Chapter 9
Card Set Information
Author:
Vcrawford
ID:
223618
Filename:
Chapter 9
Updated:
2013-06-12 18:03:37
Tags:
McMillan
Folders:
Description:
Research class
Show Answers:
Inferential statistics
infers the characteristics of a population
null hypothesis
Hypothesis that concludes there is no relationship or difference in a measure
level of significance
probability of being wrong in rejecting the Ho
reported as
p
= x; typically level of significance is <.05
Type 1 Error
Reject Ho but it turns out there is no difference or relationship in the population
Type II Error
Fail to reject the Ho when there IS a difference/relationship in population
alpha level
level of significance set prior to data collection as a criterion for rejecting Ho
level of significance affected by 3 factors
1. groups being compared - greater the difference, smaller the p value
2. degree of sampling and measurement error (SD) lower the error, smaller the p value
3. size of the sample (N) - large sample, p will have smaller value than in small sample
confidence intervals
provide a range of values in which the population or "real" trait value lies with specific probabilities
confidence intervals measured by:
1. using sample data, calculate standard error of the mean Sx
2. this value used to create intervals around sample mean that correspond to the probability of obtaining a population value in that interval
ex - sample mean is 60; researcher has 95% confidence interval of 48-72 - 95% chance that population or true mean is in that interval
Effect size
way of quantifying the degree of difference b/t two groups; also coefficient of determination
X1-X2/SD = cohen's D
parametric tests
used when certain assumptions can be made about the data - normally distributed, equal variance, interval level measures
t-test (parametric)
tests null hypothesis
1. independent-samples t-test
: different subjects in each group
2. paired dependent-samples/correlated/matched
: subjects in the groups are paired or matched in the same way
degrees of freedom (df)
used to calculate the level of significance
approximately equal to # of subjects in the study
ANOVA
simple analysis of variance
compares group means to determine the probability of being wrong in rejecting Ho (like t-test)
independent variable has multiple levels
Simple/One-Way ANOVA
single independent variable analyzed w/single dependent variable
ex
: study 3 types of students and means... students from SES h/m/l. 1x3 ANOVA
F statistic calculated from variance of the groups
Two-Way Anova
factorial analysis of variance
2 or more i.v.s are analyzed together
test for each i.v.
ex
: one i.v. has 2 levels, one has 3, 2x3ANOVA
ANCOVA
analysis of covariance
adjusts for pretest differences b/t groups
pretest is the covariate
ex
: 1 grp has mean of 15 and other has mean of 18 on a pretest; ANCOVA used to adjust posttest scores statistically to compensate for 3 pt difference
Multivariate Statistics
two or more dependent variables are analyzed together
MANCOVA
Hotelling's T
Chi Square
x
^{2}
, c
^{2}
^{}
used when researchers are interested in # of responses or cases in different categories
results reported in a contingency table
ex
: relationship b/t gender and book choice
m/f, 4 book types to choose from, 2x4 table