Chapter 4 Flashcards
Variable language
Study variations in attributes among people and artefacts. A concept that shows variation in values is a variable.
Careful thinking
Identify properties of content that represents theoretical concepts and transform into numbers
Measurement
Reliable and valid process of assigning numbers to units of content.
Measurement failure =
Unreliable and invalid data - inaccurate conclusions. CA: waste of effort.
CA: intercoder reliability:
Trained coders apply the same classification rules in a predetermined protocol to same content.
Validity:
Assignment of numbers represents the concept being studied
By counting unambiguous recognisable verbal events.
A common problem
Isolating a phenomena enables us to study it, removes it from the context, resulting in distortion of the understanding.
In CA:
Reducing a body of content into units risks losing communication context
Success of our predictions based on our use of content units must validate the choice of the variables we study and the ways measured.
Level of measurement for our variables must correspond to the phrasing of relationships among variables in our hypothesis.
Broadcast level:
Content forms, manifest ways decomposed in parts.
Units of observation
content likely to include variables of interest.
Units of analysis
Includes the variables of interest measured at a level informed by our hypothesis or questions
Using more than one communication form
Will increase since the increase of web based materials.
Problems associated with measuring non text forms
- Non-text adds dimension: can cloud the manifest content of communication. Meaning of words and symbol, involves inflection, tone and body language. Difficult to interpret and categorise
- Visual can create analysis problems - ambiguities that are not resolved within the message.
- Reliability on the expense of validity when studying content variables
Shared meaning:
Written text provided within message cues that serves to reduce ambiguity.
Shared meanings of visuals are less common.
Combinations of communication forms
Coding problems because of between forms.
Multiform communication
Requires consistency if accurate communication is to occur.
Inconsistent
Meaning of content becomes ambiguous, categorising more difficult.
Units of observation:
Access to content containing the variable data that address that concern.
More specific demarcations of content:
Focus our observation on content of interest.
No system for specifying how many levels of observation units.
Successive definitions of units of observation to map the world so researchers can follow.
Physical units:
Time and space measures. Occupy physical space measured by centimeters. Type of measure: should be based on type of information needed in the study, variables investigated and theoretical basis.
Allocation of content space or time is systematic and not random. Identifiable as nonrandom content patterns. Greater the space - greater impact on the audience.
Meaning units
Physical qualities, less standardized. Meaning of the words that are of focus - richness and ambiguity to inferences from them to antecedent causes or effects.
Symbolic units in CA are syntactical units:
Occur as discrete units in a language or medium. The word, particular verses in the bible? for example.
Sampling concerns with unit of observation
Census or purposive sample of all units of observation, reasoning to justify the decision.
random sample? keep the rules of random sampling inference. Only generalize from a sample of one type of observation units to the population of such units from which the sample was taken
Units of analysis=
Demarcated content which we can define and observe one or more variables
Data that can give us an answer. Heart of work: Define variables in our hypothesis, observable in content, quantify what we have found.
This stage: two types of decisions of critical theoretical and empirical importance:
Definitions of content (classification into categories) and measurement of content (level of measurement)
CA protocols
Descriptions for how content can be observed and classified into categories.
Each content variable
At least two categories in which the content unit are placed.
One or more types of clothing: level of measurement: dichotomous variable (revealing or not revealing) or do we have a scale
Classification
Collection of definitions that link observed content to the theoretical, conceptual variables. Nominal: categories. Translates messages into variables by assigning content units to categories.
Deese: typology in conceptualizing CA variables frequently used:
Grouping: Content place into groups when units of analysis share common attribute.
Class structure: groups have hierarchical relation (some higher than others)
Three levels of abstraction: general, doctrinal, and substantive
Dimensional ordering or scaling: Classified based on numerical scale.
Five properties involving scaling: 1. intensity, numeouristy, probability, position or length and time
Spatial representation and models: language thought of as representing a cognitive space or map.
Meaning of words: placed in a mental spatial model for a person to evaluate objects, issues and people. Content has two or more dimensions that need description. Semantic differential through CA. Explore complex meanings attached to symbols.
Abstract relations:
Tv characters? Classification system can specify and make more concrete these abstract relations expressed in content
Binary attribute structure:
Characteristics to a person often have an opposite. Good opposite to bad, this can be found in content, although concepts need to be thought
Content classification relates
To the validity of the concepts.
Selection of classification system for content should have a theoretical basis. Validity argued logically and empirical
-
Classification System requirement (Must meet particular requirements dictated by the logic of empirical inquiry.
Necessary to establish the validity of concepts
Creating categories required instructions so they can be coded reliably.)
Meet requirements: 1. Reflect the purpose of the research, 2. mutually exclusive, 3. be exhaustive, 4. be independent, 5. be derived from a single classification principle
Purpose of research
Define variables theoretically
Coding instructions must specify how and why content units will be placed in categories for variables.
Classification systems:
Mutually exclusive when assigning numbers.
Using statistics to study patterns requires
That units be unambiguous in meaning - giving more than one number to a content created ambiguity
Problem?
Select smaller units of analysis that can be classified in mutually exclusive ways. Instead of the article as unit: statements within the article are the focus.
Exhaustive:
Every unit of content must fit into a subcategory - easy for content with a lot of attention. May fall into another section instead. Don’t use others too often.
Extensive pretesting with similar content will create categories that are exhaustive.
Independence
Placing content does not influence the placement of the other units. Important in assessment of coder reliability and statistical analysis.
Coder reliability:
Both agreements and disagreements, may be forced by coding instructions.
Each category should have a single classification principle
That separates different levels of analysis.
Each separate rules for classifying units.
(Systems without classification violate the mutually exclusive rule.)
Level of Measurement
Content: one of four levels of measurement: nominal, ordinal, interval and ratio.
Nominal:
Numbers to categories of content. Arbitrary. No meaning just connecting.
Dummy variables
Calculate presence of multiple candidates in a tweet by combining these dummy variables into new variables with a statistical package.
Multivariable approach:
Each category becomes a variable with one number for having variable characteristic - same article placed in more than one classification.
CA options to create variables that control multiple meanings to occur:
Break into smaller units. Fewer alternative interpretations.
Negatives with multi variables:
Entirety artificial variety into coding, contingencies intrinsic to multiple valued accounts of phenomena are lost. Produce more unreliable variables. Less information about reliability of the protocol.
Observation units
Can be manipulated to form different analysis units. (Dummy more consistensy)
Ordinal measures:
Content units into categories, but with an order.
Intervals have an order, but differences between numbers are equal. each interval are equal to all intervals.
Ratio:
Meaningful zero point.
Ratio: allows to compare relative emphasis regardless of number of units. Interval and ratio: more sophisticated statistical procedures, computation of means and measures of dispersion (variance and SD) Multiple regression: control stat for influence of a variety of variables.
Importance of measurement levels
Two rules: theoretically appropriate and much information.
Theo; reflect the nature of the content.
Level of measurement: determined what statistical procedures can be used. Interval and ratio: parametric procedures, describe more precisely. Nominal is parametic: less information about patterns
Rules of Enumeration
Requires creation of rules coders must follow for connecting content with numbers. Number represents the level of measurement.
Nominal: arbitrary picking numbers for the groups
Interval or ratio: instructions how part of the physical content to include or exclude
COUNT words in text - description on which words to count.
Clear and consistent rules
Success: affects the reliability and validity. Consistent numbering of content.
Measurement Steps
Process of measuring content
Develop research question - identify variables (level of measurement)
Examine existing literature that has used the variable and measurement.
Build on whats already known. Theoretical definitions. Face validity in measure.
Use good previous measures, adjust your measures if not good. reviewing literature. Use it critically.
Modified: new one should have face validity consistent with existing measures. Reducing measurement error
Create coding instructions: content categories for each variable in detail.
Detailed: higher reliability. Don’t make it too difficult.
Defining: selecting among type of content, setting up classification system, and deciding enumeration rules. Create coding system for recording data into computer. Include coding sheets.
Criteria: efficacy in data input and keeping costs down. Variables arranges and numbered
Summary:
Measurement: process of moving from theoretical definitions of concepts to numerical representations of those concepts as variables.
Operationalisation. Identifying content of interest and designing classification system = content units to develop definitions of variables and categories for variables. Must be translated to numbers - level of measurement, classifying content and rules for applying numbers to the content.
Coding instructions that maximise the validity and reliability of measurements of content concepts of interest.
Coders to replicate. Measures are statistically analysed to address study hypothesis and RQ.
