EXAM 1 Goal of ALL scientific research causal inference policy implications require it Epistemology study of how and why we know things how we acquire knowledge daily experience tradition authority o o mutual agreement everyone agrees its true o observation empirics Advantage of Observation vs other methods its verifiable and repeatable you can falsify it probability Political Science study of politics based on theory and observation we seek to generalize from our observation Goal of Scientific Research inference causal and descriptive definition best guess about something unknown given known information causal inference unknown causal relation between 2 or more variable descriptive inference unknown fact about a single variable inference is by definition uncertain it contains error Causal Inference causation cant be observed just inferred correlation when values of 2 or more variables tend to move together causation change in the value of one concept tends to produce change in the value of another concept Valid Causal Inference requires o Theory that expects X to effect Y o Evidence of correlation between X and Y REMEMEMBER causation x cause y implies correlation Values of x and y tend to move together BUT CORRELATION DOESN T IMPLY CAUSATION Deterministic Causal Relation cause x is always present when outcome y occurs Outcome follows cause without exception Scientific laws are deterministic ex law of gravity VS Probabilistic Causal Relation cause x is usually present when outcome y occurs Outcome occurs with some likelihood when cause is present Political science seeks to identify probabilistic causal relationships Descriptive Inference Good description offers us a profile of interest including o Units of interests individuals states countries o Variation o Both quantitative and qualitative judgments Description isn t the goal of science o Explanation of causal relationships is Why can t we just observe as scientists Observation without theory is inadequate for causal inference Why Inaccurate observations o o Tendency to overgeneralize o o Selective observation of phenomena Illogical reasoning To draw a valid causal inference we require both logical and empirical support Criteria of Scientific Method Evidence are the relevant data consistent with expectations of your theory Reason o Do the hypotheses from your theory follow from your assumptions o Emotion and passion influences scientists but don t overrule evidence and reason if one adopts the scientific method o Mutual agreement about the criteria of the scientific method prohibits emotion and passion undermining accumulation of knowledge Requirements for Scientific Study of Politics Analyze the real world not counterfactuals Think in terms of concepts constructs instead of proper nouns o Rather than Obama say presidents Propose testable causal relationships among concepts o Must be possible to falsify the proposed relationship o Must know what evidence is needed to possibly falsify hypothesis Fallacy of Affirming the Consequent Using evidence to prove a hypothesis is a logical fallacy If A then B hypothesis o o Observe B evidence o Conclude A is true fallacy of affirming the consequent Science of Politics If evidence is consistent with our theory we infer that our theory is not wrong we just failed to falsify the theory doesn t prove that its correct Political and social worlds produce observable phenomena that we can conceptualize and then measure we can produce political and social data Political scientists seek to identify and then explain the occurrence of patterns among political and social phenomena To do so we develop causal explanations aka theory about the cause and effect relationships among concepts and then test the hypotheses from those theories by examining relevant data Moving from Description to Causal Relationship We develop models theories to explain patterns to observe Model defined Simplified snapshot of reality Provides explanation of causal relations o o o Composed of interrelated concepts Causal Theory Answers questions what causes something and why Theory is a set of interconnected statements that identify what causes something and why o DV the something being caused the effect outcome o IV hypothesized cause Provides story about how and why change in the value of one concept influences the values of another concept Should also offer additional implications or hypotheses Constructing Theory Deduction theory first then empirical testing Induction observation first then theory construction to explain what we observed Hypothesis Expectation about what we will observe o More specifically a statement derived from theory that describes the expected co variation when two variables vary together between the values of two or more concepts Positive relationship values of x y move in the same direction both up or down o o Negative relationship values of x y move in opposite directions one up and other down and vice versa Hypothesis is a testable i e falsifiable proposition Specify nature of expected relationship not just the sign o More explicit hypotheses are easier to reject Hypothesis Testing and Null Hypothesis or doesn t change value A null hypothesis is implied by every hypothesis Falsifiability Null Hypothesis Expectation of no relationship between two concepts x and y varies randomly It must be possible to collect and evaluate evidence that permits us to demonstrate that the hypothesis is not consistent with the evidence If the data permit us to reject the null hypothesis then we can infer that our hypothesis is not yet wrong we can continue to use the theory This implies that all theory is provisional future evidence may lead us to accept the null Operationalization measurement To test hypotheses we must measure our concepts o Measurement permits us to examine data and evaluate co variation o You first need an abstract definition of your concept o Second you produce an operational definition set of instructions that describe how to measure the value of that concept Evaluating Operational Indicators Validity does it match the concept well Reliability do we get the same value when we repeat the measurement Error can be broken into 2 parts Systematic tendency to assign values that are either too high low o o Random equal likelihood of assigning too high low values Theory Construction and Evaluation Identify a question or puzzle Identify the DV Propose a causal explanation theory o o Identify the IV s
View Full Document
Unlocking...