Ch5 Recap

Ch 5 Review:

Scientists like to use measurement for the word observation, because it refers to careful, deliberate, observations for the purpose of describing attributes composing a variable.

If it can be conceptualized, it can be measured.

(Note: I hate this chapter and the concept and subconcepts involved in conceptualization.)

conception - ideas about a subject, internal, individual, 'mental images'
conceptualization - process of coming to agreement about meaning of term
concept - result of conceptualization (Kaplan: is a "family of conceptions")

Kaplan's 3 classes of observables\types of observation:
Direct - simply (checkmark)
Indirect - more removed than direct observation (minutes of past mtgs)
Constructs - created, theoretical, not-directly-observable (IQ)

Reification - reguarding constructs as real

indicator - sign of absence or presence of concept (helping animal, crying during movie: indicators for concept of compassion)
dimension - grouping of indicators within a concept, a specifiable aspect of concept (compassion for animals vs compassion for humans).

Complete conceptualization involves specifying dimensions and finding indicators for each.

Interchangeability of indicators - idea that if indicators are valid, then they all rep same concept, then all will give same results. (We should get same research results no matter which indicators we use, as long as our indicators are valid).

Nominal defination - assigned to term w/o claim that def rep 'reality'. Arbritrary.
Operational defination - specifies exactly how concept will be measured. Max clarity about concept for given study.
Working defination - def for the purposes of inquiry - whatever you want it to be.

Hermeneutic circle -- cyclical process of deeper understanding

Conceptualization is also the continual refinement of the understanding of a concept.

Progression of measurement steps:

Conceptualization
|
\/
Nominal Definition
|
\/
Operational Definition
|
\/
Measurements in the real world

Srole scale: another measure of anomia (5 statements)
Durkheim theories about suicide and anomie.

Defs more problematic for descriptive research than for explanatory res. Why?
For example - descriptive: What does 'being unemployed' mean? Who qualifies? What people can be unemployed (children)?
Explanatory Res: Does conservatism inc with age? Not matter what def of conservative used, the relationship bet that and age is of interest, not the exact def used.

Conceptualization is the refinement and specification of abstract concepts, and operationalization is the dev of specific procedures.

Operationalization choices:
Range of Variation - how much is acceptable?
Variations between the Extremes - degree of precision, how fine are your distinctions among attributes composing your variable? (Do you care if person is 17 or 18?)
Be clear about which dimensions of a concept you're covering.
Attributes composing your variable should be exhaustive as well as mutually exclusive.

Levels of Measurement:
Nominal: labels for characteristics (gender, hair color), analysis available is that 2 people are the same or different.
Ordinal: rank-ordered (conservatism, alienation), analysis can say that Person A is “more” than B in terms of var.
Interval: distance separating attributes of a var HAS meaning, but lack of absolute zero. (IQ test) Can say 'how much' more A is over B, can not say IQ of 150 is 50% more intelligent than someone w/ IQ of 100.
Ratio: Intervals with a true zero point. (K temp, age, income). Can say that A is twice B.

Precision - fineness of distinctions made bet attributes of var. (43 vs 'in 40s')
Accurate - how closely something matches reality.
Reliability - does your teq give repeatable results when measuring the same object? Problem with researcher subjectivity.

Methods of ensuring reliability:
Test-Retest method: measure multiple times and compare results (survey repeated 3mo later - get same results?)
Split-Half: make more than one measurement of same concept. (Take q that test concept and analyze after splitting Q list in half).
Use Established Measures - use someone elses'
Test reliability of res workers -- call subset of samples and verify info, code results through multiple people

Validity
-- Does a measure reflect real meaning of concept?
Face validity - common sense
criterion-related validity - predictive validity - external criteria (validity of driver's test det by scores people get and rel to later driving records)

*Need help on predictive validity, in the making examples step. I understand the concept but can't make em, yet.

Construct validity - logical rel among vars. (marital satisfaction relates to cheating)
Content validity - does the measure cover range of meanings w/in concept. (does our measure consider all types of prejudice?)