Psychology

Hindsight Bias

when results come out for a certain event (ex: election), people believe they would have predicted the outcome after it has happened.

“I KNEW it would happen!”

Overconfidence

when one thinks they know more than they actually know

“I WILL do this perfectly!”

Perceiving Order in Random Events

the tendency to see patterns or meaning in random data, typically like superstitions or charms, to relieve anxiety or stress about an event.

“It MUST BE because of —-!”

Null Hypothesis

Statement: IV has no effect on DV

Alternative Hypothesis

Statement: IV has an effect on DV

Hypothesis Falsifiability

the hypothesis HAS to be able to be falsified scientifically. It cannot be too broad. (ex: dreams, religion, sigmund freud theories)

Confounding/Extraneous Variables

variables that can affect the DV other than the IV itself. They must be CONTROLLED!

Operational Definition

identification of EXACT operation on how to define/measure variables. with it, the study must be able to be replicated to be valid

Experiments must…

must measure cause and effect between IV and DV! if not, only shows an “association” and will be considered “a study” instead.

Representative Sample

represents the population at large that experiment is meant to be used for

Biased Sample

sample not representative of full diversity of the population being measured/studied

Random Sampling/Assignment

participants in study RANDOMLY ASSIGNED to groups, completely! experiments must be randomly assigned or else it will be considered a “quasi-experiment”, or not real. the only exception is if the experiment is performing a “within subjects design”.

Quasi-experiment

aka not a real experiment. it is when the experiment does not do random assignment when studying its population.

Convenience Sampling

when researchers choose participants in study because they are easier to access/reach out to or are close by. It is a low cost and faster way of collecting participants, but there is a high chance for bias affecting the results.

Single-Blind Study

when the participants are ignorant/blind/unaware about receiving treatment or placebo when in the experiment. very common method

Double-Blind Study

both research participants and research staff are ignorance/unaware which participants received treatment or placebo. for extra security. researchers will observe all unbiased and afterwards check which received treatment or not.

Placebo Effect

experimental results caused by expectations of participants alone. know of possibility of receiving treatment —> act differently bc assume they actually took the agent

Between Subjects Design

different subjects in different groups

Within Subjects Design

same subjects in different groups at different times. MUST be WELL-CONTROLLED.

Demand Characteristics

people finding out about true goal of study and will stop acting naturally and may either help prove or disprove the researchers’ hypothesis. describes the researcher’s tone of voice, mannerisms, eye contact, setting, wording, etc.

Social Desirability Bias

tendency to give researchers the more socially acceptable answer than the hard truth. typical in surveys, which is why it is important to check the wording of the prompts.

Internal Validity

how well-controlled the experiment is. High Internal Validity means all variables are well controlled, there’s random assignment, the experiment can perfectly establish cause and effect, and can rule out all other explanations. Low Internal Validity is the opposite and can’t be confident the results are due to the IV.

External Validity

if the experimental results can be applied to the rest of the population accurately. describes the artificiality of the method of experimentation.

Construct Validity

describes the validity of the instruments used for measuring the DV in the experiment. the operational definition must confirm that YES it measures the intention well and correctly. many use standard instruments and tests considered valid for measuring something. aka VARIABLE VALIDITY

Reliability

the consistency of the results. can describe the people taking the test or the people grading the test.

Test-Retest Reliability

if the person taking the test can get the same/similar results every single time over and over again

Inter-Rater Reliability

if different people grading the subjects can give the same grade to the same subject, strict rubric and little nuance.

Informed Consent

the consent given by the participant when they know exactly what they are doing and going through in this experiment and the consequences that could possibly happen if they do it.

Debriefing

required by all experimenters to debrief participant on what they are consuming, doing, and the overarching goal of the researchers. if they have to lie to participants, they have to explain why lying is necessary

Non-Experimental Research Methods

not actual experiments and are considered quasi-experiments. not all have random assignment, large sample size, etc.

Cross-Sectional Research Design

most common. groups that alr exist in nature and are measuring their differences.

can learn more about populations but not anything about cause and effect

Case-Studies

examine VERY SMALL number of subjects IN DEPTH.

reveal principles and are useful for studying rare conditions or aspects of the subjects (ex: phineas gage)

Naturalistic Observations

when researchers observe the subjects in their normal environment. researcher CANNOT interfere with environment.

describe behavior, not find causality when only doing this. can be used to make actual experiments though.

Small Science

sample observation in smaller population or a smaller space

Big Data

lots of information available for observation. aka social media

Surveys

not a research design method really. more of an INSTRUMENT. everything is self-reported. must need a representative sample for effectiveness since there is a high chance for a bias sample and social desirability bias. careful about wording

Correlational Studies

DOES NOT IMPLY CAUSATION! usually data already done in database and are just used to analyze quantitative relationship between variables. can use for future studies.

Correlation Coefficient

suggests the strength of the correlation/relationship between variables. represented by r. closer r is to 0 the weaker the relationship is. closer to +1 or -1 stronger correlation is.

+1 = positive correlation

-1 = negative correlation

0 = no correlation

Interpreting Correlational Scatterplots

make best fit line for data points.

Illusory Correlation

false perception of a relationship between two variables. caused by the cognitive bias of “perceiving order in random events”

Regression toward the Mean

a statistical phenomenon where an unusually high or low measurement tends to be followed by a measurement closer to the average simply due to chance and natural variation.

you get 100% —> naturally get lower score closer to your average next time

Meta-Analysis

when researching with no new data and instead using the data from many, many other studies already done on the subject.

a study of ALL studies alr done

Experiment v Quasi Experiment

lacking random assignment (e.g. using preexisting groups) —> less control, weak proof of cause and effect

Frequency Distribution/Polygon

show how often each score/value appears in data set. shown in graph

Measures of Central Tendency

shows the typical value in the data set: mean, median, mode

Measures of Variance

shows how much the scores differ from each other: range, standard deviation

Calculating Z-Scores

(Raw - Mean)/SD = Z. tells us the standard deviation that score is

Properties of the Normal Distrubution

68-95-99.7 = 1 SD away from the mean, 68% of data falls under that - etc.

symmetrical graph

mean, median, and mode in center highest peak

guideline for quantitative things like height, weight, intelligence, etc.

better to have a more narrow graph

Score Percentile Rank

rank shows how much of the population you did BETTER than

Skewed Distributions

there are positively and negatively-skewed distributions.

tail of the graph is going toward positive side of graph = pos

tail of the graph is going toward negative side of graph = neg

Bimodal Distributions

there are 2 modes —> 2 peaks in the graph

mean and median usually in between the peaks

mode is at both the peaks, one may be higher than the other (major v minor)

Inferential Analysis

check if there is a REAL difference or it’s just CHANCE between the results of the experiment. done after performing the experiment and received the data.

Alpha Value

threshold of certainty. always established BEFORE the experiment. and almost always set as 0.05 = 5% as the standard.

aka the difference is actual real ONLY if you are 95% CERTAIN or more

P Value

measure of uncertainty. checks the actual certainty that you have AFTER doing the experiment. want it to be VERY SMALL for the least amount of uncertainty.

Statistical Significance

P > alpha —> BAD, can’t reject the Null

P < alpha —> GOOD, can reject the Null (suggests there IS a statistical difference)

better if there are LARGER SAMPLE SIZES and getting SMALLER VARIATIONS so it’s easier to reach alpha threshold. if it’s smaller it is more likely to be impacted by outliers.

Effect Size

difference in means (usually). tells how meaningful result is for the real world. aka practical significance.

After deciding Statistical Significance..

CHECK FOR ERRORS (type 1 and type 2)

Type 1 Error

FALSE POSITIVE - tested positive but actually false. if alpha value too big

Type 2 Error

FALSE NEGATIVE - tested negative but actually positive and the test MISSED something. there IS an actual difference! if alpha value too small

T-Test

way to measure inferential statistics and significance. requires calculation.