Functional Explanation

functions = benefits

identifies the long-term benefits/advantages that accrue to whoever causes the event

Scientific Explanation

identifies causes of events (ie, behaviors) and their functions.

based on evidence from empirical research using scientific method

Falsifiable Prediction

If dog sees food then she will eat it.

describes specific pattern in research results that should occur if explanation is correct

• Marxism-bad but eventually will prevail
• Religious Doctrines
• Casual fatalisitic reasoning-wasnt meant to be
• theories (need 'special training')
• Telic- is a 'grand plan'

Scientific Method

Empirical Research

collection of new observations under conditions

allows you to answer questions about the topic of study and confirm or contradict predictions of a proposed scientific explanation

Correlational Study 

descriptive research then examine whether (2+) the objects/events being studied covary

Experiment

Empirical Research in which you manipulate features to be a cause and measure other features you believe to be an effect

to decide cause-effect relationship

Bibliographic Research

review past published writing to research topic

Research Types

• empirical
• bibliographic
• methodological

Methodological Research

Hypothesis Components

1. expected pattern of observations
2. required conditions for observing pattern
3. reason to expect pattern

pattern of observations for one variable

predicts pattern of observations for one variable

Univariate Hypothesis

Relational Hypothesis

Causal Hypotheis

relational plus claims one causes variable other(s)

Axiom

Assumption made by a theory

PARSIMONY

Ratio of what theory explains over how complicated it is.

Confounded Manipulation

(IV) A varies mroe than intended so you dont know what aspect of A is causing B.

Tom or Taiko- Was it gender or actually his name that didnt get the job?

Causal Analysis

[Steps 1-3]

Demonstrate a causal relation exists

1. Find covariation strength
2. direction of causation
3. identify 3rd variables

confounded manipulations could kill relation

3rd Variables vs. Confounded Manipulations

3rd Variable- in non-experiment research

cm- in experiments and quasi-experiments

1. Measure (C): do they covary with A and B?
2. Suppress: say it doesnt cause A and B.
3. Experimentally Manipulate: if manipulation= causation then you have sufficient evidence
4. Develop Strong Args: could be but doesnt make sense

A->B and B-> A

(Doesnt that violate nagel though?!?!?)

TIME LAGS!

Then you have reciprocal causation: A->B and (separately) B->A

ok as long as B doesnt determine CURRENT A

Discriminant Validity

Give evidence that variables arent related when they shouldnt be

(separated clouds for A and B)

Test-Retest Reliability

• Measure A twice
• Correlate 2 times

time:

too little- cause interference

lower r_aa: intentional duplication of r_aa

too much- raa can drop b/c meaningful change in attributes of A and B

Just Enough

Spurious Causality

False belief that one of 2 covarying variables causes the other

(Actually a C that causes both A and B at the same time)

Correlation thus Causation?

Nagel's Definition of Causation

1. relation is invariable and uniform
2. for events contiguous in space
3. cause precedes effect in time and are temporally contiguous
4. relation is asymmetrical

a->b and not b->a

Error

ran vs. Sys

ran- unpatterned

 over many repetitions eran= 0

sys- patterned

These errors do not cancel each other out

ex- polls how ppl vote: sys=bradley effect, ran= epichanies/ppl chaning minds suddenly

Predictive Validity

Causal Analysis Step 2

How do we tell direction?

A->B

or

B->A

Timing: Which comes first?

Manipulation: create variation in levels of A then measure B. If A and B covary then A causes B.

Validity

what one claims is true

what actually is true

(inferential v= absence of doubt about findings)

= predictive but you dont have to wait

test now and see if it matches things you already knew

Face Validity

Not scientific but political

widely agreed valid

Police example

Causal Analysis

[Steps 4-6]

4. identify variables that mediate the relation

5. variables that moderate the relation

6. A and B cause each other?

split-half reliability- on questionnaire odd items should equal even items.

coefficient alpha- formula to estimate test-retest reliability based on correlation tween individual test items in single measure

7 Approaches to Assessing Construct Validity

FCCRPD

reliability- if not reliable then construct isnt valid

content validity

predictive validity

convergent validity

discriminant validity

discriminant validity

face validity

concurrant validity

Boundary Conditions

Construct (measmnt) Validity

RELIABILITY

• Absence of eran (not esys)
• same score if same variable is remeasured under same conditions

to degree different values are produced, measurement is unreliable (repeatability)

e= observed score - true score

Observed = True +/- esys and eran

Ceiling- 'high end of measurement' doesnt reach highest Truth

Floor- 'low end' doesnt reach lowest Truth

3 Components of Reliability

Stability

Internal Consistency

Inter-Rater Agreement

Content Validity

Measurement is valid if incorporates full content (all aspects) of corresponding construct

Almost full circle is shaded

Mediating Variable

A->B

BECAUSE

A->E->B!

Internal Consistency

low- no way to predict from one answer the answer to next question

if

variable scores= add of multiple components then should score same on all components

6 Fundamentals of Covariation

1. is to a degree
2. only to a set of cases
3. can be pos/neg (or not for categorical)
4. A and B can vary without covarying with each other
5. if A or B doesnt vary then A and B cant covary
6. different forms of covariation

Convergent Validity

2 measurements = same scores for same variable in same cases

refine construct to show that it relates well to operational variable

Causal Analysis Step 1

must covary to cause each other

trivial causal relations are absolute (present/absent) but causal processes that psych study are partial and probalistic (wk/strong/moderate/not perfect)

A->B partial/incomplete= others may also cause B

Internal Validity

absence of reasonable doubt that research results accurately describe aspect of natural world.

Internal validity exists to the extent that rival explanations for research results do not exist, or are not credible

problems in empirical study's design that allow for rival explanations

ie. artifacts, 3rd variables, confounded manipulations, poor construct validity, expectancy effects, order effects

History Artifact

changes over time on a measured variable that result from events outside contect of study

ie, moods decline in more sunlight with changing seasons

History by treatment artifact

Maturation Artifact

Instrumention Artifact

Attrition Artifact

Attrition by treatment artifact

Assignment Bias

VS

Selection Bias

Assignment= assignement to IXB- nonrandom or non-counterbalanced

Selection- non random or non structured or non stratified sampling

1- construct validity

2- internal

3- statistical significance

4- external

Statistical Regression

can occur if researcher premeasures a key variable before beginning study; only high-scoring/low-scoring are selected for further study

or

as above, but all pretested cases are included in study but each case is assigned to hi or low grp based on pretest

ie, when we picked numbers

Regression ARtifcat

Main Effect

Main Effects

(read section!)pg51

External Validity

Generalizability

how?

• show construct validity
• establish internal validity
• generalize to population
• generalize beyond population

Generalizability outside population ideas

other cultures

genders

ages

measures

stimuli

procedures

settings

Point-Biserial

Eta (and eta²)

phi

Coefficient of Contigency

statistic that measures covariation between two categorical variables that have any number of categories

Correlation Coefficient

Pearson r

most common correlation coefficient for two scalar variables

version of pearson (for two scalars) calculated on ranks

Look at...!!!!

5:

6:

7: Distributions of Scores

8: Descriptive Statistics

26:

27: