Exam 2Chapter Four (Choosing Variables and Measurements)- Conceptual variable- an abstract idea that forms the basis of a research hypothesiso Ex. Study time, job satisfaction, temperatureo Circles in diagram- Measured variable- numbers that represent the conceptual variableso Ex. Seconds of study, Likert scale, monthly sales, degrees farenheito Squares in diagram- Operational definition- a precise statement of how a conceptual variable is turned into a measured variableo Ex. Conceptual- employee satisfaction Operational- # of days per month that employee shows up for work The lines from squares to circles are the operational definitions If the conceptual variables correlate and if the operational was correct, a correlation between measured variables will be seen tooo Converging operations- when there is more then one way to operationalize a conceptual variable depending on what is the focus of the research.  Helps “triangulate” the conceptual variable to get a fuller measure of itScales of Measurement- Measurement is a procedure for classifying data; the set of categories used for classification is called the scale of measurement- Nominal Scale- categories have different names but order does not mattero Reflects qualitative differences Sex, ethnicity Can be assigned numbers- Male-1, female-2 Statistical procedure- non parametric- Frequency counts, mode, chi-square (frequency differences) We can say that two people differ but you cant find an average of the differences- Ordinal Scale- categories have different names AND are organized sequentiallyo Ex. Horse raceo We can determine If there are differences The direction of the differences and/or rank of ordero We CANNOT say how big those differences are (doesn’t measure intervals)o Appropriate statistics: non-parametric Median, mode, rank-sum correlation, Wilcoxon rank sum test, non-parametric ANOVA Ex. Taste test between 3 sodas- Can tell how much more one is liked over the other but not by how much- Interval Scale- categories are organized sequentially and all categories are the same sizeo Has an arbitrary zero point (not a true zero) Temperature in Celsiuso Can determine how large or small differences areo Statistical procedures: parametric Mean, median, mode, standard deviation, correlation, t-test, anova- The Ratio Scaleo Consists of equal, ordered categories, anchored by a non-arbitrary zero point (true zero) Ex. Weight, percent correct, reaction time to a stimulus Zero is the variable being measured and can help find multiples Appropriate Statistics: Parametric- Mean, median, mode, standard deviation, correlation, ratios, t-tests, ANOVASummary of Scales- Nominalo Can say only if a difference exists (sex, religion)- Ordinalo Can tell us the direction of the difference (soda test, horse race)- Intervalo Can tell us the direction and size (Celsius)- Ratioo As above and ratio of the difference (true zero KELVIN)Factors Affecting Your Choice of a Scale of Measurement- Information Yieldedo Nominal scale yields the least informationo Ordinal scale adds some crude informationo Interval and ratio scales yield the most information- Statistical Tests Availableo The statistical tests available for nominal and ordinal data (non parametric) are less powerful than those available for interval and ratio data (parametric)- Ecological Validityo Sometimes your research requires you to use a measure that reflects what people do in the real world (a nominal guilty/not guilty verdict)o Such scales have ecological validityo When necessary, choose an ecologically valid measure, even if it means loss of infoo These are things you can do to gain needed information Include an additional scale (interval) Create a composite scale by combining a nominal and interval scale Incorporate features of ecologically valid scale into a more informative scaleAdequacy of a Dependent Measure- Adequacy of a dependent measure is related to the sensitivity of the measure and range effects- Sensitivityo Is a dependent measure sensitive enough to detect behavior change? Marital happiness- Divorce or # of arguments per montho An insensitive measure will not detect subtle behaviors- Range Effectso We often run into range effects when our tests are too easy or too hardo Occur when a dependent measure has an upper or lower limit Ceiling effect- when a dependent measure has an upper limit Floor effect- when a dependent measure has a lower limito Affect data in two ways Limiting values of your highest or lowest data point Variability of scores within affected treatments is reduced- May cause misleading results from statistical analysis of dataTypes of Dependent Variables- Behavioral Measureo Record actual behavior of subjectso Several types Frequency: count of the number of behaviors that occur Latency: the amount of time it takes for a behavior to occur Number of Errors: the number of incorrect responses madeo May not allow you to determine the underlying cause for behavior- Physiological Measureo Physical measure of body function o Typically requires special equipmento Most physiological measures are non-invasive Some require an invasive procedureo Allow you to make precise measurements of arousal of a subject’s bodyo Must infer psychological states- Implicit Measureso Implicit Associations Test (IAT) is a popular example of an impact measure ofracial bias Explained themselves as “self” or “other” and “panic” or “calm”o Disguise the goal of the test- Free-Format Self Reporto Allow them to respond with no constraintso Projective Measures Thematic Apperception Test (TAT)- Free response to picture of people that show emotions indirectly in the images instead of within themselves Rorschach Ink Blot Test- Free response to imageso Associative Lists Study stereotyping Listed races and people wrote descriptive wordso Think-Aloud Protocol Say out loud thought they have while completing a tasko Coding Issues Empirical Support is critical Hard to analyze data and give it measured variables because it reduces freedom  Content analysis- process of coding free-response data- Self-Report Measureso Participants report on their own behavior or state of mindo Self measures are popular and easy to use, but may have questionable reliability and validity You cannot be sure that a participant is telling the trutho Fixed-Format Self-Report Likert (scale