Qualitative and Quantitative Measurement Social Research Methods 2117 & 6501 Fall, 2006 November 1-20, 2006

Part I. Conceptualization and Operationalization

The meaning of the exercise: An fact: Most of the variables we want to study don’t really exist, and they seldom have a single, unambiguous meaning. [大多數我們想探討的變項實際上並不存在，這些變項也很少會有單一清楚的意義。] What can we do? Give constructs (構念) or conceptions (觀念) definitions, and communicate with others → Need measurement

The question: why measure? Measurement can: Provide information about social reality Helps us to observe, especially for those invisible things Provide clear ideas about what are being studied

Measurement Process: Conceptualization & Operationalization conceptual definition(概念定義):抽象或理論式的定義 Operationalization (操作化): 將概念定義與某種測量程序或技巧連結，以進行實證研究 Operational definition (操作定義): 研究者實際上如何測量某變數(或概念)

Develop a measure: Remember the conceptual definition (記住概念定義) Keep an open mind (保持開放心靈、活潑思考) Borrow from others (借用別人的測量) Anticipate difficulties (預期會有困難) Do not forget your units of analysis (記得分析單元)

Indicators and Dimensions of Concepts An Indicator (指標): the presence or absence of the concept [用來顯示概念存在與否] Dimension (層面). : a specific aspect of a concept [概念中的某一特別面向] During conceptualization and operationalization, we often specify different indicators to represent different dimensions of a concept. [我們可以用不同的指標來測量一概念的不同面向]

Conceptualization and Operationalization: from abstract construct to concrete measure

Quantitative and Qualitative Measurement Think about variables and measures before data collection Collect numerical data Measures are specific and precise Qualitative Think about ideas before, but measure during data collection Conceptualization flows largely from the data Operationalization before conceptaulization

Part II. Levels of Measurement

More on measurement When conceptualize and operationalize, consider the range of variation and degree of precision Good measurement: the attributes of all measures should be mutually exclusive (互斥) and exhausive (周延) Mutually exclusive attributes: 一個人或個案只能隸屬於某變數的一個屬性 Exhaustive attributes: 所有的個案都能隸屬於一變數的所有屬性

Levels of Measurement Some measures are at a higher or more refined level, and others are crude or less precisely specified.[有些較高層次的測量較精確，有些則較粗略] The conceptualization of a variable will affect the level of measurement, indicators chosen, and statistics applied. Continuous variables (連續變數) vs. discrete variables (間斷變數)

Four Levels of Measurement Nominal Measures (名義的測量): differences among categories Ex: gender, religious affiliation, college major Ordinal Measures (順序、次序的測量): categories can be ordered or ranked Ex: social class, prejudice Interval Measures (等距、間距的測量): can specify the distance between categories Ex: IQ scores Ratio Measures (等比、比率的測量): attributes are based on a true zero point Ex: age, # of times married, length of residence in a given place

Four levels of measurement

More on levels of measurement Discrete variables are nominal and ordinal. Continuous variables can be measured at the interval or ratio level. A higher level measure can be turned into a lower level measure, but not vice versa. [較高層次的測量可以轉換成較低層次的測量，但反之則不成立] More powerful statistical procedures are available for the higher levels of measurement. If possible, conceptualize and measure variables at higher levels of measurement.

Part III. Reliability and Validity

The Quality of Measurement We look at the reliability (信度) and validity (效度) of measurement. [測量的品質可由信度及效度判定]

Definition of Reliability and Validity Reliability: 對同一受訪對象，如果我們運用特定的技術，重覆進行研究，每次都能得到相同的結果，此為信度(reliability)。 Measurement reliability: 變數測量的可依賴度或一致性 Validity: 一實證測量(empirical measure)能夠反映想要研究的概念之真實意義的程度，即為效度(validity)。

Three types of reliability Stability reliability (穩定信度):信度不受時間影響 Verify by test-retest method(再試法) Representative reliability (代表性信度):信度不因受訪者群體不同而改變 Verify by subpopulation analysis (次母體分析) Equivalence reliability (等值信度):不同的指標測量一個概念的一致性 Verify by the split-half method(折半法)

Ways to improve reliability: Clearly conceptualize all constructs (概念的定義應明確清楚) Increase the level of measurement (用較高層次的測量) Use multiple indicators of a variable (同一變數用多個指標來測量) Use pretest, pilot studies, and replication (利用前測、前導測試、重複試驗來改進信度)

Four types of measurement validity Face validity(表面效度): 依字面來看，一指標真的是在測量某變數? Content validity(內容效度): 一測量應儘可能涵蓋一個概念所有的意義

Four types of measurement validity Criterion validity (效標效度): 一測量的效度有時可用另一外在的標準來決定 Concurrent validity (同時效度) & predictive validity (預測效度)

Four types of measurement validity Construct validity(建構效度):利用多個指標來判定效度，同一變數的多個指標可聚合；不同變數的指標應分歧 Convergent validity(輻合效度) & discriminant validity (區別效度)

Relationship between reliability and validity A measurement can be reliable but not valid. Usually complementary, but may conflict with each other The true essence of a highly abstract construct vs. a concrete and reliable measure Related to two different research approaches Quantitative: more reliable Qualitative: more valid Reliability and validity in qualitative research have different meanings. [信度與效度在質性研究上還有不同的意義，我們會在下學期討論。]

Relationship between reliability and validity