd ed) Survey Methodology Hoboken, N.J.: Wiley, chs 7 &
8
Moser, C. and Kalton, G. (2001) Survey Methods in Social Investigation
Aldershot: Ashgate
Oppenheim, A.N., (2000) Questionnaire Design London: Continuum