Designing Social Inquiry Understanding What to Avoid Jaechun

Designing Social Inquiry Understanding What to Avoid Jaechun Kim

n In previous class, we discussed how we should design our research… we talked about indeterminate research design (small N problem and multicollinearity), selection bias, etc. n Today, we will discuss things we need to avoid when we make inferences…when we measure variables… briefly…

n Making Observations - Two criteria for judging the soundness of measurement n Un-biasedness - validity Efficiency - reliability… n Remember dart-throwing example… ! n Increase reliability (efficiency) and reduce bias – that is the goal that you should pursue when you measure your variables… n

n Types of Measurement n Nominal measurement Observations are grouped into a set of categories n Presidential system vs. Parliamentary system or… n n Ordinal measurement Observations are grouped into categories that have ordering schemes n Examples? ? Countries with degree of industrialization n n Interval measurement – continuous variables… n n Example – annual economic growth rate or GDP, etc. Ethnicity in the US - nominal or ordinal measurement? ?

n p. 153 n Use the measure that is appropriate to your theoretical purposes… !

n Use measurement made by other researchers for entirely different purposes in order not to craft your variables to justify your hypotheses… e. g. , Democratic Peace scholarship uses Polity data set n Polity IV Project http: //www. systemicpeace. org/polity 4. htm n Grouping error. . What is it? P. 153 n n n Effect of Age on Hours of Sleep Needed… Young-aged group, middle-aged, old-aged groups… Why is it problematic? ?

Measurement n Don’t worry too about the parts on nonsystematic and systematic measurement error. Just browse the parts on these subjects in KKV…!!! n n Error… p. 155 on… Systematic measurement error n n What is it? Consequence? It will bias descriptive inference (But systematic measurement error that affects all units by the same constant amount causes no bias in causal inference…)

n Nonsystematic Error Measurement n What is it? n Nonsystematic measurement error in the DV Consequence? n Does not bias the usual estimate of the causal effect, but it does make the estimate less efficient

p. 161 figure here!

p. 165 figure here!

n n n An example of omitted variables problem pp. 177 -178 Impact of summit meetings Cooperation between the USSR and US n Any antecedent variables? n Econ interests or military interests? ? n Trust? n Convergence of geopolitical ambitions? n Any other domestic political reasons?

n Problem of Endogeneity n This problem occurs when the values of EVs or IVs take on are a consequence, rather than a cause, of DV… n Examples n P. 186 Constituency services Votes n No real positive evidence supporting this. Why?

n Max Weber’s The Protestant Ethic and the Spirit of Capitalism n What is Weber’s causal hypothesis? n What’s wrong with this? n P. 191 Causal Impact of Ideas… n Ideas n can be consequences of policy… P. 192 Halpern (1993) n What is her causal hypothesis?

n Stalinist planning doctrine (idea) economic policies after WWII n Alternative n hypothesis? Soviet military might and influence n China and Yugoslovia? ?

n pp. 193 -195, the last paragraph n What is the research question of the authors (Verba, Schlozman, and Brady)? n What is their causal hypothesis? n Going to church Political Participation n Why can it be problematic? n How did the authors resolve this ‘endogeneity’ problem?