hbs 301 exam 2

literature

a body of related work on a topic that is transmissible

literature review

a systematic examination and interpretation of literature for the purpose of informing further work on the topic; uses existing research to understand what new research is going to explore

peer review

process by which experts in a field evaluate a scholarly work to assess its quality, validity, and contributions before it is published or funded

mechanism for “vetting” the quality of work and enables the cumulative nature of science

types of peer review: blind review (single blind)

reviewers know who the author is, but the author does not know who the reviewers are

types of peer review: double-blind review

neither the author nor the reviewers know each other’s identities (most common in academia)

types of peer review: triple-blind review

in addition to author and reviewer anonymity, the editor handling the submission also does not know the author’s identity during the review process

research database

an organized collection of scholarly sources (journals, articles, and books) that can be searched electronically to find reliable info for academic or professional research

research question

a clear, focused, and specific inquiry that identifies the main issue a study seeks to answer, guides the research design, determines what data to collect, and shapes the analysis to gain new understanding or insights

theories

when we put concepts together and specify the way they fit together

specifying an explanation with fariables

independent variable x, dependent variable f(x) = y

operationalization

deciding how to record empirical observations of the occurrence of an attribute or a behavior using numerals or scores

substantive view of democracy

classify political regimes in terms of the outcomes they produce

procedural or minimalist view of democracy

classify political regimes in terms of their institutions or procedures (dahl’s view — multidimensional concept)

conditions: democracy-dictatorship

a country is a democracy if:

• a chief executive is elected

• the legislature is elected

• there is more than one party competing in elections (“contestation”)

• there has been an alternation in power under identical electoral rules

if these conditions are not satisfied, the country is a dictatorship

measure of democracy: democracy-dictatorship

dichotomous variable, minimalist

retrospective measure of democracy — cannot observe many of the conditions in real time

contestation

occurs when there is an opposition that has some chance of winning office as a consequence of elections

conditions for contestation

ex ante uncertainty: outcome of the election is unknown before it takes place

ex post irreversibility: the winner of the election actually takes office

repeatability: elections occur at regular and known intervals

conditions: polity

provides an annual evaluation of democracy and autocracy in a 10-point scale

democracy score minus autocracy score = polity score

score of 6 to 10 = democracy

score of -5 to 5 = mixed regime

score of -6 to -10 = dictatorship

attributes measuring polity

1. competitiveness of executive recruitment

2. openness of executive recruitment

3. executive constraints/decision rules

4. regulation of political participation

5. competitiveness of participation

anocracy

a form of government that mixes democratic and autocratic features, functioning as a partial democracy or hybrid regime located between democracy and dictatorship

open (1 to 5): democratic features outweigh autocratic

closed (-5 to 0): autocratic features outweigh democratic

conditions: freedom house

annual evaluation of the state of “global freedom”, not democracy

1. political rights — ability to vote and compete for office, etc

2. civil rights — free speech, etc

measures of democracy: freedom house

classifies countries as free, partly free, and not free

ordinal variable

levels of measurement

describe the amount of precision associated with a variable and the mathematical properties of the variable

nominal

describes variables that indicate a difference between categories

implies a difference in “kind” or “type”

very little precision, cannot be used with many statistical tools

example: marital status, eye color

dichotomous

subset of nominal, only two outcomes

example: gender (male / female)

ordinal

categories may be ranked in order, in addition to indicating differences between categories

implies a difference in “rank” or “order”

more precision and can be used with more statistical tools

example: education levels, likert scale

interval

includes all information of the preceding levels and adds meaningful intervals between values of the variable, but does not use a meaningful zero; cannot multiple or divide them

adds precision and can be used with most statistical tools

example: temperature (F) — 80 degrees is not “twice as hot” as 40 degrees

ratio

includes all information of the preceding levels and adds meaningful intervals between values of the variable, including a meaningful zero

most precise, can be used with most statistical tools

example: weight, age

nominal and ordinal variables

should be exhaustive and exclusive

exhaustive

all possible categories accounted for

exclusive

a single value can only fit into one category at a time, categories are distinct and do not overlap

experimental research design

good for isolating causal factors (does the independent var. actually impact the dependent var.?)

limitations to public affairs experiments

expensive, trying to study things that can’t be tested experimentally — ethical tensions / constraints or something else

often suffer from weak external validity

advantages of public affairs experiments

excellent internal validity

allows causal inferences to be made with a higher level of confidence — control over exposure to an experimental treatment

high degree of confidence in replicability, accurate representation of relationships

internal validity

the degree to which changes in the outcome are caused exclusively by the independent variable rather than outside factors

external validity

the degree to which study findings can generalize to other people, settings, and real-life environments

basic features of experimental design

1. at least one experimental group and one control group

2. randomly assigned individuals to each group, avoiding self-selection

3. controlled administration of treatment, including circumstances under which the experimental group is exposed

4. measurement of dependent variable before and after the treatment is administers (pre and post test)

• pre-test averages should be the same for control and experimental groups if true randomization was sued

5. controls the environment from the experiment (time, location, and other physical aspects)

impacts to internal validity

• history: events (other than treatment) that occur between pre and post test

• maturation: change in subjects over time that might produce differences in experimental and control groups

• test-subject interaction: process of measuring the dependent var. prior to the experimental stimulus may affect posttreatment scores

• selection bias: volunteers differ from nonvolunteers

  • experimental mortality: dropouts are likely not randomly distributed

• demand characteristics: participants may guess the investigator’s goals

alternative experimental designs: post-test

shares the characteristics of the classical randomized experiment, except no pre-test is used; sample is truly random and sufficiently large that one can assume control/experimental groups are equivalent

alternative experimental designs: repeated measurement

adds additional pre-tests, post-tests, or both in an effort to measure long-term effects of experimental treatments

alternative experimental designs: multiple groups

a modification of the classic example where more than one experimental group is created to compare the effects of different treatments

alternative experimental designs: field experiments

experiment is in a natural setting in which the investigator does not have control over group membership but does have control over one or more independent variables

research design

a plan of action for executiving a research project, specifying:

• a theory or hypothesis to be tested

• unit of analysis

• necessary observable data

research designs attempt to

• establish a relationship between one or more variables

• establish whether one phenomenon precedes another in time

• eliminate as many alternative designs as possible

• demonstrate that the results are generally true

characteristics of causal relationships

1. covariation: alleged cause varies with the supposed effect

2. time order: cause precedes the effect in time

3. isolation: elimination of alternative explanations isolate the causation to one factor

correlation

two things are related, but variation in one may not lead to variation in another

causation

implies that variation in one variable leads to variation in the other

spuriousness

two things are related, but variation in one does not lead to variation in another

Z causes X, Z causes Y —> X and Y trend together (therefore they are related), but they are not causal

validity in measurement

accuracy results

example: bullseye

reliability in measurement

likelihood of someone with the same information to reach the same conclusion

valid and reliable

centered on the bullseye

clustered on the bullseye over multiple attempts or trials

valid but not reliable

no cluster

one or few are centered

reliable but not valid

tight cluster

off-center

measures of central tendency

mean: average value

median: middle value

mode: most frequently occuring value

median

describes middle value in an ordered set of values, order of observations is important

resistant to outliers

only includes one value

mean

describes the average value

not resistant to outliers

computed using all values

mean notation

population: μ

sample: x bar

x bar (sample mean) = sum (sigma) of all values in a dataset (X) divided by the number of values (N)

mean

dispersion

how data is distributed over a range of possible outcomes

univariate (descriptive) statistics

look at one variable at a time

skewness

determines whether the mean or median is an appropriate measure of central tendency

how “lopsided” a sample distribution is compared to normal

outliers

values that do not need to go with the others

concentration

relative frequency of occurrence of a score

average deviation

0

graph is undefined at zero

standard deviation formula

STATA command: kdensity variable

displays kurtosis

STATA command: clear

clears the current dataset

STATA command: cd

changes working directory

STATA command: di

display

prints text, numbers, or results of a calculation in the Results window

STATA command: dir

lists the files in current working directory

STATA command: exit, clear

closes stata, discards any unsaved data in memory

STATA command: help

opens documentation for the command, shows syntax and options

STATA command: import delimited

reads a delimited text file into stata — like a .csv

STATA command: log

opens a log file

ex: log using filename

STATA command: log close

stops recording commands/output and saves the log file

STATA command: set obs

sets the number of observations (rows) in an empty dataset

STATA command: use

loads an existing dataset (.dta) into memory

STATA command: append

stacks another dataset beneath the current one, adding more obs.

STATA command: drop

removes variables or observations from the dataset

STATA command: generate

creates a new variable

STATA command: label variable

attaches descriptive label to a variable

STATA command: merge

joins one dataset to another by matching observations with a key variable and adding more variables (columns)

STATA command: rename

changes a variable name

STATA command: replace

changes the value(s) of an existing variable

STATA command: correlate

calculates the correlation coefficients between two or more variables

STATA command: describe

shows the structure of a dataset

STATA command: graph twoway

creates two-variable graphs like scatterplots and line plots

STATA command: histogram

draws a histogram showing distribution of a single variable

STATA command: summarize

reports summary statistics for numeric variables like mean, s.d., min, max

STATA command: tab

tabulate

produces a frequency table for one variable or a cross-tabulation for two

STATA command: ttest

tests whether a mean differs from a given value or whether two group means differ from each other

.smcl

log file

.dta

data file, data frame, data matrix

.do

do file, script file, replication file