Research Methods + Statistics

Intuition

A hunch or gut feeling

Authority

Information accepted as true because of unquestioning belief or trust in someone who is considered an expert

Rationalism

Information is accepted as true based on the rules of logic (requires two correct premises)

Empericism

Information is accepted as true because it’s based on direct observation of events

The scientific method

Must be based on systematic observation, must be falsifiable, and must be open to feedback and correction

Covariation

Causal criteria — if one variable causes a second, then scores of both variables should systematically correlate

Temporal precedence

Causal criteria — cause comes before the effect, manipulations come before measurement

Internal validity

Causal criteria — Alternative explanations have been eliminated — can be established through random assignment

Construct validity

Does your study study what it claims to study

Statistical validity

Are the statistical procedures appropriate

External validity

Are the study’s conclusions generalizable across populationsI

Internal validity

Are the causal conclusions appropriate

Nominal variable

Purely classification based, no relative standing (eg. hair color)

Ordinal variable

Based on classification with relative standing, but no equal interval (eg. grade level)

Interval variable

Has relative standing and equal intervals, but no true zero (eg. IQ score) — note: Lickert scales will be treated as interval scales in this class

Ratio variable

Has equal intervals and a true zero (eg. # of shoes)

Face validity

Does it look like what you want to measure?

Content validity

Is it comprehensive? Does it miss anything relevant to the construct?C

Criterion validity

Is your measure correlated with current or future relevant behavioral outcomes?

Convergent validity

Is your self-report measure correlated withs elf-report measures of similar constructs?D

Discriminant validity

Is your self-report measure not associated with self-report measures of dissimilar constructs?

Population

What we want to understand but cannot fully observe, parameters must be estimated (size=N, mean=μ, SD=σ)

Sample

A subset of the population we can observe, measured in calculated statistics (size=n, mean=M, SD=s)

Central tendency

What values are “typical” in our data? (mean, median, mode)

Variability

How spread out is our data? (range, IQR, variance, SD)

Skew

Is our data symmetric? — positive (right) skew: more extreme large values — negative (left) skew: more extreme small values

Kurtosis

Are the data stretched out towards the tails?

Normal distribution

Symmetric, bell-shaped distribution which follows the 68-95-99.7 rule

Simple random sampling

(Probability sampling) start from all members of the population, each individual has equal chance of being selected

Systematic random sampling

(Probability sampling) don’t need all members, choose every nth respondent with an interval

Stratified random sampling

(Probability sampling) dividing the population into strata and taking a random sample from each stratum

Cluster sampling

Several clusters, randomly choosing one cluster and asking every member

Convenience sampling

(Nonprobability sampling) just taking an easily accessible population, less likely to represent the population

Quota sampling

(Nonprobability sampling) decide on a quota for a subpopulation and recruit until each quota is met

Snowball sampling

(Nonprobability sampling) research participants find other potential participants, useful when studying a hard-to-reach population

Sample statistics

Any numerical values calculated directly from the sample data — summarizes to gain insights

Population parameters

Characteristics of the entire population (not directly measured, sample statistics are used to estimate it)

Central limit theorem

As a sample size grows sampling distribution will approach normal

Sampling distribution

A distribution of a statistic taken from many theoretical samples

Confidence interval

A range of values likely to contain the population parameter

Confidence level

The long-run probability that a series of confidence intervals will contain the true value of a population parameter

Confidence interval interpretation

___% of intervals constructed this way capture the true population parameter, I am ___% confident that this interval captures the true population parameter