2-1 information needs data collection instrumentation data processing report generation sampling...
TRANSCRIPT
2-1
Information NeedsInformation Needs
Data CollectionData Collection
InstrumentationInstrumentation
Data ProcessingData Processing
Report GenerationReport Generation
Sampling DesignSampling Design
Forward Links
Backward Links
Steps in the Survey Process
2-2
Developing the Project Outline
1. List information needs by priority
2. Indicate the value of the information
3. Identify internal resource requirements
4. Specify sample size and design
5. Provide a mock-up of instrumentation
6. Note the scope of the response task
7. Describe the data collection method
8. Outline the data processing method
9. Describe the type of reports required
10. Summarize final costs and the timetable
2-3
Information Value and Priority
• The cost of selecting a "bad" alternative or failing to select the best alternative would be high
• There's much uncertainty about which alternative to choose, based on existing information
• Research information is likely to reduce a substantial amount of existing uncertainty
2-4
Data collection costs High Medium Low Low
Data collection time required Medium Low Medium High
Sample size for a given budget Small Medium Large Large
Data quantity per respondent High Medium Low Low
Reaches high proportion of public Yes Yes No Yes
Reaches widely dispersed sample No Maybe Yes Yes
Reaches special locations Yes Maybe No No
Interaction with respondents Yes Yes No No
Degree of interviewer bias High Medium None None
Severity of non-response bias Low Low High High
Presentation of visual stimuli Yes No Yes Maybe
Field worker training required Yes Yes No No
Comparison of Data Collection Methods
Personal Phone Online Mail
2-5
When Interviews Are Required
• Interaction with respondents required
• Some opinions discourage response
• Adequate mailing lists aren't available
• Must be done at specific location or time
• Instrument can't be self-administered
• Much data needed from each respondent
2-6
When Personal Interviews Are Required
• Must be collected at a special location
• Respondents must have visual contact
• Interviewer must have visual contact
• Interviews are long or demands rapport
• Phone directories are inadequate frames
2-7
When Personal Interviews Are Permitted
• Appearance won't cause selection bias
• Personal encounter won't cause threat
• Companions aren't likely to interfere
• Respondents are closely congregated
• There is adequate time for data collection
2-8
Large Sample Sizes and Small Response Tasks
• Precise estimates of numeric values required
• Main interest in Individual items, not patterns
• Volume of info from each respondent low
• Self-administered survey requires a simple response task
Fixed level of funding
Am
ount
of
data
fro
m
each
resp
ond
ent
Size
of th
e
resp
ondin
g sa
mple
Large Large
Small Small
2-9
Small Sample Sizes and Large Response Tasks
• Only approximate estimates are needed
• Main interest is in patterns among many variables
• Amount of data needed from each respondent is high
• Interviewing permits a large response task
Fixed level of funding
Am
ount
of
data
fro
m
each
resp
ond
ent
Size
of th
e
resp
ondin
g sa
mple
Large Large
Small Small
2-10
Sampling Design Alternatives
• Random Sampling• Every sample unit has an equal chance
of selection
• Convenience Sampling• Some sample units have a greater
chance of selection than others
2-11
Sampling Design Alternatives
• Stratified Sampling• Subsample strata of specific proportions
are selected
• Unstratified Sampling• Strata represented in the same
proportions as in the population
2-12
Sampling Design Alternatives
• Clustered Sampling• Geographic areas are selected and units
sampled from each
• Unclustered Sampling• Sample units are selected regardless of
geographic location
2-13
Outlining the Questionnaire
1. Identify the topics to be measured based on the information needs
2. Order the topics in a sequence that will be meaningful to respondents
3. Tentatively select question and scale types for each topic to be measured
4. Draft sample questions and compose typical scales for each topic area
5. Organize the items into sections in a logical sequence
6. Compose a rough questionnaire draft outline to serve as a model
2-14
Statistical Analysis Programs
• Complete Data Description
• Confidence Intervals for Estimates
• Measures of Relationships Among Variables
• Statistical Significance of Relationships
• May Do Tabular Reports
• May Do Charts, Graphs and Data Plots
2-15
Project Cost Elasticity
Sample SizeSmall Large
$ $
Survey Initiation
Report Generation
Data Processing
Instrument Composition
Sampling Procedure
Data Collection
Total Project Cost
70
% o
f Tota
lP
roje
ct
Cost
30
% o
f Tota
lP
roje
ct
Cost
2-16
Mail Survey Project Schedule Chart
Information needs
Sampleselection
Questionnaire composition
Mailing piece production
Mailing and field interval
Early analysis programmingData transfer
and editingData analysis, report writingFinal report to
clientFollow up
analysis, reports
Days From Start7 14 28 35 49 56 63 70 77 840 4221