PSC 41: Unit 2

Frequency data

a count of how often smth happens

MEASURED, not manipulated

strong frequency claims need strong construct & external validity

Categorical

levels are categories

Quantitative

levels are meaningful numbers

Ordinal

numbers are in a ranked order; distance between numbers may not be equal

Interval

numbers have equal distance, but there’s no true zero

Ratio

numbers have equal distances & there is a true zero

Open-ended questions

questions that allow respondents to respond in their own words

pros: allows responder to say what they want to say

cons: hard to code

Forced-choice questions

asks respondents to choose from a list of answers

pros: easy to code

cons: options might not capture what a person thinks

Likert scale

ppl are given a statement & are asked to rate using a scale to indicate degree of agreement

a type of forced-choice question

Mistakes to avoid when creating questions

1. asking questions that people may not know the answer to or may be unlikely to provide an accurate answer to

2. leading questions/loaded questions

3. double-barreled questions/compound questions

4. use of double negatives

5. confusing or ambiguous wording

Leading questions

type of qn in a survey/poll that's problematic bc its wording encourages 1 response more than others → weakens construct validity

Double-barreled questions

type of qn in a survey/poll that's problematic bc it asks 2 questions in one → weakens construct validity

Other mistakes to avoid when creating questions with response sets

1. nay-saying & yea-saying

2. social desirability

3. faking bad

Nay-saying & yea-saying

people default to 1 answer —> can solve w/ reverse wording (aka reverse coding)

Social desirability

qns w ethically/morally “correct” answers or questions with strong experimenter preference

Faking bad

questions that the participants want to score high/low on even tho they’re seen as negative

Affective Neuroscience Personality Scales

assesses human behavioral traits related to 6 primary neural affective systems that have been identified in animal models of affect

• play, seek, care, fear, anger, sadness

• ANPS says that ppl’s emotions affect personality/behavior/mental health

Barret et. al

claimed the ANPS scale was:

• too long

• had constructs that weren’t indep

• had items w/ poor content validity

• had items targeting same traits w/ poor inter-item reliability

Direct human observation (problems w/ it)

• ppl studied will be distracted by those observing them

• they might change how they act

Hidden human observation

• can blend in with large crowds

  • ex. games, amusement parks, picnics, crowded streets

• would help to have many observers do multiple rounds of observation

  • many observers = higher inter-rater reliability

  • many rounds = higher valid results thru averages

Dispersion

extent to which scores in the distance deviate from the mean

Standard deviation

way to express how variable people are from the sample you measure from

• reps a fixed proportion of data

• helps visualize individuals in relation to the rest of the ppltn

• provides a way to talk about relative distance of a single score from the mean that is unit free

Range

difference between largest & smallest data value

Electronically Activated Recorder

would periodically turn on & record sounds without warning

• was used in a study to see if women = more talkative than men

• results —> even tho mean value differed, SD showed that distances were actually very similar & overlapped

Observer bias

• when observers expectations influence their interpretation of the participants' behaviors or the outcome of the study

• if interrater reliability = good, must make sure raters don’t share the same biases

Observer effects

a change in behavior of study participants in the direction of observer expectations

solutions:

• must blend in or be hidden

• wait until you’re no longer novel

• observer indirect measures (consequences of a behavior)

• use non-human observer

• be aware of how data is collected

Masked design

study design in which the observers are unaware of the experimental conditions to which participants have been assigned

Reactivity

a change in behavior of study participants bc they're aware they're being watched

External validity

extent to which the results of a study can be generalized beyond the current study to other ppl/situations/time periods

questions to ask to assess external validity:

1. is the research authentic?

2. do the results generalize to other situations?

3. does it generalize to other individuals?

Population

the entire set of ppl/products that are of interest to a researcher

Sample

a subset of a population that is selected for inclusion on the experiment

Non-representative sample

selection of who is in the study is biased toward some character

aka non-probability sample

Representation sample

everyone in the population has an equal chance of being in the study

aka probability sample

Inference

typically, conclusions are drawn back to the population

this depends on how the sample is selected

Non-representative samples vs. representative samples

Non-representative:

• doesn’t use random samples

• diff members of population have diff likelihood of being chosen to sample

• sample may not be similar to population

• may have sample bias

• may have weak external validity

Representative samples

• uses random sampling

• every member of population has equal likelihood of being chosen to sample

• sample will be similar to population

• no sample bias

• strong external validity

Sampling methods

probability sample methods:

• systematic

• cluster

• stratified

  • oversampling

• simple random

non-probability sample methods:

• convenience

• self selection

• judgmental

• quota

• purposive

• snowball

Systematic sampling

every nth person is selected to sample

Cluster

where ppl are divided into clusters by char, then ALL individuals within randomly selected clusters are sampled

Stratified

where ppl are divided into strata, then a certain number of ppl are randomly selected from all strata

Strata

groups usually defined by a continuous variable

Oversampling

a variation of stratified sampling where the researcher intentionally overrepresents one or more groups

Simple random sampling

sample is chosen completely at random from the population of interest

Convenience

sampling whoever is most easily accessible

Self-selection

when only those who volunteer are sampled

Judgmental

when researchers subjectively pick out whoever they want to sample

Quota

when researchers nonrandomly select individuals from categories until a target number (quota) is reached

similar to stratified

Purposive

when only certain kinds of ppl are included in a sample

Snowball

where participants are asked to get other ppl they know to sample

variation of purposive sampling

Random sampling vs. random assignment

random sampling

• every member of ppltn has equal chance of being sampled

• happens BEFORE an experiment

• important for internal validity

random assignment

• is done to avoid biases in assigning individuals to different experimental groups

• every member of sample has equal likelihood of being assigned to the conditions in the experiment

• happens in experiments with 1+ groups

• important for internal validity

WEIRD samples

Western, educated, industrialized, rich, democratic

Census

a set of observations containing all members of the population of interest