Research Methods Exam 2 Study Guide Ch 5 6 7 8 Chapter 5 Ways to Measure Variables Most psychological phenomenon are hard to measure Depression success attraction stress How can we measure Construct Validity how well the scale measures the variable of interest accuracy Our variables can be defined conceptually and operationally Conceptual the researcher s definitions of a variable abstract Operational how the variable will be measured and manipulated specific There are 3 types of operationalization self report observation physiological information Self Report people answer questions subjectively about how they feel about certain topics Usually used in a questionnaire or interview Sometimes it is more accurate to have a self report measured filled out by someone who observes behavior versus someone who engages in the behavior Observation operationalizes a variable by recording observable behavioral or physical traces physical traces allow for data to be collected after the fact Physiological measures records biological data such as brain activity hormone level heart rate blood pressure etc Some people find physiological measures best Operational definitions can be Categorical categories often assigned numbers by researchers This is not really a measurement it is just a label for purposes of differentiating groups The numbers are arbitrary and meaningless Sometimes referred to as nominal Nominal Discrete categories Quantitative coded with meaningful numbers ordinal interval ratio Ordinal usually the best is assigned to number 1 Numbers are not inherently meaningful except to indicate less versus more The distance between the numbers mar or may not reflect equal intervals The first place team could be way out in from or barely winning Interval distance between stimuli Both the numbers and the distances between them ARE meaningful There is NO ZERO The distance between 1 and 3 is the same distance between 8 and 10 Example test scores Ratio has all the properties of an interval scale BUT the zero is meaningful The zero means nothing If you got a test of course content which you got zero right that does NOT mean you KNOW nothing But if you have zero dollars that means you HAVE nothing Reliability of Measurement Reliability all the measurements are consistent THREE TYPES OF RELIABILITY 1 Test Retest Reliability consistent scores every time we test 2 Interrater Reliability two or more independent observers will come up 3 Internal Reliability people who agree with the first item should agree with the same findings with the 2nd item and so one Correlation Coefficient also know as r single number to indicate how close the dots on a scatterplot are to a line drawn through them Positive they move together Negative they more opposite Strong correlation closer to 1 or negative 1 Weak correlation closer to 0 Cronbach s alpha way to see if their measurement scales have internal validity For this to be reliable the number must be 70 or higher Validity Face Validity simply looking if it should be a good measurement Content Validity must capture all parts of a defined construct Usually subjective Usually evaluated and scored by experts in the content area Convergent Validity should correlate more strongly with other measurements of the same construct A new measure of intelligence should correlate highly with the current standard Divergent Validity Should correlate less strongly with measures of other distinct constructs A measure of depression should NOT correlate highly with a measure that has been connected to self esteem If it does we have no idea if the new measure is really measuring depression self esteem or a combination of both Relationship between reliability and validity A measure CAN be reliable but not valid and vice versa Chapter 6 Frequency Claims Remember we are dealing with ONE variable opinions about a president number of homicides in a particular area within a city apathy of a prison population data comes from surveys polls data comes through observation Survey used to ask about consumer products Poll used when asking about an opinion most often used interchangeably Choosing Question Formats Open ended questions respondents can describe how they feel about a question topic Rich thorough data Forced choice format opinion is given by selecting the best of two or more options BUT what if the answer that I connect with isn t given Likert Scale tends to be the most popular a rating scale is used to answer questions For example Divorce is made too easy in the U S then you select Strongly Agree Agree Neutral Disagree Strongly Disagree Semantic differential format numeric scale anchored with adjectives Writing Well Worded Questions What to AVOID Leading Questions questions that prime participants toward a certain answer GOAL word every question as neutrally as possible Double Barreled Questions questions that ask one or more thing GOAL ask simple yet clear questions Double Negatives creates unnecessary complication Question Order to avoid this problem we must prepare different versions of the survey Encouraging Accurate Responses Response sets answering in a consistent way even if it is less than accurate Yea Saying saying yes or strongly agree with every item Nay Saying saying no or disagreeing with every item answer reverse word the question Fence sitting answering neutral or in the middle scale for every question Answer take away the neutral option Socially Desirable Responding faking good or bad to answer in a way that makes them appear socially okay Answer add filler items questions that have no meaning to hide the true purpose of the test Other issues with self reporting Sometimes participants don t recall their own behaviors attitudes correctly Sometimes participants don t know how they feel about something based on memory Makes behavioral observations necessary Chapter 7 Bivariate Correlations Associations between two variables Are the variables measured or manipulated Measured manipulated variables for causal claims Correlation studies can include both quantitative score on exam and categorical male female variables but they are graphed differently When evaluating an association claim the two most important validities are 1 Construct 2 Statistical Describing associations between two quantitative variables scatterplot Describing associations with categorical data bar graph When both variables are quantitative we utilize a correlation coefficient r positive move in the same