W9: Data Analysis III Flashcards
What is the F ratio in an ANOVA test statistic?
F compares 2 estimates of variability (between and within )
What does ANOVA examine?
variability of samples
When is Vb greater than Vw (ANOVA table)?
if there is a significant difference between groups
The larger the F value, the greater the ratio between __ to __?
the greater the ratio between Vb to Vw; higher F means lower P
What is the F distribution used for?
to test hypotheses about population variances; defined mathematically (and theoretically based)
Do people of different age groups differ in number of hours of TV watched per day?
F = 8.903 at p< 0.001
Accept or reject null hypothesis?
Conclusion?
reject null hypothesis
Conclusion: there is a difference between age groups for TV hours watched
What is a post hoc test? When is it used?
tells you which variables are different from one another.
Used only when p<0.05 because difference is confirmed
What are the different types of post hoc tests?
Bonferroni, Tukis, Duncan (unreliable)
What is the simplest type of regression?
multiple linear regression- calculate slope and add additional variables
What is the 3 M approach?
model, measurement, method
What is the limitation of ANOVA?
F does not pinpoint which means are significantly different from each other
What are the limitations of ANOVA?
F does not pinpoint which means are significantly different from each other
need post hoc multiple comparison procedure
When is a post hoc multiple comparison procedure not needed?
when the F is not significant
What is a two way ANOVA?
extension of one way ANOVA, 2 simultaneous IVs of 2+ categories each (nominal/ordinal IV), 1DV (ratio)
what is an MANOVA?
multivariate analysis of variance; extension of ANOVA with 2+ DVs
(post hoc) If a significant main effect or interaction is found, then you can only conclude that there is…
a significant difference amongst levels of IVs somewhere
What is a bivariate correlation?
describes how closely 2 ratio level variables co-vary together
what is the difference between bivariate correlations and multiple regression?
bivariate-> analysis of 2 ratio level variables
multiple regression-> analysis of more than 2 ratio level variables (many variables analyzed together)
What can bivariate correlations and multiple regressions do in relation to causal relationships?
cannot prove causal relationships but can provide evidence to support causal arguments
When is multiple regression used?
to examine impact of several independent variables on dependent variable
- in applied research
Give an example scenario when multiple regression is used
to predict how long a senior stays in hospital post-surgery, you might consider many possible predictors (age, severity of illness, physical function scores, depression scores)
What is the difference between multiple nonlinear regression vs. multiple linear regression?
nonlinear fits variables on a best fitting “curve”
linear fits variables on best fitting line
What makes multiple linear regression a powerful tool?
-isolates separate effects of IVs on DV
-estimates which IV is strongest predictor of DV
-identifies math equation describing linear relationship between IVs and DV
What happens during the “Model” of 3Ms?
- determine which variable will be treated as DV and which will be treated as IVs
- specify greater than and less than relationships between variables
What happens during the “Measurement” of 3Ms?
identify level of measurement (nominal, ordinal, ratio) for reach variable
What happens during the “Method” of 3Ms?
determine which method (stat procedure) appropriate for examining relationships between variables
eg. Crosstabs (chi-square)
means (ANOVA)
correlations