Social Science Methods

Question 1

Social Science Research

Answer 1

1) Prediction
2) Control
3) Understanding

Question 2

Research can lead to control, without understanding:

Answer 2

1. Scurvy killed a million sailors between 17th and 19th century: they figured out that citrus fruit somehow stopped scurvy without understanding it
2. London Cholera epidemic: Figured out that Broad Street pump was spreading disease without understanding how

Question 3

Measurement of Social Science

Answer 3

a) Variables
b) Operational definition
c) Reliability
d) Validity

Question 4

Variables:

Answer 4

Things that one chooses to measure

i) Independent and dependent variables: independent is cause; dependent is effect

Question 5

Operational definition:

Answer 5

define a variable by specifying how it is measured—>allows people to replicate study

Question 6

Reliability:

Answer 6

i) Reliability
1. Interrater (similar ratings by raters is good)
2. Test-retest

Question 7

Validity:

Answer 7

If raters are getting all different scores (interrater) or different outcomes every test (test-retest), test lacks validity

Question 8

Internal validity

Answer 8

i. Derive your hypothesis, from a theory
ii. Formulate plausible hypotheses that can rival your hypothesis in explaining empirical results

iii. Eliminate these plausible rival hypotheses
1) London- Bus Drivers had more heart attacks than fare collectors
a) First hypothesis: Less exercise
b) Second: Fatter people got assigned to drive bus on first day on the job

Question 9

External validity

Answer 9

i. Validity outside of good study

i. e. research concludes something about juries in civil cases—>does it apply to criminal cases?

Question 10

Threats to External Validity

Answer 10

i. History

ii. Maturation
1) People mature and change over time

iii. Testing
1) If someone takes a test repeatedly, they will probably improve

iv. Instrumentation
1) Professor example: he used to grade flex exam in order they were submitted and the good grades were front-loaded

v. Regression

vi. Selection
1) Bar Prep course claims success but maybe people who take prep course are the type who are more likely to pass anyway

vii. Mortality

Question 11

Generalizations problems/External Validity

Answer 11

a. Generalization across people
i. Using students as jurors for studies
ii. Do they represent a real jury pool?

b. Generalization across places
i. Do studies on crime in Sweden work for a US model?
ii. Lower crime rate, fewer guns in Sweden

c. Generalization across time
i. Would a study of law students in 1970 be applicable in 2017?
Different pool of students?

Question 12

Sampling

Answer 12

i. Started with “food tester” trying king’s food for poison

b. Justification for sampling
i. More accurate than census
1) Study can be much more careful than census
ii. More stringent screening

c. Much less expensive
i. Last census cost $16B

d. Simple random sampling
i. Everyone has an equal chance of being picked
In order to make sure groups are equally represented: stratify first–>split into groups first then randomly sample–> i.e. split into groups of Catholics, Presbyterians, and Methodists first

Question 13

True Experiment

Answer 13

a. Not threatened by internal validity
i. i.e. People can’t choose their group

b. Assignment totally by chance
c. Doesn’t violate equal protection–>everyone has equal chance of being picked for a study

Question 14

Quasi-experimental studies

Answer 14

a. No random assignment

b. Before and after designs
i. toothpaste commercial

c. Simple comparison group designed
i. Selection bias because groups are not randomly assigned
ii. BAR review–>did they do better because they took it or because they were more motivated students?

d. Time series designs
Did bussing cause white flight or was it happening anyway?

Question 15

Mortality/Leave study

Answer 15

a. This causes problems with studies

b. We can’t guess what these people would have done had they lived/stayed in study

Question 16

Descriptive Statistics

Answer 16

a. Measures of central tendency
i. Mandatory mean grade (UVA)
1) Tells you nothing

b. Measures of variability
i. Grade distribution by professor
1) A lot of B+ or lots of A/B-
2) Bell shaped curve
3) Tells you a lot

c. Measures of association

Question 17

Inferential Statistics

Answer 17

1. ) Null Hypothesis
2. ) Type 1 error
3. ) Type 2 error

Question 18

Null Hypothesis

Answer 18

i. the hypothesis that an observed difference between the means of two samples is due to chance alone and not due to a systematic cause.
ii. For the Scientist: No relationship between variables
iii. For the Juror: Presumption of innocence.

Question 19

Type 1 error

Answer 19

i. report a real difference between 2 populations (i.e., reject null hypothesis), when in fact there is no real difference, only chance variation. Convention: p ≤.05
ii. i.e. UVA law students carry around more money than Darden students–> not true!
iii. To avoid Type 1 error: decrease the p-value (e.g., from p = .05 to p = .01), leaving less to chance

Question 20

Type 2 error

Answer 20

i. report no real difference between 2 populations (i.e., fail to reject the null hypothesis), only chance variation, when in fact there is a real difference
ii. i.e. UVA law students DO carry around more money than Darden students
iii. To avoid Type 2 error: increase the sample size.