Post on 29-Dec-2015
transcript
From An Introduction to Scientific Research Methods in Geography
(Montello and Sutton)
Behavioral Observation and Archives
Learning ObjectivesHow are behavioral observations and archives
used to collect data in geography?How are behavioral observations and archives
examples of nonreactive measurements, and how are they not?
How is scientific behavioral observation different than everyday behavioral observation?
What is “coding” open-ended records, and how are coding systems developed?
How are validity and reliability established for coded data?
Behavioral Observation & ArchivesBoth are methods to collect data on
people’s current or ongoing activitiesBoth can produce data without requiring
people to intentionally and knowingly supply information to a researcher (as can physical trace measures)
Behavioral Observation BasicsBehavior is overt, potentially perceptible
action or activity by people or other animals; nearly always goal-directed
Scientific observation of behavior aims to apply systematicity and objectivity to observing behavior; requires a planned strategy; initially produces data rather than inferences about the meaning or intention of the behavior
Records of behavior are made for later coding into data
Nonreactive and Reactive Observation
The nonreactive nature of behavioral observation depends on whether or not people are aware they are being observed and recorded
Observers or recording devices may be hidden, or placed in plain sight with a delay before recording (those being observed likely revert to “normal” behavior).
Participant observation is when the researcher joins an ongoing setting or subcultural group as a member
Ways to Observe BehaviorAnnecdotal records (these are rich, but lack
systematicity)Checklists or rating scales of activitiesA specimen record or “running record” (a detailed
and comprehensive record of what an individual or group does over time)
Formal observation schedules: (1) Time sampling (specified, uniform time intervals), (2) Event sampling (whenever the event occurs during the study)
Researchers must use time or event sampling when data depend on temporally recurring phenomena
Ways to Encode BehaviorCan be live and in situ; but coding reliability
is suspect; the researcher may miss critical events
Typically, researchers use recording media and code later to produce interpretable data
The recording media vary, depending on nature of the record, cost and efficiency
Examples include: still photographs, audio recordings, video recordings; radio or GPS tracking; traffic recorders; transaction logs (internet site visits)
Behavioral Observation ProblemsCoding is labor intensive and difficult to do wellIt may be reactive if those observed become
aware of researcherObservers’ and coders’ perception is subjective
and selectiveObservers have tendency to interpret the world
meaningfully (instead of just recording behavior)
Observers and recording devices have points of view (they observe from some positions and not others)
Questions: Behavioral ObservationHow does scientific behavioral observation
differ from everyday behavioral observation that we all do?
What are some different ways to record behavior for future coding and analysis?
What are the formal observation schedules of time and event sampling?
What are some strengths and weaknesses of behavioral observation as a technique for collecting data?
ArchivesArchives are existing records that were not
collected for the purpose of a particular geographer’s research; so, archives are clear examples of secondary data
Archives are nonreactive; may be biased if they result from a data-gathering method that influenced the subject to try and create a particular impression with the data they provide
Archives refer to records expressed in symbolic form (words, numbers, images); they indirectly refer to entities or processes
Archives often must be coded to provide usable data
Questions: ArchivesWhat are some examples of archival data
sources in geography? How do archives differ from physical materials?
To what extent are archives examples of nonreactive measures?
What are some strengths and weaknesses of archives as a source of data?
Coding Open-Ended RecordsOpen-ended records typically consist of words,
pictures, or intentional acts that have meaning; semiotic (symbolic) artifacts created by an entity with agency
Coding is the process of turning open-ended records in data; document final system in a coding manual; known as content analysis when applied to verbal or graphical expressions
Consists of (1) segmentation - breaking records into appropriate units, and (2) classification - grouping segments into abstracted categories that capture aspects of the the content or meaning of the records (virtually always reduction)
Categories for CodingCategories effectively and efficiently capture aspects
of the records relevant to specific theoretical interestsThey should be (1) specifically and operationally
defined, (2) exhaustive (provide coding option for every possible segment a coder will encounter in the record), and (3) mutually exclusive (each segment falls into exactly one category)
Three to ten categories are recommended (depends on nature of records and research question)
Meaning of categories should depend on prior conceptual and theoretical knowledge accumulated in the problem domain
Coding ExampleCase study: How people look at and remember
information from topographic contour maps and natural landscapes
Researchers recorded a transcription of the subjects’ verbal recall
A coding sheet was used to classify elements of the verbal recalls in the transcription record
Coding sheet terms included: demographic information, number of features (and mistakes), 8 terrain types (either single or multielement), 7 spatial terms (either ego-independent or ego-dependent)
Coding Reliability and ValidityCoded records are data; want to know if they
measure what you think or claim they doResearchers must show coding is reliable: that
measurement is consistentReliability is established if two or more coders
redundantly and independently code subsets of the records)
Researchers must show coding is valid: that the measure is truthful (has construct validity)
Validity is determined by the detail and transparency of the documented coding system, and by statistical relationships to other data already understood to be valid
Questions: Coding Open-Ended Records
What types of data sources in geography require open-ended coding and why?
What are the two parts of coding called segmentation and classification?
What are properties of a good coding system, and how does one go about developing such a system?
How does one establish the reliability and validity of a coding system?
End…