PSY 355 1st EditionExam # 2 Study Guide Psychology Research Process Logical Processes 2 major ways in Psychology 1. Inductiona. How you’re taught as a child to observe reasonb. Observation, associating with other things, correlationsc. Patternsd. Examples: Naturalistic observation, surveys, qualitative studies (focus groups)e. Strengths Helps build theoriesf. Weaknesses Unable to show cause and effecti. Unable to tell when you are wrong (don’t have a formal way of trying out your observations—correlation does not mean causation)2. Deductiona. Using premises from theory to test theoryb. Example: Rape blame study– Women are held responsible when drinking• Woman is drinking• All other things are held equal (controlled)• She will be held responsible (more than when not) c. Derive Hypothesis d. Test Hypothesis by setting up situation in which premises occur – What if hypothesis shown to be correct? – Support for theory (cause and effect)– However, problem of confounding variable – What if hypothesis found not correct? • 1. Maybe theory is wrong • 2. Maybe something you did not control for (hold constant) had a strong effect on results (caused much variability) • 3. Statistics help with #2; replication helps with #1 Lecture: 2/2Confounding variable- Can’t attribute results to variable you set upTheory: An organized, systematic explanation of a phenomenon Hypothesis: A more specific application of the theory • Independent Variable: – A (hypothetical) cause that the experimenter manipulates • Dependent Variable: – The (hypothetical) effect that the experimenter expects to see Operational Definition ―A definition that is specific enough to be used to actually perform an experiment (Pg 46)-Must be measurable Must have some validity***EXAMPLE1. 1) Effect of Alcohol Intoxication on Braking Distance When Driving 2. 2) Effect of Ambient Room Temperature on Activity Levels of Mice 3. 3) Do Specific Examples in Class Produce Better Learning in Psychology Classrooms? 4. 4) Instituting “No Smoking” Policy in a Restaurant: Does It Change the Clientele? Examples: Independent and dependent variables1. IV: BACDV: Brake distance2.IV: room tempDV: activity level3.IV: Specific examplesDV: changes in learning4.IV: Smoking policyDV: Number of clientele Lecture: 2/4Examples: Operational definitions1. Distance of how far from when you brake to when the car stops (measure of skid mark)2. Heart rate of mice3. Grade point average4. Number of customers Examples: Hypotheses-Guess based on theory-What you are testing in an experiment 1. The higher the level of alcohol, the slower a person will be to brake2. As room temperature rises, mice become less active3. In classes with more specific examples, students will have higher GPAs4. Instituting a smoking policy will cause more people to enter a restaurant Between Groups Vs. Within Groups Designs • Example: Braking Distance study -100 undergrads (random assignment) – 50 sober; 50 intoxicated -Distributions do not overlap—everybody who is intoxicated takes way longer to brake-Therefore, hypothesis is supported. - Example: Braking Distance study - 100 undergrads—random assignment – 50 sober; 50 intoxicated - But what if distributions overlap because, even sober, braking distance times vary so much... - How about a Within Groups Design?? - How would that work? - But what might be the problems with it? Review• Induction vs. Deduction• Theory • Hypothesis• Independent Variable• Dependent Variable• Confounding Variables• Operational Definition• Between Group vs. Within Group design Descriptive Statistics- describe and summarize the dataMeasures of central tendency Mean Median ModeMeasures of variability RangeVarianceStandard Deviation -Measures of Association between two or more variables -Pearson Product-Moment correlation - (symbolized as r). -Most common measure of association -Describes how strongly variables are related to one another. Inferential Statistics- test the hypothesis- Hypothesis testing (What does it all mean?) Your hypothesis versus the null hypothesis o t-test – to compare means of two groups o F- test – to compare means of more than two groups. (Analysis of Variance) o Chi-square – to compare frequencies (e.g., how many men versus how many women?) o Pearson’s r-test – to investigate whether there is a linear relationship between two continuous variables. - Regression – to use one predictor variable to predict a criterion variable. - Multiple Regression – to use more than one predictor variable to predict a criterion variable. - Partial correlation – to partial out the effects of a third variable that is influencing the relationship between two variables. - Semi-partial correlation – to partial out the effects of a variable that is influencing only one of the other variables. Data Analysis and Presentation -The statistical power of the study:-Your hypothesis says that (all other things being equal) if person hears conversation, participant will make more mistakes than if he/she does not hear conversation (Experimental hypothesis) -Null hypothesis says (what?) -Your Finding: If no difference, does not support your hypothesis— BUT does not prove the Null Hypothesis (you would say that the experimental hypothesis was ―not supported) Suppose instead 1. You did not run the session long enough? (e.g. instead, participants had 500 math problems and heard conversation for 5 minutes—more chance to actually make mistakes) and/or 2. There was a between-group difference that was consistent but very small***, so you needed more participants to find it (to separate the ―signal from the ―noise) If you concluded that your hypothesis was wrong THEN YOU HAVE COMMITTED A TYPE II ERROR Type I Error is: -Thinking you disproved the Null Hypothesis because you hit that one chance in 20 Type II Error is: -Accepting the Null Hypothesis when, in fact, your hypothesis was correct To avoid Type II Error: Increase statistical power How would you increase statistical power?1. Run more participants (computer programs help you decide this before you begin and/or 2. Use a better, perhaps more precise, operational definition of the dependent variable As Power increases, the probability of Type II Error decreases Type I vs Type II ErrorsTrue State of Affairs Null is true Null is false Reject the null Type I error (α) Correct inference (power)Fail to reject the null Correct