GVPT100 Midterm Outline I. The definition and measurement of concept A. Concept- an idea or mental construct that represents phenomena in the real world1. Conceptual question- a question expressed using ideasa. Unclear, difficult to answer empirically 2. Concrete question- a question expressed using tangible properties a. Can be answered empirically 3. Conceptual definition- describes the concept’s measurable properties and specifies the units of analysis to which the concept applies a. Problems that arise with finding a concept’s properties i. Properties might refer to a different concept ii. Properties might include immeasurable conceptual terms iii. Properties might represent different dimensions of the concept b. Properties must be concrete and must vary c. Conceptual dimension- a set of concrete traits of similar type i. Multidimensional concept- a concept that has two or more distinct groups of empirical characteristics; groups must be measured separately ii. Must communicate the variation within a measurable characteristicor set, the subjects to which the concept applies, and how the characteristic is being measured iii. “The concept of _______ is defined as the extent to which _______ exhibit the characteristic of ________. iv. Unit of analysis- entity (person, city, country) we want to analyze; can be at an individual level (lowest possible level) or aggregate level (a collection of individual entities) v. Ecological fallacy- problem that arises when an aggregate-level phenomenon is used to make inferences at the individual level4. Operational definition- describes the instrument to be used in measuring the concept and putting a conceptual definition into operation a. Measurement error i. Intended characteristic- characteristic we want to measureii. Unintended characteristic- characteristic we don’t want to measure iii. Operational definition should measure only the intended characteristic iv. Systematic measurement error- produces operational readings that consistently mismeasure the intended characteristic. - Hawthorne effect- phenomenon describing the inadvertent measuring of a subject’s response to the knowledge that he is beingstudiedv. Random measurement error- introduces chaotic distortion into the measuring process producing inconsistent operational readings of aconceptb. Reliability- the extent to which the measurement is a consistent measure of a concept; gives the same reading every time is it taken i. A reliable measure can have systematic error ii. Test-retest method- the investigator applies the measure once and then again to the same unit of analysis; same results=reliable measurement iii. Alternative-form method- the investigator administers two different but equivalent versions of the instrument to the same unit of analysis - Remedies weakness of test-rest in that the concept rather than theindividual’s memory is tested iv. Drawbacks -Hard to distinguish random error from true change -Data is obtained from panel studies that are more expensive -Panel studies- information on the same unit of analysis at two pointsin time-Cross-sectional study- contains information on units of analysis at one point in time, less expensive v. Internal consistency methods -Split-half method- an operational measurement obtained from halfof a scale’s items should be the same as the measurement obtained from the other half- the investigator divides the scale, calculates, and compares; if measurements are the same, then they are reliable. -Cronbach’s alpha- compares consistency between pairs of individual items and provides an overall reading of a measure’s reliability (0-1 scale of reliability)c. Validity- the extent to which a measurement records the true value of the intended characteristic and does not measure any unintended characteristic i. A valid measure contains no systematic error ii. A measure van be valid but not reliable iii. Face-validity approach- the investigator uses informed judgment todetermine whether an operational procedure is measuring what it issupposed to measure iv. Construct-validity approach- the researcher examines the empiricalrelationships between a measurement and other concepts to which is should be related v. Randomizing question order in a survey can convert systematic error to random error II. Measuring and Describing Variables A. Variable- an empirical measurement of a characteristic 1. Has one name and at least two values2. Central tendency- typical or average value of a variable 3. Dispersion- amount or variation in a variable’s values 4. Nominal-level variable- a variable whose values and codes only distinguish different categories of a characteristic (least precise, more common in individual-level studies)5. Interval-level variable- has values that tell us the exact quantity of the characteristic (most precise, more common in aggregate-level studies)6. Ordinal-level variable- communicates relative differences between units ofanalysis/ values can be ranked (more common in individual-level studies)a. Index- additive combination of ordinal variables, each of which is coded identically, and all of which are measures of the same variable b. Additive index/ summative scale/ ordinal scale- more reliable measurement of characteristic c. Likert scale- additive index of 5-7 value ordinals that capture a scale ofagreement/ disagreement B. Descriptive variables /central tendency1. Central tendency a. Mode- (most basic) most common value of the variable, can describe the central tendency of any variable, only central tendency that can be used for nominal-level variables b. Median- (ordinal/interval) the value of a variable that divides the casesright down the middle, impervious to amount of variation “resistant measure of central tendency” c. Mean- (interval-level only) arithmetic average, sensitive to skewness2. Frequency distribution- a tabular summary of a variable’s values a. First column- variable’s valueb. Second column- raw frequency/ number of individuals giving each response c. Third column- total frequency/ percentage of cases falling into each of the variable’s values d. Fourth column in ordinal-level values- cumulative percentage/ % of cases at or below any given value e. Bar chart- graphic display of data f. Bimodal distribution- frequency distribution having two different values that are heavily populated with cases; two modes should be separated by at least one nonmodal category g. Percentile-