If you can't read please download the document
Upload
james-neill
View
10.594
Download
0
Embed Size (px)
Citation preview
Lecture 9 Week 14
Survey Design Workshop
Inter-University Research Workshop Program
Dr. James Neill
Centre for Applied Psychology
University of Canberra
1 February, 2011
Survey Design WorkshopUniversity of Canberra, ACT, AustraliaJames T. Neill
This workshop was previously presented by Dr. Brent Ritchie, 2008 who is now at the School of Tourism at The University of Queensland he kindly gave me a copy of his slides, which have been adapted and expanded each year since.
Image sources:Questionnaires are by James Neill (License: Public domain)
Scissors are by Gracenotes - http://commons.wikimedia.org/wiki/File:Edit-cut-mod.svg - (License: - Creative Commons by SA 2.5)
Further info: http://en.wikiversity.org/wiki/Survey_design/Workshop
Outline
Objectives
Introductions
Logins & Resources
Research methods
Questionnaire design
Levels of measurement
Sampling
Evaluation
Objective 1
Understand the importance of a rigorous, step-by-step process in planning, developing & implementing research questionnaires
Objective 2
Consider the pros and cons of common methods for survey administration Face-to-face interview
Telephone survey
Mail survey
Internet/mobile survey
Objective 3
Examine nuts & bolts of questionnaire design e.g.,Question style,
Response formats,
Layout, and
Pre- and pilot testing
Objective 4
Consider implementation issues Sampling methods
Sample size
Objective 5
Critically review example surveys.Existing examples
Student in-progress examples
with a view towards planning, drafting, and/or revising of an initial draft (pilot) survey.
Resources
Survey Design Workshop Notes (Wikiversity)
Readings
Books about surveys design and survey research (check library)
Image: James Neill, from Flickr, cc-by-a
Introductions
Introductions
Types of Research
(Research Methods)
There are 3 main research methods:Experimental
Quasi-experimental
Non-experimental
Image source: http://en.wikiversity.org/wiki/File:Nuvola_apps_edu_science.svgLicense: GFDL
Types of Research -
Experimental
Characterised by:Random assignment
Control over extraneous variables
Types of Research -
Quasi-experimental
Characterised by:Non-random assignment
Control over some extraneous variables
Groups are naturally occuring
Types of Research -
Non-experimental
Characterised by:No groups or conditions are created or used (i.e., no full experimental or quasi-experimental groups)
Minimal control over extraneous variables
Survey Research Characteristics
Surveys are widely used in
non-experimental social science research.
Often use interviews or questionnaires.
Involve real-world samples.
Often quantitative, but can be qualitative.
History of Survey Research
Survey research methodology was initially developed in the 1940's 1960's.
Since the 1980's, theories and principles evolved to create a unified perspective on the design, conduct, and evaluation of surveys.
Groves et al. (2004)
8 Survey Research Characteristics
Backstrom & Hursh-Csar (1981, pp. 3-4)
Systematic: follows a specific set of rules, a formal and orderly logic of operations
Impartial: selects units of the population without prejudice or preference
Representative: includes units that together are representative of the problem under study and the population affected by it
Theory-based: operations are guided by relevant principles of human behaviour and by mathematical laws of probability (chance).
Quantitative: assigns numerical values to nonnumerical character
Self-monitoring: procedures can be designed in ways that reveal any unplanned and unwanted distortions (biases) that may occur
8 Survey Research Characteristics
Backstrom & Hursh-Csar (1981, pp. 3-4)
Contemporary: it is current, more than historical, fact-finding
Replicable: other people using the same methods in the same ways can get essentially the same results
8 Survey Research Characteristics
Backstrom & Hursh-Csar (1981, pp. 3-4)
Advantages of
Survey-Based Research
Ecological validity
Access to wide range of participants
Potentially large amounts of data
May be more ethical
(than experiments)
Image source: http://commons.wikimedia.org/wiki/File:Crystal_Clear_action_edit_add.pngLicense: GFDL
Disadvantages of
Survey-Based Research
Lack of control
less internal validity
Data may be 'superficial'
Can be costly to obtain representative data
Self-report data only
Potentially low compliance rates
Image source: http://commons.wikimedia.org/wiki/File:Crystal_Clear_action_edit_remove.pngLicense: GFDL
Example Surveys
Unit Satisfaction Survey
Community Library Use
Image source: http://commons.wikimedia.org/wiki/File:Hyperlink-internet-search.svgLicense: GFDL
The research process
1. Establish
need for info/
research2. Problem
definition/
Hypotheses3. Researchdesign4. Sampling/Data collection5. Data
analysis6. Reporting
Image source: James Neill, Creative Commons Attribution-Share Alike 2.5 Australia, http://creativecommons.org/licenses/by-sa/2.5/au/
The research process
Image source: http://www.socialresearchmethods.net/kb/Assets/images/hourglas.gif
Survey construction:
Overview
What is a survey?
Types of questionnaires
Questionnaire development
Writing questions
Types of questions
Response formats LOM
Survey formatting
Image source: Unknown.
A standardised stimulus
A measuring instrument
A way of converting fuzzy psychological stuff
into hard data
for analysis
What is a survey?
Image source: Unknown.
Types ofsurveysSelf -administeredInterview -administeredPostalDelivered andcollectedTelephoneFace to facestructuredinterview
Web-basedTypes of surveys
1. Formulate
generic questionnaire2. Expand
the
questionnaireTurn into separatesections based onstudy objectives.
Draft qs &
response
formats
4. Finalise
questionnaire
& implementQuestion
order &
funnel qs
Questionnaire development
3. Pre-test,
pilot test,& redraft
Formulate Generic Questionnaire
Turn objectives into sections of the survey
Ensure all questions relate to research objectives
For explanatory objectives or hypotheses ensure both dependent and independent variables exist
Cover Letter & Ethics Statement
Introduction or cover letter:Outline details of research project and allow for ethical informed consent.
Few will read it without good prompting and being easy-to-read
Instructions
Provides consistency - helps to ensure standard conditions across different administrations
Explain how to do the survey in a user-friendly manner
Example:
Life Effectiveness Questionnaire
Instructions: Example
Image source: Life Effectiveness Questionaire (Neill, 2009)
Cover letter / ethics statement checklist
Outline details of research project e.g.,:Who are you? Are you bona fide?
Purpose of survey?
What's involved?
Explain any risks/costs/rewards
How will results be used?
Human ethics approval #
How is consent given / not given?
Voluntary - can choose not to continue anytime
More info: Complaints, how to obtain results, contact details etc.
Font (type, size)
No. of pages
Margins
Double vs. single-siding
Colour, etc.
Layout
Image sources:Questionnaires are by James Neill (License: Public domain), based on the Life Effectiveness Questionnaire
Demographics single section, usually at beginning or end of questionnaire
only include relevant questions
Layout
Image sources:Questionnaires are by James Neill (License: Public domain), based on the Life Effectiveness Questionnaire
Space for comments?
Indicate the end
Say thanks!
Layout
Image sources:Questionnaires are by James Neill (License: Public domain), based on the Life Effectiveness Questionnaire
Flow/Structure
Logical order of questions
(use sections)
Ask screening questions first, rather than later. Does the participant qualify for the survey? (esp. for internet surveys)
Use funneling/branching questions to move respondents through survey
Start off with easy to answer and engaging questions
More controversial questions in middle section
Expanding the Survey
Image source: http://www.flickr.com/photos/peretzpup/3059447579/
Personal Information
Generally, researchers put personal information at beginning of survey (if required). However, this may put off respondents, so also consider uncluding at towards end.
Consider response format e.g, Income in categories or ranges
Personal Information
More likely to respond to personal questions for anonymous or mail surveys as opposed to face to face or telephone
Show cards for responses may help for face to face interviews
Pre- & pilot-testing
Pre-test items on convenient others - ask for feedback
Revise items e.g.,
Which dont apply to everybody
Are redundant
Are misunderstood
Are non-completed
Reconsider ordering & layout
Pilot test on a small sample from the target population, analyse, & revise
e.g., skewed response items
Types of Questions
Be able to justify and defend your choices...
Imag sourcese: http://commons.wikimedia.org/wiki/File:Aiga_information_.svghttp://www.flickr.com/photos/laffy4k/404298099/
Writing questions - Dos
Define target constructs
Check related research & questionnaires
Draft items
(for important, fuzzy constructs aim to have multiple
indicators)
Pre-test & revise
Writing questions - Dos
Focus directly on topic/issue
Be clear
Be brief
Avoid big words
Use simple and correct grammar
Writing questions Don'ts
Inapplicable
must apply to all respondentsOver-demanding
e.g., recall of detail or time-consuming, unnecessary
questioningAmbiguous
meaning must be clear to all respondentsDouble negative
e.g., Do you not disapprove of tax reforms?
Writing questions Don'ts
Double-barrelled -
e.g., Do you think speed limits should be lowered for cars &
trucks?Leading -
e.g., dont you see some danger in the new policy?Loaded
e.g., Do you advocate a lower speed limit
to save human lives? vs Does traffic
safety require a lower speed limit?
Response biases
Social desirability
Acquiescence yea- and nay-saying
Self-serving bias
Order effects
Social desirabilityAcquiescence or Yea- and Nay-saying - tendency to agree or disagree with everything, use reversed items to controlSelf-serving bias - tendency to enhance selfOrder effects - routine, fatigue
Demand characteristics
InterviewHigh demand characteristics
Can elicit richer information
QuestionnaireLower demand characteristics
Information may be less rich
Types of Questions
Image: http://commons.wikimedia.org/wiki/File:Aiga_information_.svgThis example actually has many elements of a well-structued/designed survey what are they?
Accuracy of recall
decreases over time
Image source:http://books.google.com/books?id=E-8XHVsqoeUC&pg=PA58&lpg=PA58&dq=survey+design+types+of+questions+objective+subjective&source=bl&ots=fwFJdFznwn&sig=7Sil7P1uq4j6ctHzw3oKqliUhB4&hl=en&ei=KBSqSd6NJpm0sQPI3pzlDw&sa=X&oi=book_result&resnum=2&ct=result#PPA59,M1
Objective questions
A verifiably true answer (i.e., factual information) exists for each unit.
The question could be accurately answered by an observer.
e.g.,How times during 2009 did you visit a G.P.? ______
Subjective questions
Asks about fuzzy personal perceptions.
There is no true, factual answer.
Many possible answers per unit.
Can't be accurately answered by an observer. e.g.,
Think about the visits you made to a G.P. during 2010. How well did you understand the medical advice you received?perfectly very wellreasonably poorly not at all
See alos: The Power of Survey Design By Giuseppe Iarossihttp://books.google.com/books?id=E-8XHVsqoeUC&pg=PA58&lpg=PA58&dq=survey+design+types+of+questions+objective+subjective&source=bl&ots=fwFJdFznwn&sig=7Sil7P1uq4j6ctHzw3oKqliUhB4&hl=en&ei=KBSqSd6NJpm0sQPI3pzlDw&sa=X&oi=book_result&resnum=2&ct=result#PPA59,M1
Objective vs. subjective questions
Both types of questions may be appropriate; depends on the purpose of the study.
One criticism of this distinction: There is no such thing as objective and that all responses are subjective.
Open-ended Questions
Rich information can be gathered
Useful for descriptive, exploratory work
Difficult and subjective to analyse
Time consuming
Open-ended questions
Rich information can be gathered
Useful for descriptive, exploratory work
Difficult and subjective to analyse
Time consuming
Open-ended questions: Examples
What are the main issues you are currently facing in your life?
How many hours did you spend doing university study last week? _________
Closed-ended questions
Important information may be lost forever
Useful for hypothesis testing
Easy and objective to analyse
Time efficient
Levels of Measurement
=
Type of Data
Stevens (1946)
Image source: Unknown.
Levels of measurement
Nominal / Categorical
Ordinal
Interval
Ratio
Nominal/Category - measures identify categories e.g., sex, ethnicity.
Ordinal - relative ordering of responses e.g., rankings in an exam
Interval - scores stand in a quantitative relationship to one another, adjacent scores are separated by an equal interval
Ratio - like interval but with a true zero value e.g., height, speed
Discrete vs. continuous
Discrete- - - - - - - - - -
Continuous___________
Discrete data: finite options (e.g., labels)Continuous data: infinite options (e.g., cms)Discrete data is generally only whole numbers, whilst continuous data can have many decimalsDiscrete: nominal, ordinal, intervalContinuous: ratio
Each level has the properties of the preceding
levels, plus something more!
Image source: Unknown.
Categorical / nominal
Conveys a category label
(Arbitrary) assignment of #s to categories
e.g. Gender
No useful information, except as labels
Categorical / nomimal example:
Phrenological labels
Image source:L.N Fowler & Co. c. 1870.
Ordinal / ranked scale
Conveys order, but not distance
e.g. in a race, 1st, 2nd, 3rd, etc. or ranking of favourites or preferences
Image: Cropped version of http://www.flickr.com/photos/beatkueng/1350250361/?addedcomment=1#comment72157605326099631CC-by-A by Beat - http://www.flickr.com/photos/beatkueng/
Ordinal / ranked example:
Ranked importance
Rank the following aspects of the university according to what is most important to you (1 = most important through to 5 = least important) Quality of the teaching and education Quality of the social life Quality of the campus Quality of the administration Quality of the university's reputation
Image source:L.N Fowler & Co. c. 1870.
Interval scale
Conveys order & distance
0 is arbitrary
e.g., temperature (degrees C)Usually treat as continuous for > 5 intervals
Interval example:
8 point Likert scale
Image source:L.N Fowler & Co. c. 1870.
Ratio scale
Conveys order & distance
Continuous, with a meaningful 0 point
e.g. height, age, weight, time, number of times an event has occurredRatio statements can be made
e.g. X is twice as old (or high or heavy) as Y
Ratio scale:
Time
Image source:L.N Fowler & Co. c. 1870.
Why do levels of measurement matter?
Different analytical procedures
are used for different
levels of data.
More powerful statistics can be applied to higher levels
Image source: Unknown.
Levels of measurement:
Revision question
Fill in all cells
Descriptive Statistics
Ratio
Interval
Ordinal / Rank
Nominal / CategoricalGraphsExamplesProp-ertiesLevel
Closed-ended rating scales
Dichotomous
Multichotomous
Verbal frquency scale
The list (multiple response)
Ranking
Likert scale
Graphical rating scale
Semantic differential
Non-verbal (idiographic)
Image source: Questionnaire by TuppusLicense: Creative Commons Attribution 2.0
Dichotomous
2 response options e.g.,
Excluding this trip, have you visited Canberra in the previous five years?__ Yes __ No
Provides the simplest type of quantification
Multichotomous
How many hours did you spend doing university study last week?__ less than 5 hours__ > 5 to 10 hours__ > 10 to 20 hours__ more than 20 hours
eg. Which of the following statements best describes your reasons for taking a holiday to Canberra? (please tick one only) to visit friends and relatives for business for educational purposes for holiday/ sightseeing
Multichotomous
More than two possible answers e.g.,What type of attractions in your current trip to Canberra most appeal to you?__ historic buildings__ museum/art galleries __ parks and gardens
eg. Which of the following statements best describes your reasons for taking a holiday to Canberra? (please tick one only) to visit friends and relatives for business for educational purposes for holiday/ sightseeing
Verbal frequency scale
Over the past month, how often have you argued with your intimate partner?1. All the time2. Fairly often3. Occasionally4. Never5. Doesnt apply to me at the moment
Consider number of points (avoid over ~10)Consider directionConsider layout
The list (multiple response)
Provides a list of answers for respondents to choose from e.g.,Tick any words or phrases that describe your perception of Canberra as a travel destination:__ Exciting __ Important__ Boring __ Enjoyable__ Interesting __ Historical
The list (multiple-response)
What are the main issues that you are currently facing in your life? (tick all that apply) __ financial __ physical / health __ academic __ employment / unemployment __ relationships __ other (please specify)
Ranking
Helps to measure the relative importance of several itemsRank the importance of these reasons for visiting Canberra (from 1 (most) to 4 (least)):__ to visit friends and relatives__ for business__ for educational purposes__ for holiday/ sightseeing
Likert scale
Measures strength of feeling or perception.Indicate your degree
of agreement with this statement:I am an adventurous person.
(circle the best response for you)
12345stronglydisagreeneutral agree stronglydisagreeagree
12345stronglyagreeneutral disagree stronglyagree disagree
Consider number of points (avoid over ~10)Consider directionConsider layout
Graphical rating scale
How would you rate your enjoyment of the movie you just
saw?
Mark with a cross (X)
not enjoyable very enjoyable
What is your view of smoking? Tick to show your opinion.
Bad ___:___:___:___:___:___:___ GoodStrong ___:___:___:___:___:___:___ WeakMasculine ___:___:___:___:___:___:___ FeminineUnattractive ___:___:___:___:___:___:___ AttractivePassive ___:___:___:___:___:___:___ Active
Semantic differential
Non-verbal scale
Point to the face that shows how you feel about what happened to the toy.
Also called an idiographic scale.
Image source: Unknown.
Verbal frequency scale
Over the past month, how often have you argued with your intimate partner?1. All the time2. Fairly often3. Occasionally4. Never5. Doesnt apply to me at the moment
Sensitivity & reliability
Scale should be sensitive yet reliable.
Watch out for too few or too many options.
How many response options?Minimum= 2
Average= 3 to 9
Maximum= 10?
Basic guide: 7 +/- 2
Number of response options?
AGREEMENT ABOUT SOMETHING2-CategoriesDISAGREE AGREE3-Categories DISAGREE NEUTRAL AGREE
4-Categories STRONGLY MILDLY MILDLY STRONGLYDISAGREE DISAGREE
AGREE AGREE
5-Categories STRONGLY MILDLY MILDLY STRONGLY DISAGREE DISAGREE
NEUTRAL AGREE AGREE
Number of response options?
Likert scale example
Watch out for too many or too few response options
Capital punishment should be reintroduced for serious crimes
1 = Agree2 = Disagree
1 = Very, Very Strongly Agree 7 = Slightly Disagree2 = Very Strongly Agree 8 = Disagree3 = Strongly Agree 9 = Strongly Disagree4 = Agree 10 = V. Strongly Disagree5 = Slightly Agree 11 = V, V Strongly Disagree6 = Neutral
Example: How could
this question be improved?
How old are you?
___ 18-20
___ 20-22
___ 22-30
___ 30 and over
Are you satisfied with your marriage and your job?__________________________
Example: How could
this question be improved?
You didnt think the food was very good, did you?_____ Yes _____ No
Example: How could
this question be improved?
Environmental issues have become increasingly important in
choosing hotels. Are environmental considerations an important
factor when deciding on your choice of hotel accommodation?
____ Yes ____ No
Example: How could
this question be improved?
What information sources did you use to locate your restaurant
for todays meal?
(please tick appropriate spaces)
____ yellow pages
____ Internet
____ word of mouth
Example: How could
this question be improved?
Comparison of
data collection methods
Alreck and Settle (1995:32)Alreck & Settle (1995; 32)
Finalise Questionnaire Draft
Questions need to be exhaustive and mutually exclusiveInclude other (please specify)
Ensure categories do not overlap
Finalise Questionnaire Draft
LengthTry to keep them as short as possible
Only ask questions that relate to objectives
Tricks? Font size/double sided photocopying/numbering sections
Maximising Response Rate
Layout and design is key
Respondents level of interest
Colour of paper
Accompanying letter / introduction
Mail surveys - self-addressed stamped return envelope
Rewards
Reminders or follow up calls
Pre-testing and Pilot testing
Pre-test try out on convenient others & revise
Pilot test try out on a small sample from the target population & revise
Be assertive and interactive about seeking feedback ask questions & observe
The customer is always right.
Pre-test & Revise
Pre-test items and ask for feedback
Revise:
items which dont apply to everybody
redundancy
skewed response items
misinterpreted items
non-completed items
Reconsider ordering & layout
LUNCH BREAK
Survey Design Critique
In pairs, look through the example questionnaires, and highlight aspects which:could be improved
are particularly good
you would like to ask about
Examples
Examine questionnaire examples
Examine structure, design issues and question styles
Note cover page and details provided
Sampling
Image source: http://commons.wikimedia.org/wiki/File:Marbles_canicas.PNG
Sampling:
Overview
Sampling terminology
What is sampling?
Why sample?
Sampling methods
Example: Shere Hites survey
Image source: Unknown.
Sampling terminology
Target population To whom you wish to generalise
Sampling frame Those who have a chance to be selected
Sample Those who were selected and responsed
Representativeness The extent to which the sample is a good indicator of the target population
Population - set of all individuals having some common characteristic, e.g., Australians
Sampling Frame subset of the population from which the sample is actually drawn e.g., White pages
Sample set of people who actually contribute data to e.g., Every 1000th person in the white pages who answers the phone and responds
Representativeness How similar is the sample to the population with regard to the constructs of interest?
What is sampling?
Sampling is the process of selecting units (e.g., people, organizations) from a population of interest so that by studying the sample we may fairly generalize our results back to the population from which they were chosen.- Trochim (2006)
Why sample?
Reduces cost, time, sample size etc.
If the sample is representative, the sample data allows inferences to be drawn about the total population.
Representativeness of a sample depends on:
Adequacy of sampling frame
Sampling method
Adequacy of sample size
Response rate both the % & representativeness of people in sample who actually complete survey
It is better to have a small, representative sample than a large, unrepresentative sample.
Sampling methods
Probability sampling Random
Systematic
ClusterMulti-Stage Cluster
Non-probability samplingQuota
Convenience
Snowball
Probability sampling - each member of population has a specific probability of being chosen.Random Sampling - everyone in population has an equal chance of being selected.
Systematic Sampling - e.g., every 10th student ID number
Stratified Random Sampling - population divided into strata, then random sampling from within each stratum (e.g., an equal number of males/females are selected)
Cluster Sampling - identify clusters of individuals & sample from these (e.g., 1 person per household)Multi-Stage Cluster Sampling (e.g., 1 person per selected household per selected suburb)
Non-probability sampling - arbitrary, sample not representative of populationQuota Sampling - e.g., 50% psychology students, 30% economics students, 20% law students
Convenience Sampling - take them where you find them method e.g., at shopping mall
Snowball Sampling - ask each respondent if they know someone else suitable for survey e.g., studying drug-users.
Random/probability sampling
Each unit has an equal chance of selection
Selection occurs entirely by random chance
Also called representative sampling
Simple random sampling
Everyone in the target population has an equal chance of selection
Useful if clear study area or population is identified
Similar to a lottery:List of names are assigned #s and randomly select #s of respondents
Randomly select # through table of random #s or by computer
Systematic random sampling
Selecting without first numbering
Respondents (units) selected from a list/file.
Useful when survey population is similar e.g. List of students
Select sample at regular intervals from the population e.g., every 5th person on a list, starting at a random number between 1 and 5
Sometimes called file sampling
Stratified random sampling
Sub-divide population into strata (e.g., by gender, age, or location)
Then random selection from within each stratum
Improves representativeness
e.g., Telephone interviews using post-code strata
Non-random / non-probability
Also called purposive or judgemental sampling
Useful for exploratory research and case study research
Able to get large sample size quickly
Limitations include potential bias and non-representativeness
Convenience sampling
Sampling is by convenience rather than randomly
Due to time/financial constraints
e.g. surveying all those at a tourist attraction over one weekend
Purposive sampling
Respondents selected for a particular purpose e.g., because they may be typical respondentse.g., select sample of tourists aged 40-60 as this is the typical age group of visitors to Canberra
e.g., Frequent flyers to contact regarding service quality in an airline setting
Snowball sampling
Useful for difficult to access populations e.g., illegal immigratnts, drug users
Respondents recommend other respondents
e.g., in studying ecstasy users, gain trust of a few potential respondents and ask them to recommend the researcher to other potential respondents
Sampling process
Identify target population and sampling frame
Select sampling method
Calculate sample size for desired power.
Maximise return rate
Summary of sampling strategy
Identify target population and sampling frame
Selection sampling method
Calculate required sample size
Maximise return rate
Sampling Example:
Shere Hite
American Sexology
Image sources: Unknown.
Hite's survey of American
male-female relations (early 1980's)
Shere Hite doyenne of sex polls
Media furors & worldwide attention
127-item questionnaire about marriage & relations between sexes
Sample: 4500 USA women, 14 to 85 years
Conclusion: Society and men need to change to improve lives of women
Some of Hites findings about American women....
Only 13% married for 2+years were still in love
70% married for 5+ years were having affairs... usually more for 'emotional closeness than sex
76% of these women did not feel guilty
87% had a closer female friend than husband
98% wanted basic changes to love relationships
84% were emotionally unsatisfied
95% reported emotional & psychological harassment from their men
Some of the critical comments....
She goes in with
prejudice & comes
out with a statistic.The survey often
seems merely to
provide an occasion
for the authors
own male-bashing
diatribes.Hite uses statistics to bolster her
opinion that American women are
justifiably fed up with American men.
Hite's response rate &
selection bias
100,000 questionnaires were sent to a variety of womens
groups
(feminist organisations, church groups, garden clubs etc.)
4,500 replied
(4.5% return rate)
We get pretty nervous if respondents in our survey go under 70%.
Respondents to surveys differ from nonrespondents in one important
way: they go to the trouble of filling out what in this case was a
very long, complicated, and personal questionnaire.
- Regina Herzog, University of Michigan Institute for Social
Research
Hite's response rate &
selection bias
Regina Herzog, University of Michigan Institute for Social ResearchTo learn more about Shere Hites research, visit her website: http://www.hite-research.com
Sample size it's not how big, it's how representative
Objectivity watch out for manipulating the survey questions and results interpretation to suit your personal conjectures
Lessons from Hite's male-female relations survey
Regina Herzog, University of Michigan Institute for Social ResearchTo learn more about Shere Hites research, visit her website: http://www.hite-research.com
Measurement
Error
Image source: http://commons.wikimedia.org/wiki/File:Noise_effect.svgImage license: GFDL 1.2+
Measurement error
Observed score =
true score + measurement errorMeasurement error =
systematic error + random errorAny deviation from the true value caused by the measurement procedure.
Sources of measurement error
Non-sampling
(e.g., unreliable
or invalid
tests)Sampling
(e.g., non-rep. sample)Personal bias
(e.g., researcher favours a hypothesis)Paradigm
(e.g., Western focus on individualism)
Image source:s Unknown.Paradigm (e.g., assumptions, focus, collection method)
Personal researcher bias (conscious & unconscious)
Sampling (e.g., representativeness of sample)
Non-sampling (e.g, non-response, inaccurate response due to unreliable measurements, misunderstanding, social desirability, faking bad, researcher expectancy, administrative error)
To minimise measurement error
Use well designed measures:Multiple indicators
Sensitive to target constructs
Clear wording on questions/instructions
Reduce demand effects:Train interviewers
Use standard protocol
To minimise measurement error
To minimise measurement error
Maximise response rate:Pre-survey contact
Minimise length / time / hassle
Offer rewards / incentives
Coloured paper
Call backs / reminders
Ensure administrative accuracySet up efficient coding, with well-labelled variables
Check data
To minimise measurement error
Summary of sampling strategy
Identify target population and sampling frame
Selection sampling method
Calculate required sample size
Maximise return rate
Sampling Task
A research project's aim is
To identify the behaviour and attitudes of UC students with regard
to its computing services.What is the research population?
How might you get hold of a sample frame?
What sampling technique would you use?
Confidence Levels
/ Margins of Error
Relates to representativeness of a sample of a target population - to what extent can we be confident about the results?
Gives the estimated range of values into which we expect other samples to fall say 95% of the time
Social sciences = at least 90% or above, preferably 95%
Example 95%
Confidence Level Graphs
Source: http://en.wikipedia.org/wiki/File:Marginoferror95.PNG
Example 95%
Confidence Level Table
Confidence Level Example
Survey of visitors to Canberra between June 1 - August 31, 2004 = 10,000 visitorsWant to work with 95% confidence level and 3% margin of errorHow many to survey?Answer: Sample size = 964 peopleQ: What are your favourite attractions?
Following Results
+ / - 3% error @ 95% confidence level
Confidence Intervals /
Margins of Error
One time out of 20, we expect the answers may be greater than +/- 3%Establish the confidence level, margin of error you want to work with
Identify the number of surveys to be done
Once you have completed that number, do not do any additional surveys
HOWEVER.
Confidence Intervals /
Margins of Error
Sometimes we do surveys and we do not know how many will be returned until later, as with postal surveys
Thus you have to calculate the margin of error afterwards.Count up # of returned surveys
Identify target population and confidence level
Calculate margin of error
Confidence Intervals /
Margins of Error
All this assumes you know your target population and can get a sample frameIf not, a non-random sampling technique is best
Also consider whether you want to report results for sub-groups the margins of error will be widerConsider stratified random sampling
Summary - 1
Survey research has developed into a popular research method since the 1940's.
A survey is a standardised stimulus designed to convert fuzzy psychological phenomenon into hard data.
Summary - 2
Survey development - types of questions and response formats.
Sampling - probability & non-prob.
Levels of measurement & parametric / non-parametric stats
Ethical considerations
Sources of measurement error
Evaluation
Please complete the workshop evaluations
References
Alreck, P. & Settle, R. (1995). The survey research handbook (2nd ed.). New York: Irwin.Stevens, S.S. (1946). On the theory of scales of measurement. Science, 103, 677-680.Trochim, W. M. K. (2006). Sampling. In Research Methods Knowledge Base.Wikipedia (2009). Shere Hite - Methodology.See also Readings
Image source: Questionnaire by TuppusLicense: Creative Commons Attribution 2.0
Open Office Impress
This presentation was made using Open Office Impress.
Free and open source software.
http://www.openoffice.org/product/impress.html
Click to edit the outline text formatSecond Outline LevelThird Outline LevelFourth Outline LevelFifth Outline LevelSixth Outline LevelSeventh Outline LevelEighth Outline LevelNinth Outline Level
Click to edit the title text format
Click to edit the outline text formatSecond Outline LevelThird Outline LevelFourth Outline LevelFifth Outline LevelSixth Outline LevelSeventh Outline LevelEighth Outline LevelNinth Outline Level
Click to edit the title text format
Click to edit the outline text formatSecond Outline LevelThird Outline LevelFourth Outline LevelFifth Outline LevelSixth Outline LevelSeventh Outline LevelEighth Outline LevelNinth Outline Level