Ch 5 Conceptualization, Operationalization, and Measurement- Socypsy 2K03 PDF

Preview

Summary

Download Ch 5 Conceptualization, Operationalization, and Measurement- Socypsy 2K03 PDF

Description

Conceptualization, Operationalization, and Measurement Where Concepts Come From ● ● ● ●

●

●

● ●

Concepts gave 2 impt features: (1) Nonempirical (2) are meaningful Empirical experiences are sensate ones: five senses → concepts do not have this Meaning gives concepts life → If terms arent meaningful, it seems like nonsense to us Concepts in our mind are what we used to make our life meaningful ○ As you grow older and learn more concepts, the potential meaning in our lives expand Concepts — abstract categories for organizing sensory experience; a product of conceptualization ○ Something we actively create, not discovered ○ Fundamental components of thinking Conceptualization — process by which concepts are formed through the selective organization of sensory experience ○ Classification of different concrete objects into the same abstract category ○ Ignoring unique empirical properties or phenomena to focus on shared ones in order to categorize them into premade concepts When expressing relationships between concepts, we are creating ideas or propositions  Meaningfully communicate with others, you use the concepts in your mind ○ No guarantee that other people hold the same meanings for the same concept



Concepts, Meanings, and Definitions ● Reification — mistake of treating a conceptual construction as something real ○ Must understand that concepts are constructions which may have problems of communication ○ Cannot assume that others share the same meaning for a concept as you do ● Specification — process of clarifying the meaning of concepts ● Form 2 types definitions: c onceptual and o perational  ○ Conceptual Definitions — statement that indicates the meaning of an object term by expressing it in other abstract terms ■ Neither true nor false, but not all equally useful ■ Useful conceptual definitions are those that specify clear boundaries → informs what is and is not included from the concept ○ Operational Definitions — the specific steps (operations) of observing abstract concepts at the concrete level ■ Includes operationalization, instrumentation, and measurement ■ Connecting an abstract concept with concrete observations ■ Specifies the meaning of a concept by specifying the steps used to experience it empirically

● Conceptual definitions provide the dimensions of the concept that need to be articulated in an operational definition ○ Dimensions a specifiable aspect/facet of a concept  AN EXAMPLE OF CONCEPTUALIZATION AND DEFINITION: THE CONCEPT OF ANOMIE ● Anomie — first used by Emile Durkheim and his studies of suicide ○ Was a re-conceptualization of anomy, with the purpose of social scientific use ● Compared govt publications on suicide rates as a result of various factors in different regions and countries ● Says suicide reflected the extent to which a society’s agreements were clear and stable ○ Times of social upheaval and change often present uncertainties about what is expected → uncertainty may cause confusion, anxiety, and self destruction ● Anomie used to describe the societal condition of normlessness ○ Results from a disparity btwn the goals of society and the means prescribed to achieve them ○ Eg. Widely shared goal of monetary success, yet not all have the resources to achieve it through acceptable means ● Anomia used to refer to indv characteristics of feelings of loss of orientation, emptiness, apathy, and meaninglessness  CREATING CONCEPTUAL ORDER  ● Continual refinement of concepts as your understanding of each aspect of it grows ● Must address each concept for any study design, esp rigorously structured designs ○ Eg. Operationalization of surveys results in a commitment to a specific set of questionnaire items that will represent the concepts under study ■ Without that commitment, the study cannot proceed ● Must begin with initial set of anticipated meetings that can be refined during data collection and interpretation ○ Be conscious of an explicit about these starting points 

Definitions in Descriptive and Explanatory Studies ● 2 general purposes of research: description and explanation ● Descriptive studies → use concepts to classify and give meaning to observations under consideration ○ Does not help us understand w  hy  certain phenomenon exist or h  ow it occurred ○ Descriptive statements depend directly on the conceptual and operational definitions used ● Explanatory studies → focus on multiple concepts rather than a single one ○ Explanations require systematically linking different concepts together ○ Must account for something by linking it to something else → an event or experience does not explain itself ○ Tautology — the thinking error that claims to explain something by referring to itself (Circular reasoning)

Operationalization Choices ● Conceptualization → the refinement and specifications of abstract concepts ● Operationalization → the development of specific research procedures or operations that will result in empirical observations representing those concepts ● Variety of choices when operationalizing a concept  RANGE OF VARIATION ● Must be clear about the range of variation → to what extent are we willing to combine attributes in fairly gross categories ● Eg. Measuring attitudes toward the expanded use of nuclear power generators ○ Eg. answer categories ranging from “favour it lots“ to “don’t favour it at all“ ○ Conceals half the attitudinal spectrum → many people in more or less degrees of intensity actively oppose it ■ Considerable variation on the left side of zero ○ Measure full range of variation → operationalize attitudes with range from favouring very much, through no feelings, to opposing it very much ● Shld always consider whether you really need to measure the full range of variation ○ If not relevant to research, don’t bother ● Decisions on range of variation shld be governed by the expected distribution of attributes among the subjects of the study  DIMENSIONS & VARIATIONS BTWN THE EXTREMES ● Degree of precision → how fine will you make distinctions among the various possible attributes composing a given variable ○ Eg. Does it matter whether someone is 17 or 18 yrs old, or could you still get the results you need by grouping them together ● If unsure, always better to get too much than too little ● Should be clear about which dimensions of a concept you’re interested in measuring ○ Eg. May accidentally measure how people feel about something when you really wanted to know how much they think there is ● After everything, may have to decide what l evel of measurement to use  DEFINING VARIABLES AND ATTRIBUTES ● Attribute — a characteristic or quality of some object (eg. female, old, student, etc) ● Variables — logical sets of attributes (eg. gender, age, status, etc) ● Every variable and its attributes must have two qualities: ○ Exhaustive — a property of a variable ensuring that all objects can be classified ■ Eg. Adding option “other” ■ Must be able to classify every observation ○ Mutually Exclusive — a property of a variable ensuring that every object can be classified into only one attribute ■ Eg. Must define “employed“ and “unemployed“ in such a way that nobody can be both at the same time

LEVELS OF MEASUREMENT ● Attributes composing variables may represent different levels of measurement ● Four levels: nominal, ordinal, interval, ratio Nominal Measures  ● Nominal measures — a variable whose attributes have only the characteristics of being jointly exhaustive and mutually exclusive ○ Lvl of measurement describing a variable that has attributes that are merely different from each other ● Merely offer names or labels for characteristics ● Eg. Gender, religious affiliation, birthplace, political affiliation, hair color, etc ● Eg. (with exceptions) male and female compose the variable gender → are mutually exclusive + exhaust the possibilities of gender ● Can only say, in terms of a nominal variable, is that they are either the same or different  Ordinal Measures  ● Ordinal Measure — lvl of measurement describing a variable with attributes one can rank order along some dimension ● Eg. Social class, conservatism, alienation, prejudice, intellectual sophistication, etc ● Eg. Variable Socio economic status are composed of the attributes high, medium, and low ● When comparing in terms of an ordinal variable, can say one is “more“ than the other ○ Eg. More conservative, more religious, more prejudiced, etc  Interval Measures  ● Interval Measure — lvl of measurement describing a variable whose attributes are rank ordered and have equal distances between adjacent attributes ○ Where the actual distance separating attributes does have meaning ● Eg. The Celsius temperature scale ○ The difference between 30° and 40 is the same as between 10° and 20 ○ However, 40° is not twice as hot as 20 because the zero points in the Celsius scale are arbitrary → 0° does not mean lack of heat ○ -30° doesn’t represent 30° less than no heat ● Eg. Standardized IQ tests ○ Interval btwn score of 100 and 110 can be seen as the same as the interval separating 110 and 120 ○ But, would be incorrect to infer that someone with IQ 150 is 50% more intelligent than someone with IQ 100 ■ A person who received a score of zero cannot be regarded as having no intelligence ● When comparing two things using an interval variable, we can say they are different from one another (nominal), that one is more than another (ordinal), and say h  ow much  more  

 Ratio Measures  ● Ratio Measure — a lvl of measurement describing a variable whose attributes have all the qualities of nominal, ordinal, and interval measures, and are based on a true  zero  point ● Eg. Age, income, number of times married, number of friends, length of residence, etc. ● Allows us to say that one person is “twice as” old as another ● Comparing two people using a ratio variable then allows us to conclude: ○ (1) they are different or the same ○ (2) One is more than the other ○ (3) how much they differ ○ (4) The ratio of one to another  IMPLICATIONS OF LEVELS OF MEASUREMENT ● The lvls of measurement are embedded in one another → higher lvls can be converted to lower ones → can always make a measure less complex ● Eg. Ratio measurement of “number of dollars per year” can be converted to an ordinal measurement into categories like low, medium, and high ● Lvl of measurement determined by the analytical uses planned for a given variable ● If a variable is to be used in a variety of ways (or if unsure), the study should be designed to achieve the highest level required (though not always)

 ● Most research projects will use variables at different levels of measurement  SINGLE OR MULTIPLE INDICATORS ● Many variables have obvious single indicators ○ Eg. Gender (with some exceptions) is a rather straightforward variable usually turning out to be a matter of male or female ● Many concepts are subject to varying interpretations, each with several possible indicators

○ Can combine pieces of info to create a composite measurement of a variable 

Criteria for Measurement Quality PRECISION AND ACCURACY ● Precision — the property that refers to the fineness of measurement distinctions ○ Eg. saying 43 yrs old is more precise than saying “in their 40s” ○ Not always necessary or desirable ● Operationalization of concepts must be guided partly by an understanding of the degree of precision required ● Accuracy — the property that refers to the correctness of measurements ○ Some less precise descriptions or a more accurate, better reflection of the real world RELIABILITY ● Reliability — whether the same data would have been collected each time in repeated observations of the same phenomena ● Does not always ensure accuracy → must account for possible biases and subjectivity ● Several techniques for cross-checking the reliability of the measures devised Test-Retest Method  ● Sometimes appropriate to make the same measurement more than once ● If do not expect info to change, should expect the same response both times ● If answers vary, measurement method may be unreliable depending on extent of variation Split-Half Method  ● Make more than one measurement of any social concept ● Measures the extent to which all parts of the test contribute equally to what is being measured ● Compare results of 1/2 of a test with the results from the other half ● Each set should provide a good measure of the variable and should classify respondents in the same way ○ If the two sets of items classify people differently, there may be a problem of reliability Using Established Measures  ● Use measures that have proven the reliability in previous research ● Eg. Use of the Minnesota Multiphasic Personality Inventory (MMPI) as a standard in the psychological assessment of individuals Reliability of Research Workers  ● Possible for measurement and reliability to be generated by research workers ● Use of replication to see if results come out similar ● Clarity, specificity, training, and practice as prevention methods  VALIDITY ● Validity — extent to which an empirical measure Adequately reflects the concept it is intended to measure ○ Operates on agreements about the terms we used in the concepts they represent

● Face Validity — extent to which a test is viewed as covering the concept it purports to measure → transparency/relevance of a test as it appears to test what its supposed to ● Criterion-Related Validity — degree to which a measure relates with some external criterion → AKA predictive validity ○ Validity shown in its ability to predict future events or outcomes ○ Consider how the variable ought to relate to other variables, theoretically ● Construct Validity — degree to which a measure relates to other variables as expected within a system of theoretical relationships ○ The degree to which inferences can legitimately be made on the basis of observations or measurements ○ Based on the logical relationships among variables ● Content Validity — degree to which a measure covers the range of meanings included within a concept ○ Eg. A test of mathematical ability cannot be limited to addition alone, but also needs to cover subtraction, multiplication, division, etc. 

Indicators, Indexes, and Scales ● Both qualitative and quantitative reports use concepts to express and interpret their understanding of social reality ○ Since concepts are abstract, empirical investigation must rely on things that reflect what the concepts mean ● Indicators — an empirical specification of some abstract concept ○ Quantitative → variables used as indicators spelled out through operationalization ○ Qualitative → indicators generated from observation + interpretive task of constructing concepts that capture their meaning ● Shld look for m  ultiple indicators of concepts in order to capture meanings of concepts ○ Different indicators capture different facets of a concept ○ Explore how multiple indicators may be combined into overall, summary measures ● Quantitative data analysis use indexes and skills (composite measures) for combining indicators into a single measure ○ Often wish to study variables that have no clear and unambiguous single indicators ○ Single data item might not have enough categories to provide desired range of variation ○ May give a more comprehensive an accurate indication ● Are DATA-REDUCTION DEVICES → allow us to summarize several indicators in a single numerical score while nearly maintaining the specific details of all individual indicators  INDEXES VS SCALES ● Both ordinal measures of variables → rank order units of analysis using specific variables ● Both composite measures of variables: measurements based on more than one data item

● Index — composite measure that summarizes and rank orders several specific observations and represents some more general dimension ○ Constructed by accumulating scores assigned to individual indicators ○ Eg. Measure prejudice by adding up the number of prejudiced statements each respondent agrees with ● Scale — composite measure composed of several items that have a logical or empirical structure among them ○ Constructed by assigning scores to patterns of responses ○ Recognize that some items reflect a relatively weak degree of the variable while others reflect something stronger ○ Takes advantage of differences in intensity among the attributes of the same variable to identify distinct patterns of response ● Scales generally better → considers the intensity different items reflect of the variable ● Two misconceptions about scaling: ○ Whether the combo of data items result in a scale usually depends on the particular sample of observations under study ■ Certain items may form a scale within one sample but not within another ○ Do use of specific skill and techniques do not ensure the creation of a scale ■ Rather lets us determine whether a set of items constitutes a scale 

Typologies ● Typology — typically nominal classification of observations in terms of their attributes on 2+ variables or concepts ○ When summarizing the intersection of 2+ concepts or variables, creating a set of categories/types (a nominal variable)

 ● When the items you felt represented a single variable appear to represent two ● Extremely difficult to analyze typologies as a dependent variable ● Often used in qualitative analysis in theory building  Ethics and Measurement ● Trade-offs made between choosing the “better“ measure and the more ethical one ● Must consider how much discomfort or harm a particular measure may cause

● Could the measure put people at risk? How much deception is allowable for the sake of gathering more accurate data? ● Often disagreement and debate over what actions are considered harmful and what topics are impt enough to warrant some degree of potential risk or harm ● The determination of what is or is not ethical practice and research is now made collegially...