PSYC 2100WQ Terms, Lecture Exam #1

Empirical science

Empirical science

Science based in numerically represented observations of our five senses

construct

provides a definition of what is to be observed before data collection

Operationalization

how a construct is to be observed and numerically recorded

statistic

numeric summary of numerically recorded observations

Variable

event/object/attribute where numeric observations show variation

covariation

the variation of one thing is shown to be related to the variation of another thing

Simple covariation

Absence of specified order in covariation (use non-experimental designs)

Ordered covariation

Demonstrates an independent/dependent variable relationship (aka causality, use experimental design)

Data

numerically recorded observation

Hypothesis

A formal and public statement/declaration about what a researcher things is true

Knowledge claim

based on observations consistent with hypothesis

causality

for every cause, there is an effect, and for every effect, there is a cause

Population

unspecified N# of individuals for which we have numeric observations, numeric summaries only exist in theory

sample

participants drawn from a population

Descriptive statistic

Used to describe a sample

Inferential statistics

Use numeric observations about the sample to make inferences about the population from which the sample was drawn

skewness

in reference to a frequency distribution, when the line departs from what is bell shaped

Standard deviation

indicates average variation of scores from the mean

case study

non-experimental design where n=1, used when permission is difficult to obtain

trial

indicates in advance where and when data are going to be collected

Naturalistic observation

record observations in real-word (natural) settings (can include non-humans)

Experimental designs

designs in which the independent variable is manipulated in order to identify causal relationships

control group

group of participants not exposed to IVxperime

experimental (treatment) group

group of participants exposed to IV

4-group experimental design

use of two control groups and two experimental groups

skepticism

suggestion of doubt about a knowledge claim

Threat to validity

Is the design of the study (the basis of the claim) sufficiently valid for providing an empirical basis for the knowledge claim?

To address internal validity threats…

…build better study design that addresses the something else

Internal validity questions

Is the variation in the DV due to the IV, or due to something else (must be identified)?

External validity question

Does the IV-DV relationship as observed in the study generalize beyond the study (to other settings, sample, points in time)?

To address external validity threats…

…build a better study design that addresses the other areas

Measurement

measure/scale/test used to assign numbers to observations (how we operationalize)

Reliability of a measure

scores yielded by the scale (measure/test) show consistency over time, scaling formats, or with respect to the responses to items that comprise the scale

Validity of a measure

scores yielded by the scale (measure/test) provide a correct representation of the low-high variation of the variable of interest (do they do what we say they do)

1, 2, 3 of person’s correlation

Departure from zero indicates (1) - the presence of covariation (2) direction of covariation and (3) how much covariation

Test-retest design

data collected at two points in time to examine the consistency of scores (covariation at two time points indicates consistency)

Equivalent-forms design

data collected from multiple scaling formats to examine the consistency of scores (covariation across two scaling formats indicates consistency)

consistency can be indicated by

covariation

Internal consistency design

Answers to each item should correlate as highly as possible with every other item (as seen using the average of r across all items)

Predictive Criterion-related design (validity of measure)

Couples two hypotheses. One states simple covariation between the variable of interest (predictor) and a second variable (criterion), and the other predicts that the variable of interest predicts the second. Demonstrates a predictor-criterion relationship.

Convergent-discriminant validity design (of a measure)

Tests the variable of interest in relation to two other variables: one that goes with it/is related to (convergent) and one that doesn’t go with it/is not related to (discriminant)

Correlation is derived from…

covariance

Covariance can indicate _____ and ______, but not ______.

Presence of covariation, direction of covariation, how much covariation

linear

in reference to a straight line

best fitting line

a straight line closer to all points than any other straight line

r = undefined, indeterminate

when there is variation in one variable but not the other, so covariation cannot be indicated

Statistical rule of life about interpreting r

plot the data before interpreting (is there an absence of covariation or the absence of linear covariation?)

when r=.00, there are two possibilities:

non-linear covariation or no covariation

slope is…

the departure from the absence of slope (positively or negatively)

r²

indicates the proportion of variation in X that has been observed in relation to the variation of Y(how much of the variation in the data is explained by this relationship)

Venn Diagram

Each Venn indicates the proportion of variation of the variable. Overlap indicates the amount of shared variation (variation explained by this relationship)

Based on an r value, one cannot tell…

the order of covariation (causality)

Problems of range restriction/selection effect/small sample

creates an artifact through a limited range of variation (fix through larger sample)

When your expected r vales doesn’t match your observed, you’re not an idiot. You should…

Check and see if the scatterplot suggests an artifact of range restriction/selection effect/a small sample