PSYCH 109: Research

Merton’s Norms for Science

Communality, universalism, organised scepticism

Communality

Belonging to all

Universalism

Validity independent to researchers

Organised scepticism

Available to criticism

Dependent variable

What’s measured

Independent variable

Manipulated to see how the DV changes

Reliability

Replication of results when repeated

Validity

DV is the thing being measured

Population

Every data point the research covers, what the DV measures

Sample

Representative of the population, generalised conclusions for population

Sampling error

results of repeated samples from the same population differ

Sampling bias

Population is systematically misrepresented, random sampling solves this

Observational design

Correlation between 2 DV, correlation doesn’t equal causation

Experimental design

IV’s effect on DV, can imply causation

Confounding variable

Anything other than the IV that could affect the DV

Standardisation

Eliminates confounds by maintaining their conditions

Randomising

Helps minimise internal/subject-based confounds

Within-subjects

All subjects are expoed to all experimental conditions, minimises internal confounds

Between-groups

Every subject is exposed to one experimental condition, minimises environmental confounds

Matched pairs

Creates pairings with similar abilities randomly divided across conditions

Single-blind design

Eliminates subject expectation when subjects don’t know what group they’re in

Double-blind design

Eliminates experimenter and subject expectation, do not know what groups subjects are in

Significance level

0.05

Sagan’s balance

Extraordinary claims require extraordinary evidence

H0

Null hypothesis, no real effect, results due to chance or sampling error

H1

Experimental/alternative hypothesis, real effect

Formal logic of inferential statistics

Assume H0 is true, if the probability of results drops below the significance level H0 can be rejected

H1: two-tailed

No specification of the effect’s direction

H1: one-tailed

Predicted direction of the effect

Type I error

Rejecting null hypothesis when perceived effect was due to chance

Type II error

Accepting the null hypothesis when it’s false

Cautious compromise

Reducing the significance level reduces one effect but raises the other

Mean

Location/average

Standard Deviation

Variability/spread

t value

Compares difference between means with SD

Finding significant difference

Finding t’s probability of H0 being true

Degrees of freedom

Number of scores that can be changed while the means stay the same

Calculating df

df = N1+N2-2