Agenda 2 28 08 Review of key concepts from recent material New lecture material Causality and experiments 2 28 08 1 Warming up What if I told you that studying with a partner would help you do better on your final exams and you were a little skeptical about my claim and interested in better understanding the how this is supposed to work What kinds of Qs about the causal relationship would help you figure out whether or not you think this is something that you d like to try 2 28 08 2 What sorts of things would you be curious about Is the ultimate payoff big enough to be worth going through all that trouble I m a little skeptical about the connection here does this help mainly by forcing me to schedule in study time or is there something about talking to another person that is special How much does this study method help all other things being equal How much will it help as compared to pulling an intense all nighter or any OTHER special efforts I make how much MORE powerful is this method than others How focused and consistent could I really expect myself to be at this in normal life 2 28 08 3 1 Quick Story Younger me yikes 2 28 08 4 New Topics to be Covered Experiments Confounding Variables Subject Procedural Other Controlling for Confounds Strategies Study design features Experimental Validity Internal Validity External Validity 2 28 08 5 Quick Review 1 of 7 Just to prime your memory for these things but let s not get too hung up on them for now Testing for differences between sample means what is this How are the variables treated What question do you ultimately want to answer Type 1 error and Type 2 error what are they how do researchers try to reduce the risk of them Statistical significance P value Alpha levels Power How can researchers increase power 2 28 08 6 2 Quick Review 2 of 7 Independent variables dependent variables Operational definitions a k a operationalization Necessary and sufficient causes Why are they useful to identify when possible How can you evaluate claims about necessary and sufficient causes Deterministic vs indeterministic causes Partial or contributory causes How can you evaluate claims about these when they re concerning types of events and concerning particular individuals Counterfactual analysis of causation What is it When might you try to use it 2 28 08 7 Quick Review 3 of 7 Proximal vs Ultimate causes Causal overdetermination what is it how it relates to the counterfactual analysis of causation Mill s methods for forming causal hypotheses Be able to read and draw diagrams of causal relations according to the system articulated in the course reader 2 28 08 8 Quick Review 4 of 7 Be able to identify examples of some of the common errors of reasoning about causal relations wrongly assuming or neglecting common causes mistaking causes for effects neglecting circular causal relationships post hoc ergo propter hoc 2 28 08 9 3 Quick Review 5 of 7 Match struck yes no Match tip temperature 350 350 Match lit yes no Diagramming causal relations Variables as nodes boxes Causal relations as arrows Not tracing the flow of activity but causal relations If there are conditions under which changing one variable will result in change of another variable include a arrow between the variables Sometimes there are important intermediate causes such that a more ultimate cause only produces its effect through a more proximate cause 2 28 08 10 Quick Review 6 of 7 Common cause A positive correlation between two variables may be the result of a common cause for both Pine needles on tree dropped Fish alive dead Toxic waste no yes 2 28 08 11 Quick Review 7 of 7 If a causal relation is direct there should be no way to screen off the effect from the cause 2 28 08 12 4 Experiments QuickTime and a TIFF Uncompressed decompressor are needed to see this picture 2 28 08 13 The basic idea of an experiment If the independent variable is a cause of the dependent variable manipulating the independent variable should change the value of the dependent variable If it weren t a cause we wouldn t expect such a result from manipulation Manipulation Independent variable values Dependent variable values 2 28 08 14 The basic idea of an experiment 2 Just to keep you on your toes What if the dependent variable is causally overdetermined Is it always easy to manipulate JUST the independent variable you re interested in and nothing else Manipulation Independent variable values 2 28 08 Dependent variable values 15 5 Experiments on regular deterministic systems When there is no variance in the population being studied statistical analysis isn t necessary The main danger is affirming the consequent The key is to test a causal hypothesis in which it is unlikely for the effect to occur unless you were right about the cause 2 28 08 16 Variability in nondeterministic systems Different systems of the same type or the same system different times will vary in their responses to a manipulation depending on Their particular composition and history Effects of a prior manipulation Interaction of the manipulation with other relevent variables You might also see variability in your data due to Imprecision in the manipulation or in your data collection Unknown extraneous variables affecting responses Challenge how to detect and learn about causal relations in the face of background variability 2 28 08 17 Pretend we tried our study buddy experiment on everyone in the class Randomly we assign 1 2 the students to study in groups 2 hrs wk the other half don t do this No other manipulations or instructions We will compare the Midterm Score with the Final Exam Score Even assuming that studying in groups has some some effect on the Finals scores would you expect everyone s scores in each group to change exactly the same amount e g everyone in the study buddy group improving by 10 points everyone in the other group only improving by 2 points what sorts of things help explain why that wouldn t be likely to happen 2 28 08 18 6 Variability can be your friend On one hand it can make things complicated On the other hand you can LEVERAGE variability in data to get a lot of information In my opinion the only kind of variability you shouldn t like is the kind that s just noise or error due to sloppy experimental design and procedure In fact decreasing this is another way to increase power But variability that s due to real differences between individuals can be incredibly informative 2 28 08 19 Using variability for information