13. Assessing Relationships 2 Flashcards
measure (or questionnaire) consistently reflects the construct it’s measuring
reliability
most common measure of reliability
Chronbach’s alpha
useful with questionnaires
Chronbach’s alpha scores
should be .7 to .8 for test retest reliability
ICC
measures relationship between 2+ variables that measure the same thing
Measurements ICC can be used with
- single measurement
- mean of several measurement
ICC can be used to assess:
- inter-rater reliability
- intra-rater reliability
3 models of ICC: model 1
each subject assessed by different set of raters
3 models of ICC: model 2
- each subject assessed by same set of raters
- raters are representative of the population
3 models of ICC: model 3
- each subject assessed by same set of raters
- raters only reliable for their own study
Which ICC model is least common?
model 1
When is regression used?
when you’re trying to predict an outcome (or DV)
variables used to predict the outcome
predictor variables (or IV)
Where do predictor variables come from?
previous research
types of regression
- simple linear
- multiple linear
- logistic
1 IV
1 categorical DV
logistic regression
1 IV
1 continuous DV
simple linear regression
> 1 IV
1 continuous DV
multiple linear regression
data for regression
linear
assumptions of a regression: variables
- must be continuous or categorical (with only 2 categories)
- DV is continuous
assumptions of a regression: no perfect multicollinarity
two or more predictor variables shouldn’t be highly correlated (i.e. IVs should be measuring different things)
multicollinarity and simple linear regression
not a problem
assumptions of a regression: confounding variables
predictors are uncorrelated with confounding variables
assumptions of a regression: homoscedasticity
- similar to homogeneity of variance
- variance should be equal among predictor variables
assumptions of a regression: independence
- values of outcome variable are independent
- come from a separate entry
assumptions of a regression: linearity
relationship of IVs and DV is linear
What are the assumptions of a regression?
- predictor variables continuous or categorical and DV is continuous
- no perfect multicollinarity
- no confounding variables
- homoscedasticity
- independence
- linearity
b_1
slope of the line
b_0
y-intercept
What are b_1 and b_0?
regression coefficients
steps to performing a regression analysis
- assess the model as a whole
- assess individual predictor variables
How to assess the model as a whole
goodness of fit
