Transcript Unit 2 Research Methods

Unit 2
Research
Methods
AP Psychology
The need for psychological science

Intuition and common sense – not in psychology!!!

Hindsight bias – widespread “I knew it all along”; should’ve seen it coming

Overconfidence – leads us to overestimate our intuition

Perceiving order in random events – we want to see correlations
The Scientific Attitude & Critical thinking
Curiosity, humility & skepticism prepares us to think smarter
Generalizability – to what degree the results of a study can be applied to different
types of populations
Critical thinking – smart thinking
The Scientific Method

Hypothesis – (in psychology) a statement of relationships between or among variables.
Psychologists generate statements after researching what other scientists have learned
and observing variables in the environment.

Then test it!

Theories – generally highly researched, rigorously tested, framework that organize multiple
studies under one umbrella of ideas; they are tested, not just questions
Operational Definitions & Replication

Needs to be measurable and manageable

Most important part of study; it defines what
researchers will be observing & manipulating

Experiments are more likely to be replicated
effectively with clear operational definitions used

The more a study is replicated, the more the
findings can be applied to the general public.
Generalizability of studies helps make them
applicable to real world.
Description – case study & naturalistic observation

Case study – one of oldest methods – Freud (Anna O.), Jung, Watson &
Rayner (Baby Albert)

Naturalistic observation & case studies – describing behavior, not explaining
it – anthropology, sociology, zoology
Description - survey

Random sampling usually generates representative sampling
*Choosing participants
random from larger
population
*Taking a randomly chosen at
sample & assigning
participants at random to
*Chosen for many types
either experimental
of research including
control group
surveys, interviews,
*only applies to
experiments
experiments
or
**Best basis for generalizing is from a representative sample of cases
***Representative samples represent the population
****percent of male & female participants should match between sample
and population as well as racial/ethnic identity
*****Always be critical of survey findings!!!*****
Correlation

Scatterplots – plots are never perfect (scattered)

How closely two things are related to one another

Positive or Direct correlations – as one increases or decreases, so
does the other

Negative or Inverse correlations – as one increases, the other
decreases

Correlation coefficient - number represents strength of correlation;
sign represents direction of correlation; -.70 is stronger than .65

Correlation does not mean Causation; tells you that they are related,
not why they are related

Correlation indicates possibility of cause-effect relations but does not
prove causation

Illusory correlations are often influenced by confirmation bias; we
may notice random coincidences but forget they are random & instead
see them as correlated

Caution: media reports of experiments: are experiments actually
experiments or correlations?
Experimentation

2 main ways to group participants in a study:

1. within subject design – comparing participants to themselves (pretest v. post-test)


Less efficient but more resistant to effects of individual differences
2. between subjects design – comparing one group of participants to
another (most common set up)


Only one group (experimental group) sees independent variable
More efficient b/c you can conduct research in one sitting
*Only type of study that can determine cause & effect is experimental
*Do not assign causal conclusions for experiments only
Experimentation

Random Assignment - often need 2 sample groups – can determine if its one factor
or another

Placebo Effect – health benefit from treatment by trusted source; can be used with
any treatment not just drugs; treatment can help even if not finished
Variables

Independent variable
Dependent variable

Manipulated variables
What is measured

Its what researchers change
The expected
make dependent variable
outcome
occur

Look for this 2nd
Look for this 1st

What will cause DV
this study?
What are they
Looking for?
to
in
Confounding variables – other factors that can possibly
affect outcome of experiment
-controlling for confounding variables is goal of
experimental method
-more controlled the environment, more likely you
can tell if IV caused DV
Variables

Reliability – consistency; retesting; e.g. SAT

Validity – measure what’s supposed to measure versus outside norms e.g.
SAT/PSSA/AP/school grades
Statistical Reasoning in Everyday Life

Types of Data:

Nominal Data – data that indentifies categories


Interval Data – data falls within a number line & has a zero point


e.g. gender, class level in school, yes/no survey answers
e.g. weight, height (no weight=0)
Ratio Data – data falls in number line where zero is just another
number on the line

e.g. 0 degrees/test scores
Measures of Central Tendency – mean, median, mode
Measures of Central Tendency

Single score that represents whole set of scores

Simplest yet most often confused type of basic statistics

Central tendency refers to how the data measure the center of a
set of data

Mean, median and mode all point to where middle of data should
be

If mean, median, mode are all same #, graph of data will look like
normal curve

If mean, median mode are different, graph will be skewed or off
center in some way

Mean most susceptible to extremes in data – gets pulled up or
down depending on extreme data points
Positive and Negative Skew

Positive Skew – when scores pull mean
towards higher end; the mean is more
“positive” or greater than the rest of the
scores

Negative Skew – the opposite; the mean
is pulled down towards lower end of
scores
Standard Deviation

Can tell how much scores deviate from
one another

Knowing standard deviation for a set of
data can reveal how similar scores are –
the higher the standard deviation, the
less similar the scores are

*AP test doesn’t usually test how to
calculate standard deviation, but how to
interpret it*

**AP test tip – knowing percentage of
scores that fall 1,2,3 standard deviations
from mean is important for AP exam**

For bell curve or normal curve – 68% is
one standard deviation, 95% 2 standard
deviations, 99.8% 3 standard deviations
Inferential Statistics

Representative Samples are better
than biased samples

Less-variable observations are more
reliable than those that are more
variable

More cases are better than fewer

**Generalizations/anecdotes are
unreliable**
Statistical Significance

How 2 groups’ means are different

If you graphed data from 2 groups & the the graphs did not overlap or
only overlapped a little, the difference would be significant; if graphs
overlapped a lot, the difference would not be significance

Significance allows you to say how likely the difference in means is due to
chance

The usual goal is to get a significance level of .05 which says results are
only 5% due to chance (or 95% due to your independent variable)

Effect size (measure of strength of relationship between variables) – new
way for researchers to report results; tells more about quality of data then
significance along
Psychology Applied

General principles that help explain many behaviors most important

Psych studies have been criticized in past for using samples that are
not representative of general public e.g. not including both genders
or multiple cultures

Famous example: Kohlberg moral development
Ethics in Research

APA ethics code requires that institutions that allow animal
experimentation to take place establish an Animal Care and
Use Committee to review whether studies done on animals
follow ethical standards

Milgram’s obedience and Zimbardo’s prison guard studies
forced researchers to consider long term psychological effects
on participants who may commit acts the didn’t realize they
were capable of

Confidentiality – keep participant info private; info may not be
released and data must not be presented so individuals can
be identified through any kind of process of elimination

Anonymity – participating in study without providing identifying
info; often helps participants respond more truthfully
Drs. Kenneth & Mamie Clark

1st psychological research cited in Supreme Court decision – black children
prefer white dolls over black dolls

Highlighted the importance of examining controversial and societal issues
with scientific methodology