exam 17 - 3

Question 1

cluster sampling

eg if have 30 students 15 fem and 15 male

Answer 1

would of had a total of 15 schools with two students per high school (presumably one boy and one girl from each)

Question 2

nominal, numeric, ordinal

Answer 2

gender = niminal
no. of friends = numeric
group (small/med/large) = ordiinal
reading comprehension score = numeric

Question 3

3 assumptions for one-way ANOVA

- comment on the validity of each assumption w/ reference to box plot or overall study design

Answer 3

1. independant groups - since students were randomly chosen from different schools, we can assume scores were independent
2. Normality - whether symmetrical or post/neg skewed
3. Equal variances - seem similar, if have large tail and outlyer - can give overall impression of unequal variances

Question 4

How would omitting/excluding an outyer affect the outcome of the analysis.

Answer 4

omitting the outlyer will riduce the residual variability and so would likely give stronger evidence of a difference

Question 5

grouping the no of people into groups (small/med/arge) as explanatory variable instead of individiual score numbers
- One-way ANOVA
why would this one-way ANOVA been preferable to a linear regression model using the number of friends ?

Answer 5

Improves power since relationship may not be linear

Question 6

when looking for p’s

Answer 6

sig (2-tailed) in independant samples test