Effect Size

small- .01 medium- .09 large- .25

Factorial Design Hypothesis

1 hypothesis per factor + interaction & 1 f-ratio per hypothesis H0: no main effect (no difference) H1: main effect (is a difference)

2 x 2 Factorial Design

2 factors, 2 levels, 1 DV Ex. Factor A: weapon present -> Level 1: present Level 2:absent Factor B: lighting quality -> level 1: good level 2: poor DV: eye witness

Small N design

usually repeated measures, data reported independently, compare treatment & control, usually in applied settings -when to use: study a rare phenomenon, understand a single individual

Quais-Experiment: Use

research questions with policies/legal implications, voluntary medical procedures, and trailing large-scale interventions

Quasi-Experiemnt

Some predictors cannot be manipulated (IV: IV that can't be randomly assigned) - no true control group

Experiment 1 IV

1 IV, 2 conditions: between-subjects -> independent t-test 1 IV, 3+ conditions: between-subjects -> one-way ANOVA

placebo effect

solution- chose comparison results

demand characteristics

The features of settings influence results

observer bias

Researchers want to influence the results

Confounds

Systemically vary along with IV (internal validity problems)

Null results

finding no effect/relationship -why: true null effect, obscuring variables

Worst design

one group, pre-post design -pre-test -experiment (1 level) -post test (doesnt include comparison group)

double-blind

Neither the experiments/particpants know the conditions

masked design

Raters don't know the condition

Selective reporting

Reporting only successful studies/ significant analyses -gives an inaccurate picture of current knowledge

Inaccurate records

Keeping poorly detailed/inaccurate records of the procedure -makes replication hard/more risk to participants

Questionable Research Practices

decisions made during the research process that compromise research integrity& validity of the conclusion (usually unethical)

Within Group

different types of "noise" -> irrelevant external factors, measurement error, individual differences -solution: w/in subject manipulation & more participants, more precise/ take more measurements]

Between Group

between-group, weak manipulation/intensive DV measure/ floor & ceiling effects -solution: manipulation check

Collecting Additional Data

analyzing data/deciding to collect more based on results, often done when results are approaching significance - amplifies sampling error

HARKing

hypothesis results after results are known, presenting results like they were hypothesized, or excluding the initial hypothesis -reduces replicability

P-hacking

manipulating the analysis conducted to find one statistically significant result -inflates the false positive rate

Open Science Practices

preregistration, making data/analysis code available, preprints, & expanded supplementary materials

Systemic Misconduct

Predatory journals, research paper mills, and networks of individuals

Scientific Misconduct

violating codes of research misconduct, distortion/fabrication of data, or inaccurate credit on research

Plagerism

appropriating someone's info without giving them credit

Citation

not crediting prior research

Plagiarism/Fabrication

reusing unrelated figures

Self-plagiarism

publishing results multiple times

Falsafication

manipulating the process of changing data/ results

Fabrication

making up data/results

mauration

spontaneous change in behavior over time

history effect

A bunch of participants change from an external factor

regression to the mean

Extreme outliers tend to become more average over time

attrition

Participants drop out over time - solution: remove pretest scores & analyze for systemic difference

Testing effect

change due to pretest -solution: comparison group & remove pretest

Instrumentation

measure change over time -solution: recalibrate measure, retain coders, remove pretest, & counterbalance