Geog 339
–
Definitions

Wilcoxon Rank Sum Test
: data are ordinal; compare two independent random sample rank
sums for difference.

ANOVA
: determine which is more dominant; between group variability or within.
F>F
cutoff
: reject null

Betweengroup variability
: how the sample mean of each group differs from the total mean
when all categories are grouped together.

Withingroup variability
: measures the variation of observations in each group about the mean
of that group.

Independent Samples
: items collected from one sample do not relate to items collected in the
other sample.

Dependent Samples
: matched pairs of observations (sampling the same thing before and after
event Matchedpairs).

Parametric
: interval and ratio scale, require population parameters to be estimated using
samples; make assumptions about the underlying distribution (normal).

NonParametric
: ordinal scale (ranked), nonnormal distribution, no parameter estimates.

PVE
: pooled variance estimate  when equality of variances is assumed in a twosample
difference of means ttest, longer formula for denominator.

SVE
: separate variance estimate  when variances are not assumed equal

Type I error
: the probability of rejecting a true NULL hypothesis 
more serious, = α.

Type II Error
: the prob
ability of accepting a false NULL hypothesis, = β.

Onetailed test
: directional 
area to right of α is rejected, are to left accept, H
O
: u=u1, H
A
: u<u1.

Twotailed test
: nondirectional 
area= α/2, area between the two α's accept H
O
, outside area
reject H
O
.

INFERENTIAL STATISTICS
: methods used to make inferences about a population from a sample.

Inferential Test
: a specific test chosen because it suits the particular problem at hand (e.g.,
difference of means).

Null Hypothesis
: Claim for equality  no significant difference between samples.

Alternate Hypothesis
: condition under which null is rejected, and a statement of difference
between samples.

Goodnessoffit
: Actual frequency distribution tested against an expected frequency
distribution.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
KS test for Normality
: compares the cumulative relative frequencies (CRF) of observed sample
data with that of a perfect normal distribution. If the two OGIVEs “match”, the sample
distribution can be considered normal. CRF
e
= CNV in formula sheet, CRF
o
= x/n, 1/7, 2/7 cumula

CHISquare GoF
: determine if a truly significant difference exists between observed and
expected frequencies

Contingency Analysis
: Frequency distributions between two variables are compared with one
This is the end of the preview.
Sign up
to
access the rest of the document.
 Fall '10
 staff
 Normal Distribution, Frequency distribution, clearly identified subset

Click to edit the document details