Methods and Designs, week 6
Methods and Designs, week 6 PSYC 328
Popular in Methods and Design in Behavioral Science
Popular in Psychology (PSYC)
This 4 page Class Notes was uploaded by Hope Good on Sunday October 9, 2016. The Class Notes belongs to PSYC 328 at University of South Carolina Aiken taught by Dr. Weed in Fall 2016. Since its upload, it has received 4 views. For similar materials see Methods and Design in Behavioral Science in Psychology (PSYC) at University of South Carolina Aiken.
Reviews for Methods and Designs, week 6
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 10/09/16
Chapter 5: Variables and Measurements in Research Variables can be described in several ways! o Labels: independent, quasi-independent, subject, predictor, dependent, outcome, extraneous, confounding. o Factors= independent, quasi-independent, subject and predictor Categories: presence or absence, type or amount Levels: bivalent or multivalent Independent and other variables o Bivalent: Independent variable with 2 levels; design is considered bivalent if it only contains one independent variable Present IV or absent IV o Multivalent: independent variable with 3 or more levels; design considers multivalent of there is only one independent variable containing many levels Using a medicine in three groups with different mg o Presence/absence: typically, 2 levels with one being present and the other absent Present: experimental group Absence: control group Bivalent o Type: different type of attribute being manipulated in each level Bivalent or multivalent Ex: different drug/therapies being compared different types of people o Amount: involves manipulation Can investigate amount of drug optimal for relieving symptoms (given different amounts) Bivalent or multivalent Describing o How many factors? o What are the factors? o Present/absent/type/amount? o Levels? (bi/multi) Scales of Measurements o Nominal: non-ordered category response How do you feel? Happy, sad, anxious, excited, mad o Ordinal: ordered category response How do you feel? Very sad, sad, neutral, happy, very happy o Interval: equally spaced, numerical responses, values are not ratio relations Rate how happy you are from 1-5 1 = least happy; 5= most happy o Ratio: equally space, numerical response, ration relations Time Height Accuracy o EXAMPLES: Amount of mount money paid for tuition? Ratio Hobbies listed from least to most interest? Ordinal Graduate school; yes or no? nominal Percent correct on exam. Ratio Enjoying class rating (1-5). Interval List of career options. Nominal My class rank; freshman to senior year? Ratio Validity Scale o Construct Validity: scale measures the behavior it was designed to measure A measure with high construct validity will provide an accurate measure Need to consider the developmental level of participants (children, teen adults) How would love be measured? Back Depression Inventory Sample item: circle number that applies o 0 don’t feel sad o 1 sad o 2 feel sad all the time and can’t snap out of it o 3 so sad and can’t stand it o Face Validity: intuitively, scale appears to have high validity (accomplishments, surveys) Means on the surface the scale seems to measure what you think it does Reliability Scales o Inter-rater reliability: different observers are measuring behavior consistently Naturalistic observation Internal v. External o With more control over factors in a study, internal will increase but behavior may become more artificial and decrease external validity. o Experimental validity: bias may limit internal and external validity Source Bias: Internal Validity o Decrease internal validity o Good independent variables: good test for: Causal hypothesis Controls for those extraneous factors that could affect data but not of interest o Good independent variable with good causal relationships by removing alternative explanations of data o 5 SOURCES OF BIAS OF INTERNAL VALIDITY Experimenter Bias: occurs when a researcher treats group differently in the study due to knowledge of hypothesis for the study Single-blind: used to combat effects of subject’s knowledge of their group assignment. Double-blind: neither subject nor experimenter are aware of assignment Order effects: some easy or difficult tasks administered first can affect how participants do on later tasks Test effects: can be an issue when multiplying test sessions occurs in study More likely to when studies test more than once or in a series within subject since every person receives level of independent variable Regression: towards means occurs when manipulation participant obtains an extreme score on a questionnaire/task at one testing session but regress towards means score at another test session Regression towards mean can be problematic in studies where a test or questionnaire is given more than once st Can score higher on 1 test and lower on test the next time Group differences: despite random assignment conditions, sometimes groups are not equal on important characteristics Sometimes when you flip a coin 100 times you gets 80% heads an another times it is less or more. Source bias that impacts External Validity o Hawthorne Effect: where participants in the study know that there are participating in a study and may adjust behavior accordingly May worker harder of knowing they are being observed or monitored o Demand characteristics: occur when participants in a study attempt to figure out purpose of study and change their behavior based on their view of the purpose Can be correct or incorrect on their belief of purpose