Bivariate and Trivariate Relationships Flashcards
what is covariation?
association/ a relationship
what is independence when regarding variables?
no pattern/association
what is a hypothesis?
a tentative statement about a relationship between variables
what is a directional/one-tailed hypothesis?
a hypothesis that states the direction of a relationship
what is a non-directional/two-tailed hypothesis?
a hypothesis that doesn’t state the direction of a relationship
are most hypotheses directional or non-directional?
non-directional
what is a positive relationship?
variable A increases when variable B increases and vice versa
what is a negative relationship?
variable A increases when variable B decreases and vice versa
how can you establish causation?
-replication
-controlling for a 3rd variable
-manipulate the IV
-use logic and reasoning
what is induction?
qualitative and exploratory approach; start with specific observations to general theory
what is deduction?
quantitative approach; start with general theory to specific observations
which levels of measurement can contingency tables be used for?
any, but primarily…
- nominal (x) and nominal (y)
- ordinal (x) and ordinal (y)
- nominal (x) and ordinal (y)
in a contingency table, what is the maximum amount of difference allowed between DV values?
it should be less than or equal to 10%
in a contingency table, what is the maximum amount of categories a DV and IV can have?
DV should have less than 7 categories and IV should have less than 4 categories
in a contingency table, what are the marginals?
the totals
in a contingency table, where should the DV and IV be placed?
DV should be at the very left and IV should be at the very top
what is a measure of association (correlation coefficient)?
a single # expressing strength and direction of a bivariate relationship
what are characteristics of correlation coefficients?
- some only detect linear, others only detect non-linear
- can include direction of relationship when using minimal ordinal level
- strength is always been 0 and (-)1
what is a symmetric coefficient?
unaffected by which variable is the IV; value remains the same whether X is IV and Y is DV or X is DV or Y is IV
what is an asymmetric coefficient?
affected by which variable is the IV; value of coefficients are different and depend on which variable is the IV
which level of measurement is most appropriate for correlation coefficients?
interval/ratio
what is proportion reduction error (PRE)?
looks at how well we can predict the DV value using the IV value
how does PRE determine strength of a relationship?
fewer errors in predictions means a stronger relationship
PRE compares errors in which 2 situations?
- using the IV to make predictions
- not using the IV
which levels of measure is lambda appropriate for?
- nominal (x) and nominal (y)
- nominal (x) and ordinal (y)
- ordinal (x) and nominal (y)
which measure of central tendency is best to use with lambda?
the mode
does lambda show direction and/or strength?
lambda only shows strength, not direction
which level of measurement is most useful in lambda calculations?
nominal
what is E1 in the lambda formula?
errors made without using the IV for prediction (found in the marginal)
what is E2 in the lambda formula?
errors made using the IV for prediction
for lambda, how do you find errors in contingency tables?
sum of the cells excluding the mode
what are limitations of lambda?
- less useful for variables beyond nominal
- doesn’t show distribution among categories, thus can’t detect direction of association
what is gamma?
predicts the ordinal arrangement of values
in gamma, what is the pairs concept?
looks at relative ranking of 2 parts of the pair on both variables
what is the unit of analysis in gamma?
the pair
what are the types of pairs?
concordant, discordant, tied
what is a concordant pair?
in accordance with each other; two IV levels go up and down together on the DV
what is a discordant pair?
not in accordance with each other; two IV levels inconsistently go up and down together on the DV
what is a tied pair?
exact same values; two IV levels have the same value on one/more DV levels
what is Ns and Nd in the gamma formula?
Ns is the number of concordant pairs and Nd is the number of discordant pairs
how do you calculate Ns in gamma?
using raw data, multiply the cell frequency by sum of all frequencies below and to the right of it, then add all products
how do calculate Nd in gamma?
using raw data, multiply the cell frequency by sum of all frequencies below and the to the eft of it then add all products
what indicates a positive relationship in gamma?
if there are more concordant pairs than discordant
what indicates a negative relationship in gamma?
if there are more discordant pairs than concordant
what are the characteristics of gamma?
- best to use when both variables have few categories
- symmetric (doesn’t matter which variable is the IV)
- assumes linearity
- ignores pairs tied on X, Y, or both X and Y
what is an elaboration model?
looks at what happens to a zero-order when you control for a 3rd variable
what is a zero-order relationship?
the original bivariate relationship; without the 3rd variable
which criteria of causation does the elaboration model determine?
non-spuriousness
in an elaboration model, what the do the partial results include?
the 3rd variable
what are the 4 elaboration models?
replication, explanation, interpretation specification
what is the replication elaboration model?
partial and original order are the same; all relationships are small in zero-scale
what is the explanation elaboration model?
partial relationships are 0 or less than original order, and the 3rd variable is antecedent
what is an antecedent variable?
comes before the IV in time
what is the interpretation elaboration model?
partial relationships are 0 or less than original order, and the 3rd variable is intervening
what is an intervening variable?
comes after the IV in time
what is the specification elaboration model?
one partial is the same or more than the original order and the other partial is 0 or less than the original order