Data Collection With Surveys

Unit 2: Describing Transit Systems with Data
Data Collection With Surveys
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Outline
• Types of surveys
• Methods of sampling
• Survey content
• Survey instruments
• Data collection plan
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Last time we discussed the type of information we collect.
Now, let’s discuss how we go about collected detailed survey data.
TYPES OF SURVEYS
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Types of Surveys
• Longitudinal survey
– Over time
• Cross-sectional survey
– Single point in time
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Survey Modes
1.
2.
3.
4.
5.
On-board/Intercept
Telephone
Mail
Web-based
Smartphones (emerging)
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
On-Board & Intercept Surveys
Overview
• Conducted on-board
vehicles, in stations or at
stops
• Target (study) population is
current riders
• Very common: 96% of
transit agencies conducted
on-board surveys between
2002-2004*
Pros & Cons
+ Easy to administer
+ Inexpensive
+ Relatively high response
rates (33%-67%*)
- Constraint of trip time on
length of survey instrument
(mail-back option)
- Non-riders excluded
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
On-Board Survey Example
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Telephone Survey
Overview
Pros & Cons
• Sampling frame is the
household
• Often use Random Digit
Dialing (RDD)
• Most common in cities with
high use of transit
+ Riders and non-riders
+ Medium to long survey
instruments
+ Good response rates
– 71% of transit agencies
conducted telephone surveys
between 2002-2004*
- Interviewer training
(Computer Aided Telephone
Interviewing, or CATI, can
help)
- Respondent needs a (local)
telephone
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Mail Survey
Overview
Pros & Cons
• Sampling frame is the
household
• Somewhat common
+
+
+
+
– 38% of transit agencies
conducted mail surveys*
Riders and non-riders
Longer survey instruments
Easy to administer
Eliminates interviewer bias
- Response rates can be very low;
typically below 20%*
- Usually not used for origindestination surveys
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Web Survey
Overview
Pros & Cons
• Sampling frame is riders
(usually)
• Increasingly common
+
+
+
+
+
– 44% of transit agencies
conducted mail surveys
between 2002-2004*
Very inexpensive
Very easy to administer
Longer survey instruments
Eliminates interviewer bias
Data already coded
- Respondent needs to have
internet
- Difficult to achieve random
sample
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Smartphone Survey
Overview
Pros & Cons
• Native smartphone
applications can be used
for travel surveys
• Geocoded travel
information
supplemented with
survey questions
• Still in research &
development
+
+
+
+
– Some state DOTs; limited
(if any) use in transit
Geocoded travel information
Longer survey instruments
Relatively easy to administer
Data already coded
- Respondent needs to have a
smartphone
- Difficult to achieve random
sample
- Privacy concerns
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
METHODS OF SAMPLING
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Methods of Sampling
• Simple Random Sampling
• Systematic Sampling
– every nth unit selected
• Stratified Sampling
– divide the population into homogenous groups
(e.g. areas of the city)
• Cluster Sampling
– natural groups, sample within groups
• Probability Proportionate to Size
– selection probability proportional to size
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Sample Size Determination
• Depends on the population size (e.g. number of
riders) and the level of precision desired
• Increase sample size  increase costs
• Example for Simple Random Sampling
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
What is commonly included in transit surveys?
SURVEY CONTENT
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Research Goals & Content
• Origin & Destination Surveys
– “This trip”: Origin, destination, purpose, access mode,
egress mode, etc.
• General Travel Behavior
– Frequency of travel, how long using transit, alternative
modes
• Demographics
– Gender, age, race, driver’s license, HH income, HH size,
employment status
• Satisfaction
– Attributes of service: timely service, speed of service,
cleanliness, comfort, safety, crowing, information
• Other
– Fare payment, new technologies, communications, etc.
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Translating Research into Survey Questions
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
How to write a good survey
SURVEY INSTRUMENT
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Types of Questions
• Closed-Ended
– Dichotomous (2 answers, e.g. Yes/No)
– Nominal (Categorical)
– Ordinal (a.k.a. Likert, Ranking)
– Filter/Contingency (Qualified)
• Partially-Closed
– Last answer is “Other______”
• Open-Ended
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Do’s and Don’ts
• Questions should be:
–
–
–
–
• Questions should NOT be:
Concise (multiple choice)
Understandable
Useful
Exhaustive
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
–
–
–
–
–
–
Double-barreled
Double negative
Technical jargon
Long-winded
Redundant
Obtrusive or embarrassing
Now, its time for you to practice
IN-CLASS EXERCISE
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Putting it all together in order to create a…
DATA COLLECTION PLAN
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Creating a Data Collection Plan
1.
2.
3.
4.
5.
6.
7.
8.
9.
Determine research goals
Select survey mode & sampling methodology
Determine necessary sample size
Create detailed sampling plan
Draft questionnaire
Pre-test the questionnaire
Conduct staff training
Distribute survey & revise sampling plan
Code/process data & error check
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Other Considerations
• Language
– English & Spanish
• Special groups
– Those who cannot read/write
• Longitudinal comparisons
– Consistency of questions with prior transit surveys
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
What does a (good) transit survey look like?
CHICAGO TRANSIT AUTHORITY
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
2003 Chicago Transit Authority
Customer Satisfaction Survey
• Objective: Comprehensive review of customer
satisfaction conducted every 2 to 3 years
• Content Included:
– General travel behavior (modes, trip purpose)
– Satisfaction with service delivery attributes
– Loyalty towards CTA
– Technology use (internet access, CTA website)
– Fare payment
– Demographics
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
CTA: Mode & Sampling
• Mode: Household Telephone survey
– Random Digit Dialing
– Computer Assist
• Stratified Sampling
– Area of residence (N, S, NW, SW, W, Suburbs)
– Mode (rail or bus)
• Sample Size
– Minimum of 200 respondents per strata
– Total of 2,577 responses
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
CTA: Questionnaire
• Question types
– Mostly closed-ended questions (scales typically 5
points)
– A few partially closed (other) & open ended
questions
• Interview length
– Average of 21.4 minutes to complete
– Pre-tested to assure length
• Response rate
– 64,986 telephone numbers attempted
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
CTA: Results
• Satisfaction with bus service:
– Has improved terms of reliability, information services, and comfort,
among other things.
– Has decreased in regards to access to service, personal safety, and
intramodal transferring (within CTA).
• Satisfaction with rail service:
– Has improved or stay constant in terms of operator attributes
(personnel), information sources, reliability and fare payment, among
other things.
– Has decreased in regards to access to service, intramodal travel,
personal safety, and appearance.
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
CTA: Results
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Conclusion
• Surveys are a good way for agencies to know
who their riders are and what they want.
• There are several survey types with varying
costs, penetration and bias. They should be
chosen to match their target.
• Surveys should be easy to understand and
should encourage participants to complete
them.
Materials developed by K. Watkins, J. LaMondia and C. Brakewood
Reference
Materials in this lecture were taken from:
• Schaller. “On-board and Intercept Transit
Survey Techniques”. Transit Cooperative
Research Program Synthesis 63, National
Research Council, Washington, DC (2005).
• CTA, Customer Satisfaction Survey, 2003.
Materials developed by K. Watkins, J. LaMondia and C. Brakewood