Research Methods - Google Docs.pdf
Transcript
esearch Methods 1 R Two Reasons to Love Research Methods It allows you to get at the why question It allows you to critically evaluate the conclusions that others have reached Step 1: Coming up with a research question, and making a prediction about the expected r...
esearch Methods 1 R Two Reasons to Love Research Methods It allows you to get at the why question It allows you to critically evaluate the conclusions that others have reached Step 1: Coming up with a research question, and making a prediction about the expected relationship between variables (i.e., a hypothesis) Example: Research question: Is there a relationship between watching violence on TV and aggressive behavior in children? Translate into a testable prediction: ○ H0: There will be no differences in the number of aggressive behaviors between children who watch a violent TV show and children who watch a nonviolent TV show. ○ H1: Children who watch a violent TV show will exhibit more aggressive behaviors than children who watch a nonviolent TV show. Step 2: Design a study to test your hypothesis! First, operationalize your variables ○ Define your variables in a way that allows you to test your hypothesis ○ Example: H1: Children who watch a violent TV show will exhibit more aggressive behaviors than children who watch a nonviolent TV show. How should we operationalize each of these variables? Different types of designs: ○ Experiments ○ Correlational studies Determine whether a causal relationship exists between two or more variables. Hallmarks of Experiments ○ Independent Variable – manipulated by experimenter, hypothesized to cause some effect on another variable Different conditions are called different levels Important to have a control group, which does not get the manipulation, as a comparison ○ Dependent Variable – also called the “outcome” variable, it is the variable that is hypothesized to be affected ○ “We want to test the effects of ________ on ________.” Example: Children who watch a violent TV show will exhibit more aggressive behaviors than children who watch a nonviolent TV show. I ndependent Variable: Type of television show (violent vs. nonviolent) Dependent Variable: Number of aggressive behaviors (as operationally defined) ○ Random Assignment: each person has an equally likely chance to get placed in each condition ○ In experiments, we try to vary one factor (the IV) and keep other aspects of the situation constant. Only then can we say the IV “caused” the DV. ○ Studies can be between-subjects or within-subjects Between-subjects = participants are in separate conditions (i.e., they are exposed to only 1 level of the independent variable Within-subjects = participants are in multiple conditions (i.e., the same subjects are exposed to multiple levels of the independent variable) ○ Studies can be blind or double-blind Blind experiment – subject is ‘blind’ to treatment condition Double-blind experiment – Observer and subjects are both blind to treatment condition Causation 1. Covariance ○ The IV and the DV should covary/co-occur/be correlated 2. Temporal precedence ○ The IV (the cause) should clearly come first in time, before the DV (the effect) 3. No plausible alternative explanations ○ Keep as much constant as possible between the experimental and control conditions Research Methods 2 Correlational Designs Correlational research measures statistical relationship between two or more variables. (No experimental manipulation of variables; no IV) Correlation coefficient (“r”) – ranges from -1.0 to +1.0 Positive correlation = when one variable increases, the other variable increases Negative correlation = when one variable increases, the other variable decreases Correlation does not equal causation Key Concepts for Research Reliability→Does your measure consistently achieve similar results? Validity→Is your measure accurate? ○ Construct validity→Do you have good operationalizations? ○ External validity→Do your results generalize? ○ Internal validity→Can you rule out alternative explanations? Step 3: Analyze the data Results support the hypothesis or results do not support the hypothesis? Statistics = measurements of samples, in order to make an inference about the broader population Descriptive Statistics Central tendencies = summarizes the entire data set ○ Mean = “average” (sum of scores / number of scores) ○ Median = when data are ordered from lowest score to highest score, median divides the group of scores in half ○ Mode = most frequently occurring score(s) Variability = how the sample is spread out from the mean in one or both directions ○ Standard deviation (average deviation, or difference, of a score from the mean) 68% of data falls within one standard deviation of the mean 95% of the data falls within 2 standard deviations of the mean Inferential statistics Allow us to generalize findings from our sample to our population Establishing the confidence that results are not due to chance (determine the likelihood of obtaining a particular value for a sample, given that the null hypothesis is true) Probability of a chance finding (p-value) is the significance level Convention is that we accept 5% chance or less Step 4: Interpret the results and plan further research
