Research Methods in Psychology Exam 1

three types of claims

1. frequency claims

2. association claims

3. causal claims

four types of validity

1. construct validity

2. statistical validity

3. external validity

4. internal validity

construct validity

how well something is measured

statistical validity

what is the margin of error? what can the statistics of a claim tell us about its accuracy

internal validity

the ability to rule out alternative explanations for causal relationships between two variables

external validity

a measure of how well the study generalizes to areas outside the study itself

goals of research in psychology

describing behavior

understanding behavior

predicting behavior

solving applied problems

research designs in psychology

survey and observational

correlational

experimental

quasi-experimental

frequency claims goals and design

goals: describing

design: survey, observational

association claims goals and design

goals: understanding predicting, solving problems

design: correlational, quasi-exoerimental

causal claims goals and design

goals: understanding, predicting

design: experimental

scientific method

particular way of arriving at knowledge

ways of arriving at knowledge

authority

logic

experience

knowledge from authority

based on what authority figures tell us, is not always accurate

logical reasoning

1. assumptions (ex: primates are capable of using language, Bozo is a primate)

2. conclusions: Bozo is capable of using language

can also be false

knowledge from experience

-what we experience may have other confounds

-what we experience can be biased, this is called "intuition"

social cognition biases

1. belief perseverance

2. availability heuristic

3. conformation bias

belief perseverance

ignoring contrary evidence

availability heuristic

occurs if one's experience is unrepresentative, the what comes easily to the mind becomes explanation

confirmation bias

selectively seeking out evidence that supports one's belief

the scientific method relies upon

data: role of observation, i.e. experience

logic: to draw logic about what the data means

characteristics of scientific method

1. systematic empiricism

2. public verifiability

3. hypothesis testing

4/ examines only empirical questions

5. conclusions are always tentative

systematic empiricism

-data driven

-empirical: relying on observation to draw conclusion

-systematic: systematic collection of data, precise definitions (so no alternative interpretations), reliable valuable measures

public verifability

-findings must be observable and recognizable by others

-procedures used must be systematically reported (APA style)

testability

-examining only empirical questions ( those that can be tested in a systematically empirical and publicly verifiable manner

-an example of a non-empirical question would be "do angels exist" not publicly verifiable, not replicable, not systematic examination

why we need tentative conclusions

we can never be sure a result will hold up

variables

-what is being studied, unit

-in psychology, an aspect of human behavior or something that affects human behavior

-varies among individuals or situations

types of definitions

1. conceptual

2. operational

conceptual definition

dictionary definition

operational definition

defined in terms of a procedure or set of operations, or how the concept is measured

how to frame a research question

issue of interest -> empirical question -> hypothesis

empirical question

must be answered by observable data and terms must be clearly defined.

hypothesis

conjecture regarding the relationship between two variables that can be tested empirically

theory

a set of propositions specifying interrelationships between constructs and or variables. Should be tentative and falsifiable

construct

hypothetical factor or variable that cannot be directly observed but is inferred from certain behaviors, an abstract factor/variable

alternative definition of theory

-summarizes empirical knowledge

-organizes knowledge in the form of precise statements of relationships among constructs or variables

-provides tentative explanations

-basis for predictions about behavior

deduction

reasoning from the general to the particular

-theory->hypothesis->experiment

induction

reasoning from the particular to the general (think mexican food example) *ideas are derived from a theory through induction

example of a logical inference

-if p then q is true: if p is not true, q is not true

what does an experiment show?

-since theories cannot be logically proven they are always tentative

-can support or confirm

-same for disproof, single study is not disproof

frequency claims

describe rate, level, frequency, number, or amounf of something.

-involves only one variable

-that variable is a measured variable not a manipulated one

association claims

claim that one level of a variable is likely to be associated with a particular level of another variable

-involves more than one variable

-these variables are measured not manipulator

association claims in research designs

-many research designs lead to association claims, should to just be linked to correlation/regression designs

causal claims

-claim that variation in one variable causes variation in another variable

-involves more than one variable

-must satisfy certain criteria of causality

criteria for establishing causality

1. causal and outcome variable must covary (as one changes so does the other)

2. variation in causal variable must precede variation in outcome variable (temporal precedes criterion)

3. other causes of variation in the outcome variable must be ruled out (internal validity criterion, have to rule out third variable)

construct validity

does the operational definition actually measure the construct of interest

statistical validity

extent to which conclusions/claims are based on sound statistical analysis

external validity

degree to which claims/results of study can be generalized to other populations

internal validity

degree to which variation in the outcomes variable was really due to variation in the causal variable

correlations

-can be calculated between any two variables

-only provides measurement of linear relationship

-greatly affected by a range of values

-unaffected by linear transformations

-have no units

-bidirectional

r

-correlation coefficient

-varies between 1 and -1

r^2

-the coefficient of determination

-varies between 1 and 0, the smaller the value the less correlation between the two variables

face validity

extent to which a measure appears to measure the construct it is supposed to measure

content validity

extent to which a measure includes all relevant aspect of the construct

construct validity

the most important kind of measurement validity

-convergent validity (certain things SHOULD always correlate with)

-discriminant validity (certain things should not correlate with)

criterion validity

whether the measure is related to a concrete outcome that it should be related to if its valid

ordinal scale

numbers indicate rank ordering

numeral scale

numbers are just a label

interval scale

equal differences between numbers reflect equal differences in the thing measured

ratio scale

properties of interval scale, and a true zero

descriptive statistics

-measures of central tendency

-measures of dispersion

inferential statistics

-confidence intervals (estimation

-hypothesis testing

convergent validity

a type of measurement validity that represents the extent to which a measure is associated with other measures of a theoretically similar construct

discriminant validity

-also divergent validity and empirically support type of measurement validity that represents the extent to which a measure does NOT associate strongly with measures of other theoretically different constructs

mode

most commonly occurring observation

-unaffected by extreme values

measure of central tendancy

median

score at or below which 50% of observations fall

-unaffected by extreme values

(N+1)/2

measure of central tendancy

mean

the average

-affected by extreme values

measure of central tendancy

all psychological research questions are about what kind of variability

behavioral

how to measure variability

-range

-interquartile range

-variance

-standard deviation

range

-distance between the highest and lowest values

-affected by extreme values

inter-quartile range

-first quartile (Q1) score at or below which 25% of observations fall

-second quartile (Q2) (median) score at or below which 50% of observations fall

-third quartile (Q3) score at or below which 75% of observations fall

SO Q3-Q1 is the range of the middle 50% of observations

-less affected by extreme values

steps in calculating variance

1. calculate mean

2. calculate deviations from mean

3. square deviations

4. add squared deviations

5. adjust for number of subjects in sample by dividing by # of ss

SD

square root of variance (so variance measured in squared units)

-indicated representativeness of mean

systematic variance

the part of the total variability in subjects' behavior that is related in a predictable way to the variables being studied

error variance

the part of total variability due to individual differences or random unpredictable factors (not related to errors made by subjects)

variance formulas

r= variation shared by x,y/ total variation in x,y

OR

r=covariance x,y/SDx x SDy

properties of a normal distribution

within 1 SD 68% of data

within 2 SDs 95% of data

how many sd's from the mean?

take the difference and divide by the SD

z table

-follows normal distribution mean=0 and SD=1

key idea in calculating variability

variance of the observation from the distribution mean/variance of distribution

the t statistic

-use sample SD to calculate SD's from mean

-as sample size increases, t distribution becomes closer to normal distribution

-degrees of freedom = N-1

-use t instead of z when we don't know population SD

types of t-tests

-tests for differences between sample means

--when the samples are independent (different subject in two groups)

--when samples are not independent (same subject in both groups)

how to calculate SD

find distance from mean

square value

take sum of squares

divide by number of entries

take square root