Textbook Notes (270,000)
CA (160,000)
UTM (8,000)
SOC (1,000)
Chapter 5

SOC221H5 Chapter 5: Ch. 5 SOC221.doc


Department
Sociology
Course Code
SOC221H5
Professor
Ivanka Knezevic
Chapter
5

This preview shows pages 1-2. to view the full 7 pages of the document.
Qualitative and Quantitative Measurement
Why measure?
-To test hypotheses, evaluate explanations, provide empirical support, study an issue
-Must be Mutually exclusive ( when an individual or case fits into only one attribute of a variable), or
exhaustive attributes ( all cases fit in one of the attributes of a variable)
- Measurement links 3 levels: abstract theory, specific measures/indicators, and concrete social reality
Qualitative measurement: Inductive approach, alternative to numbers, measures aspects of social life, integration
of measurement with data collecting and theories.
Quantitative: measurement is very crucial, occurs prior to data collection and there are measurement techniques.
Deductive, starting with abstract concept, and then the creation of empirical measures which precisely and
accurately capture concepts in forms which can be expressed in numbers.
Measurement is fundamental to natural science: It extends the senses, varies less with different observers,
yields exact information, and provides precise information about social reality. Observes the invisible and unseen,
the difficult-to-observe aspects of the social world.
4 approaches to measurement: 1. Timing (quantitative: converting variables into actions before gathering data,
qualitative: measures during the data collection process). 2. Data form (quantitative: moves from abstract ideas to
specific data collection to precise numerical empiricism. Qualitative: written or spoken words, symbols for data.
Measurement is an ongoing process that produces data). 3. Linkages (quantitative: to follow a clear sequence, with
preplanned measurement bridging concepts and data. Qualitative: reflect, refine on, and develop concepts while
gathering and examining data). 4. Direction (quantitative: starts with abstract ideas, creating ways to measure,
ending with empirical data. Qualitative: starts with empirical data, generating ideas based on data, ending with
ideas).
Parts of the measurement process: measurement enables you to capture or observe abstract ideas in data, and you
can adjust the measurement technique to align with or capture details of the data. Measurements are limited based
off of the research techniques you know, and indicates the presence of an abstract idea which can be observed.
There are two processes to measure ideas or constructs of the world: 1. Conceptualization 2.
Operationalization.
Conceptual definition: abstract, theoretical explanation, refers to other ideas or constructs, possible alternative
definitions. A measure must be fit to the specific conceptual definition
Constructs: have multiple dimensions or topics, and one must consider the units of analysis that fit definitions the
best.
Conceptualization: thinking through a construct’s meaning, to be clear in meaning.
Operationalization: links conceptual definitions to specific things you do (techniques or procedures of
measurement). Connects the language of abstract ideas with that of concrete empirical measures. An operational
definition refers to definitions in terms of specific actions, and could be a method of observing events, and could
reflect, document, represent the abstract construct in a conceptual definition.
Quantitative measurement: Conceptualization (causal relationship between two abstract constructs, a conceptual
hypothesis) operationalization (testing an empirical hypothesis to look for associated measures) empiricism
(interpretation of data).
Hypothesis
-Has at least 2 variables, and is conceptualized and operationalized
Qualitative measurement
-Ideas are conceptualized during the data collection and analysis, not before, which allows for new concepts
to develop, definitions, and relationship among the concepts. New constructs are also developed as well.
There are clear, explicit definitions, which are linked to other ideas, and which are molded by data.
Operationalization precedes conceptualization in qualitative measurements, and it reflects and describes
how observations and thoughts about data become conceptual definitions (data is gathered before
operationalization).
Reliability and validity

Only pages 1-2 are available for preview. Some parts have been intentionally blurred.

-How one connects concrete measures to abstract concepts, though perfect reliability and validity are
impossible to measure. Establishes truth, credibility, believability.
Reliability: Easier to achieve than validity, but not necessarily valid in results. Relates to dependability,
consistency, repeated, stable outcomes under similar conditions. Can be replicated based on method and results.
For quantitative measurements, this means that numerical measures do not vary due to measurement processes.
Reliability can be improved through Conceptualized constructs (developing unambiguous, clear definitions).
Validity: truthfulness, matching constructs or conceptual definitions with a measure, and how well and idea fits
with empirical facts. How well social reality matches ideas used to understand it. Indicators can be valid for one
definition, but less for others. More difficult to achieve than reliability. As validity increases, reliability becomes
difficult to obtain if the construct is vague.
There are 4 ways to improve reliability in Quantitative research: 1. Conceptualizing constructs (measuring a
construct of a construct, developing unambiguous definitions to eliminate noise from other constructs). 2.
Increasing the level of measurement (indicators at more precise levels of measurement are more reliable, to
measure more specific information). 3. Multiple indicators of a variable (two indicators are better than 1, and this
lets you measure from a wide range of the content of a conceptual definition, and different aspects of the construct,
and to decrease errors and increase stability). 4. Pilot studies and replication (drafting measures before the final
version, replicating used measures to improve reliability).
Measurement validity: how well conceptual and operational definitions mesh. There are Three types: 1. Face
validity ( scientific community judges indicator to really measure the construct, based on a consensus). 2. Content
Validity ( captures the entire meaning of the definition in a measure, contains ideas and concepts. 3. Criterion
Validity ( uses a standard to indicate constructs accurately, compares indicators with another measure of the same
construct).
There are 2 types of Criterion validity: 1. Concurrent ( indicator is associated with pre-existing indicator which is
valid). 2. Predictive ( predicts future events which are related to the same construct).
Qualitative reliability and validity
-Reliability: techniques are used to record observations consistently, though processes are not stable.
Research is an ongoing process, and evolves through a range of data sources and multiple measures which
are impacted by social context and are not fixed. Data collection is interactive, and data measures bring
alternative results.
-Validity: qualitative researchers want authentic, fair, honest accounts of social lives, rather than connecting
data with concepts. Attempts to view the insider’s perspective.
Types of validity: 1. Internal ( no errors are internal to the design or project, few errors). 2. External ( generalizes
finding from a specific setting and small group to a broad range of people or settings. Generalizes to the real world.
High external validity= generalized results. Low= results apply to only one setting). 3. Statistical validity
( researcher has used the correct statistical procedure and met the assumptions. Robustness, or the degree to which
procedures can be violated are also studied).
Variables: Continuous ( infinite number of attributes or values, divided into smaller increments). Discrete
variables ( fixed set of values or attributes, distinct categories). Continuous variables can be conceptualized into
discrete variables, but not the other way around.
Levels of measurement: levels depend on how you conceptualize constructs ( assumptions about whether the
construct has particular characteristics), there are many ways to measure constructs, an you can collapse higher
levels of measurement to lower levels, but not the other way around. 4 levels exist: 1. Nominal measures
( difference among categories) 2. Ordinal measures ( difference plus rank among categories) 3. Intervals
(difference between categories of rank such as temperature, with an arbitrary zero) 4. Ratio ( includes the
true zero among differences and ranks of categories, but is rarely used in studies).
Scales and indexes
Index
Combines indicators of a construct into a single score, used for content validity, components are measured and
combined, has face validity. Must be unweighted, for each index to have equal weight, unless you value some
items over others. Missing data can impact indexes, due to lowered validity and reliability. Scales and indexes
should be unidimensional ( all items in a scale or index fitting together, measuring a single construct, measuring
You're Reading a Preview

Unlock to view full version