a. Highly precise measures eliminate accuracies. Precisions leaves out key points.
b. Grading someone as an 88 out of 100 is precise, but difference between 87 and
88 is not much.
2. Orders of measurement
a. Categorical: assign people assigned entire to o
a. Select examples that are not representative. Select where we see the dependent
variable not all the possible situations.
b. Can reweight the sample to try and correct for some of the problems.
2. Phillip Zimbardo
a. Random assignment of subjects altere
a. They can suggest hypothesis for future reference to test. They can also better
understand the causal mechanism through which a phenomenon occurs.
b. Case studies can better establish theory. They can help prove causation. They can
better understand out
i. Convergent validation (with what does the measure covary): if I got
multiple indicators of the same concept, does my indicator say pretty
much the same thing as the other measures. Are other democratic
measures in general agreement with other attempts
a. States what she is going to be able to prove and what she cannot prove with her
2. Ziblatt deals with controls by introducing controls after research design.
3. Headings and subheadings are your friend
4. If true we would see, positive correlat
1. Literature review helps you understand what has already been done. That there is a
relationship. Paths not taken or paths not taken.
2. Concept formulation: need to make sure we have a specific concept. Some ideas are hard
to measure like power, human
a. Radical relativism: that all things are relative. Remington does not accept this.
We are not just imposing our own values, but these are objective values.
2. Observations are not always independent. Can have self-selection, hidden effects.
a. Did not buy arguments about cultures or that some leaders are better than others,
so the question became what makes leaders promote good economic policies?
i. Looking at what influences the preferences of leaders?
2. Reading correspondence was good rep
1. Survey research is a way to describe beliefs and behavior of individuals by asking
questions. Has to be systematic in order to allow generalizability
2. Important questions before doing survey. Also before choosing what dataset to use
a. What is the po
1. Statistical analysis wants statistically significant results. The results are not obtained by
chance but we are confident there is a relationship between what we are seeing.
a. Most of the time people stick with a 95% confidence interval.
b. When picki