anylzing politics studyguide

regression table

Regression is a way to see how one thing affects another — like how your hours of sleep might affect your test score.

positive number on regression table

More of this ➡ more of the thing we’re measuring

indicates a direct positive relationship between variables.

purpose of a regression table

A regression table is like a report card that shows how each of these factors is related to your test scores.

coefficients

Values representing the relationship between independent variables and the dependent variable, These (coefficients) numbers tell you how much that factor affects what you’re studying

stars on a regression table

Indicate the significance level of the coefficients, showing whether the relationships are statistically significant.

ex.The more stars, the more statistically significant the result. 3 stars = very important, no stars not important

Regression vs. Correlation

negative number on a regression table

More of this ➡ less of the thing

What’s the Central Limit Theorem

If you take a bunch of random samples from a big group, and you calculate the average of each one, those averages will start to form a normal (bell curve) shape, even if the big group is messy or weird

exThis means that as more samples are taken, the distribution of the sample means approaches normality, facilitating inference about population means.

Limitations of experiments

Researchers can't always assign a particular treatment, ex. Studying war is not something you can apply as a treatment

Ethical constraints (must make sure the experiment brings no harm to recipients, intentionally or unintentionally

Experiments are low in external validity (how general the results are)

TREATMENT GROUPS & CONTROLL GROUPSED

The treatment group gets the thing being tested (like a new medicine, teaching method, or product).

The control group does not receive the treatment and serves as a baseline to compare the effects of the treatment.

the controlled group

in an experiment that does not receive the treatment, allowing researchers to measure the treatment's effects against it.

does not get the thing being tested—they stay the same, so you have something to compare the treatment group to.

the treatment group

is the part of an experiment that receives the intervention or treatment that researchers are testing, allowing for comparison to the control group.

gets the thing being tested (like a new medicine, teaching method, or product).

the gold standard in research design

Randomization experiments

Eliminates all alternative explanations and allows the treatment to happen before the outcome

probability

A study of events and outcomes involving some element of uncertainty doesn't tell us what is going to happen, but tells us what is likely to happen and less likely to happen

how to calculate probability (practice on paper)

Calculate by #number of outcomes that satisfy the condition/ possibly outcomes

The calculation involves dividing the number of favorable outcomes by the total number of possible outcomes. ex: rolling a die, the probability of getting a 4 is 1/6.

monty hall theory

A game where switching your choice after one wrong option is shown gives you a better chance to win — from 1 out of 3 to 2 out of 3.

independent events in probability

Don’t affect each other ex.Rolling a die twice

Flipping a coin and then drawing a card

Picking a random student in class, then picking a different one without telling the first

• “Rolling a 5 doesn't change my chances of rolling a 2 next.”

Populations and samples: population:

the totality of the events we wish to explain both observed and unobserved

ex. if youre studying Health Outcomes After a Vaccine

Observed: People in your clinical trial

Unobserved: People who got the vaccine but weren’t part of your study

Population = everyone who received the vaccine, whether you studied them or not.

Let’s say you're baking cupcakes and each one turns out a little different — some are too small, some too tall, some burnt (the population is messy).

Now imagine:

You make 30 batches of 10 cupcakes each

Each time, you measure the average height of that batch

Even though your cupcakes are all over the place, the average heights from each batch will start to look like a nice smooth curve

central limit theorem

why is central limit theory important

Use averages to understand a big group

Make predictions, even if the original group is not perfect or normal

Create confidence intervals and do hypothesis testing

ex.The more samples you have the higher change the data will reflect the true population

what needs to be true for central limit theory to work

The samples must be random (to make sure there arent any other factors that are determining the outcome)

The sample size should be big enough (at least 30 is best)

You should take many samples (not just one)

what is a weird population

a group of data that doesn’t follow a nice normal (bell curve) shape. Instead, it might be, skewed, spiky, flat, or randomly lumpy

why does weird population matter for central limit theorem

Even if the population is weird:skewed, chunky, or uneven

The sample means will still form a normal distribution when you take enough random samples

Random samples

systemic samples

A systematic sample is when you pick every nth person from a list or group.

You have a list of 100 students.

You decide to survey every 10th student

Stratified

when you split a big group into smaller sub-groups (called strata) based on something they have in common, and then you randomly sample from each group.

Politics:

Split voters into political parties (Democrat, Republican, Independent), then randomly sample from each.

A non-random sample

when people are not chosen by chance — meaning some people are more likely to be picked than others;picked by ease, not chance

✅ Example: Surveying your friends because they’re nearby.

Snowball Sample:

You ask one person, then they refer you to others.

✅ Example: Interviewing one protester, who introduces you to others.

Voluntary Sample:

Only people who choose to respond are counted.

✅ Example: Online polls where only interested people vote.

ex.

risks of non random samples

It may not represent the whole population.

It can lead to biased results.

common polling and survey errors

Bad wording ex.“Don’t you agree the president is doing great?” = biased

Priming ex. One question changes how someone answers the next

("How unsafe do you feel?" followed by "Should crime policies be stronger?")

Aggregation Bias ex. One group is overrepresented and skews results (like if 80% of your sample is from one major)

Non-response bias ex. Only certain people respond — others don’t (and that matters)

Qualitative research

does not use numbers and analyzes things like interviews and such, interprets nonnumerical data

ex.ethnography, interviews, case studies

Empirical research

focused on validating and analyzing data, ex. Math and numbers

traits and categorization of qualitative research

Offers a more in-depth nuanced understanding

Easier to trace causality

Harder to generalize

Analysis can be biased through subjectivity

Does not always follow scientific process

traits and categorization of quantitative research

Always and clearly follows scientific process

General and more broad patterns

Identifies exact changes

Is the standard method of analyzing

Difficult to numerically capture certain concepts in political science

Small N observational studies

1–2 cases

Used in comparative politics

Good for rare events

Qualitative-heavy

Large n Observational studies

Many observations

Used in generalizable studies

Risk of omitting key variables

Quantitative-heavy

P-value

probability that result happened by chance

Lower p =

greater statistical significance, stronger evidence against null hypothesis

higher p =

weaker statistical significance, weaker evidence against null hypothesis

null hypothesis (H0)

No difference, no effect “Nothing’s going on”

alternative hypothesis (h1)

There is a difference, Something’s going on!”

observational

taking the world as it is and observing it, explanation to how the real world works, by collecting data from real-world events

when a researcher manipulates something (the treatment) and then measures the effect it has (often done in a lab setting) to establish cause-and-effect relationships.

The experimental design ensures that the test group and the control group are the same in every way except one:

the treatment applied to the test group.

independent variable

the factor that is manipulated to observe its effect on the dependent variable.

dependent variable

The thing you measure, ex. if your independent variable is How much someone studies, the dependent variable is Their test score

spurious variable

fake connection that makes two things look related, A spurious variable makes it look like there’s a connection, but really, something else is behind it.

a third thing is causing both., It’s like a sneaky third wheel

methods of analysis

a variety of techniques used to analyze data and draw conclusions in political studies.

Ensuring confidentiality during interviews can protect respondents from all of the following except:

Emotional trauma from recall

___________ research tends to be more in-depth and offers a more nuanced understanding of its subject.

Qualitative

What is a Recall Question?

a question that asks you to remember and state a fact — something you've already learned.

What’s one advantage of qualitative research over quantitative research?

It provides deeper, richer, and more nuanced understanding of complex topics

What does subjectivity mean in qualitative research?

That the researcher’s background or perspective can influence how they interpret data

Ethnography

“Immersing yourself in a group or culture”

Long-term, hands-on observation

Researcher may live with or interact deeply with a group

Goal: understand norms, values, behaviors from the inside

Can involve deception (e.g., not revealing you’re a researcher)

In-depth Interviews

“Asking open-ended questions, one-on-one”

Researcher talks directly to participants

Usually semi-structured or unstructured

Great for personal stories, experiences, or insider perspectives

Example: Interviewing voters about why they distrust the government

Focus Groups

“Guided group discussions”

6–10 people talking together about a topic

Researcher acts as a moderator

Shows social dynamics, group opinions, and how people influence each other

Example: Gathering a group of young voters to discuss political ads.

Case Studies

“Deep dive into one case, country, person, or group”

Can be a single case or comparative

Great for studying unique events, policy decisions, or rare outcomes

Combines interviews, documents, observations

Example: Studying how one city implemented universal basic income

Content Analysis (Qualitative Style)

“Analyzing texts or media for patterns and meanings”

You code and interpret themes in speeches, news articles, social media, etc.

Not about numbers — it’s about how ideas are framed or discussed

Example: Looking at how political candidates use “freedom” in their speeches.

Complete Observer

Participants don’t know they’re being observed

Observer as Participant

Participants know they’re being observed

Participant as Observer

Participants know the researcher is involved

Complete Participant

Participants don’t know researcher is studying them

Standard Deviation (SD):

ells you how spread out individual data points are in a single sample.

Standard Error (SE)

Tells you how much the sample means vary from the true population mean.

Quasi-Experiment

looks a lot like a real experiment, but there’s one big difference:

❌ No random assignment to treatment and control groups.

Field Experiment

“Real world, still experimental.”

Done in real-life settings (like schools, towns, online).

Still uses random assignment.

Has both control and realism.

More generalizable than lab, but harder to control everything.

Example:

Randomly assigning neighborhoods to receive political campaign flyers to test voter turnout.

To calculate conditional probabilities, we generally need to look at

sub populations

sub population

a smaller group within a larger population that shares something in common.

Selection bias is when:

Individuals in the test group differ from those in the control group

The limitations of observational studies include the following

Omitted variable bias

Limits in data collection

Reverse causality

Selection bias

interpretivism

a way of doing research that focuses on:

Understanding how people see and make meaning of the world around them.

Instead of measuring facts or stats, interpretivists ask:

“What does this mean to the people involved?”

“How do people understand their political reality, identity, or experiences?”