Exam 2 Research Psychology

Internal validity

determines if changes in the independent variable caused changes in the dependent variable

history

external events influence results; control groups help prevent this

maturation

natural participant changes; use control groups

testing

practice effects from pretests; use alternate forms

instrumentation

tool or rater changes; recalibrate instruments

statistical regression

extreme scores tend toward the mean; avoid selecting extremes

selection bias

groups differ initially; use random assignment

mortality

dropouts bias results; track and analyze attrition

diffusion of treatment

groups share info; separate groups

randomization

equalizes unknown variables

matching

pairs subjects on key characteristics

within-subject design

each person serves as their own control

control of subject effects

deception or blind designs

control of experimenter effects

double-blind methods

demand characteristics 

keep hypotheses hidden 

double-blind

prevents bias from both sides

external validity

extent results generalize to other settings or populations

tradeoff

higher internal validity can reduce external validity

independent variable

clearly defined and manipulated?

dependent variable

reliable, valid, and objective?

comparison group 

control included?

random assignment

properly used?

experimenter/subject bias

controlled (e.g., double-blind)?

sample

representative and sufficient size?

descriptive statistics

summarize and organize the data (mean, median, SD, range)

mean

uses all data, sensitive to outliers

median

not affected by outliers; best for skewed data

mode

simplest; used for nominal data

range

quick estimate; sensitive to extremes

variance

uses all data; less intuitive

standard deviation

common and interpretable

Z-score

shows how far a score is from the mean in SD units 

inferential statistics

make population conclusions from samples

• we can’t prove hypotheses— only test likelihood(probability) 

central limit theorem

1. sampling distribution approaches normal as sample size increases

2. mean of sampling distribution = population mean 

• larger samples reduce sampling error

seven step hypothesis testing procedure

1. state the research problem 

2. state the null and alternative hypothesis

3. choose significance level 

4. select the proper test

5. compute the test statistic 

6. make the decision to fail or reject to fail the hypothesis

7. interpret in context

• decision errors- type 1(false positive), type 2(false negative)

z test

one group; known

t-test

one or two groups; unknown

ANOVA

3+ groups

correlation

relationship between variables

chi-square

frequency counts

writing an journal article summary 

1. citation in APA format

2. purpose and hypotheses

3. method: participants, design, and procedure

4. results: key findings

5. discussion: interpretation and implications

6. critique: strength, weaknesses, future research