CHAPTER 5 Correlation Research Correlation two measures of the same person or thing may vary o Magnitude strength of the relationship 1 to 1 o Direction of r o A positive correlation means that as one increases so does the other negative correlation means that as one increases the other decreases zero correlation means there is no relationship o Scatter plot shows relationship in a graph Shared Variance correlation correlation Correlation mean association not causality Reliability consistent o Test Retest consistency of responses over time o Internal Consistency consistency of items within a measure 80 or better means the reliability is okay o Increase increase number of items used diverse sample rather than restricted testing environment must be free of distractions Validity truthful o Construct extent to which a measure accesses what is intended o Convergent extent to which a new measure correlates with measures of the same construct o Discriminant extent to which new measure correlates poorly with measures of different or unrelated constructs Nomological network and Validity Population Sample Representative Sample o Simple Random Sampling each person has equal chance o Stratified Random Sampling population divided into categories o Convenience Non probability Sample easiest sample to obtain Response Bias threatens representativeness of a sample when not all sample respondents complete the surveys Selection Bias selection procedures that under represent the population Margin Error estimated difference between sampled results and true results Cross Sectional Designs survey one or more samples at one time o Descriptive goals cant assess change predictive interpreting correlations easier and cheaper to do Successive Independent Samples multiple samples at 2 or more times o Access change in population rather than the individual representative sample is crucial for good results Longitudinal Designs same sample measured multiple times o Best way to access change respondents may be sensitized to survey can be expensive and time consuming can see changes in population Respondent mortality o What design does it affect CHAPTER 6 Experimental Research Designs True Experiment most are done in labs setting can help control of variables o Manipulate one or more independent variable to create at least two more conditions control all other variables by holding constant or balancing across conditions so that they differ only on independent variable measures effects on the dependent variable o Lab Experiments mostly done in labs so they control variables o Field Experiments real life only true experiments if conditions above Independent Variable manipulated variable o Experimental group treatment variable o Control group no treatment Dependent Variable effects of the independent variable Condition Placebo substance that looks like a drug or another active substance but is actually inactive and is used to manipulate the experiment Wait Control Internal Validity conclude that the IV caused differences on the DV o Covariation time order relationship eliminate plausible alternative causes of the outcome by using control techniques o Threats of Internal Validity Intact groups assigning existing groups to conditions but using individual scores as DVs Subject loss differential loss of participants Demand characteristics and Demand Characteristics cues and other information used by individuals to guide their behavior Confound something that varies along with the IV and could influence DV o Control confounds Hold conditions constant by having the treatment group have the same experiment as the control group and balancing by having participants in each condition similar Independent Groups Design separate groups of people in each condition o Random groups equate groups with random assignment o Matched groups equate groups with administrating a protest o Natural groups groups based on the individual difference variable o Intact Group assigning existing groups not individuals to conditions but using individual scores as DVs o Subject Loss differential loss of participants Repeated Measures Experimental Design participants are both experimental and control groups o Advantages fewer participants so it is more efficient less subject loss o Disadvantages practice effects balancing average practice effect across conditions know what they are going to do so they prepare o Counterbalancing determine all possible order of conditions randomly assigning participants to one of the possible orders o each condition must appear to each ordinary portion equally Differential transfer practice effects External Validity extent to which results can be generalized beyond the sample or experiment obtained by seeing if results replicate in all samples Analyzing Experimental Data o Outlier errors not okay to take out of data without explaining why o Effect Size strength of the relationship between IV and DV Null Hypothesis assume IV has NO effect on DV o If p inferential statistic is less than 05 reject IV and DV have effect o If p is less more than 05 do not reject null no IV and DV effect o Type I Error falsely conclude an effect occurred when it didn t o Type II Error falsely conclude that no effect occurred Bubble Hypothesis Statistically Significant differences between observed means is larger than would be expected by chance if null hypothesis were true Treatment Studies compare one or more treatments to each other and to an additional control group o Issues usually the control group is designed to mimic the treatment except for the effective ingredient in order to get thru ethical design often we used method of wait control Good Experiments internal external validity reliable sensitive CHAPTER 7 Complex Designs Factorial Designs conditions in which your experiment where each level of an IV is paired with each level of another IV Main Effects IV effects DV mean of the two Interaction Effects effects of one IV depends on the level of another IV o Occurs when effect of one IV on the DV differs depending on level of different IV occurs when IV alters effect of another IV on a DV o Test for interactions by comparing effects of one IV within each level of the other IV o If effects are different subtract the s than an interaction exists o Parallel lines say no interaction intersecting lines mean interaction 2X2 Design two IVs each has 2 levels Complex Designs studying 2 more IV at the same time o ANOVA analysis of variance used