NON-PARAMETRIC

STATISTICS
PART II

AEC 132: STATISTICS WITH

SOFTWARE APPLICATION
(1)KAPPA MEASURE OF AGREEMENT

¡ It is used in the medical literature to assess

inter-rater agreement ( e.g diagnosis from two
different clinicians) or the consistency of two
different diagnostic tests ( new developed test
versus a global standard)
EXAMPLE
¡ Research Question
¡ How consistent are the diagnostic classifications of X Depression
Scale and the Depression, Anxiety and Stress Scale?
¡ What you need:
¡ Two categorical variables with an equal number of categories
(e.g. diagnostic classification from rater 1, 0 = not depresses, 1 =
depressed; and the diagnostic classification of the same person from
rater 2)
¡ Assumptions: Assumes equal number of categories from Rater 1
and Rater 2
PROCEDURE FOR KAPPA MEASUREMENT OF
AGREEMENT
¡ Open data set (KappaTest.sav)
¡ Click analyze-descriptive-statistics-crosstabs
¡ Click on one variable ( DASSdepgp2), move it to box marked
Row(s)
¡ Click on one variable ( EPDSgp2), move it to box marked Column(s)
¡ Click on the Statistics – Kappa – Continue
¡ Click Cells button
¡ In the Counts box, click on Observed
¡ In the Percentage section, click on Row. Click on Continue and then OK
INTERPRETATION OF OUTPUT
¡ The symmetric measures
shows that the Kappa
Measure of Agreement value
is .56 with a significance
value of .000
¡ According to Peat (2001) CONCLUSION: The level of
agreement between the
value of .5 is moderate
classification of cases as depressed
agreement, above .7 good using the EPDS and DASS-Dep is
agreement and .8 very good moderate.
agreement.
(2) MANN-WHITNEY U TEST

¡ Used to test for differences between two independent

group on a continuous measure.
¡ Example: Do males and females differ in terms of their self-
esteem?
¡ An alternative test for t-tests for independent samples.
However, instead of comparing means, it actually
compares medians.
(2) MANN-WHITNEY U TEST

¡ It converts the scores on the continuous variable to

ranks, across the two groups. It evaluates whether the
ranks for the two groups differ significantly.
¡ As the scores are converted to ranks, the actual
distribution of the scores does not matter.
EXAMPLE
¡ Research Question
¡ Do males and females differ in terms of their level of self- esteem? Or
¡ Do males have higher levels of self-esteem than females?
¡ What you need:
¡ One categorical variable with two groups ( e.g. SEX)
¡ One continuous variable ( e.g. total self-esteem)
¡ Parametric Alternative: Independent-samples t-test.
PROCEDURE FOR MANN-WHITNEY U TEST
¡ Open data set (Mann-WhitneyUTest.sav)
¡ Click analyze-Nonparametric Tests- Legacy Dialogs- 2
Independent Samples
¡ Click on continuous (dependent) variable (e.g. total sel-esteem) and move
it into the Test Variable List box.
¡ Click on your categorical (independent) variable (e.g. sex) and move it into
the Grouping Variable box
¡ Click on the Define Groups button. Type in the value for Group 1 ( e.g. 1)
and for Group 2 ( e.g. 2).
¡ Tick Mann-Whitney U under the section labelled as Test Type
¡ Click on Options and choose Descriptive
¡ Click on Continue and then OK
INTERPRETATION OF OUTPUT
¡ Note: If your sample size is
larger than 30, SPSS will give
the value for a z-approximation
test, which includes a
correction for ties in the data.
¡ The z-value is -1.227 with a
significance level of p=.220. The CONCLUSION: There is no
probability value is not less statistically significant difference in
than or equal to .05, so the the self-esteem scores of males
result is not significant. and females
INTERPRETATION OF OUTPUT

¡ Note: If the result is statistically difference between

your groups, you will need to describe the direction of
the difference (which group is higher).

¡ Refer to the Ranks table under the column Mean

Rank. Report the median values for each group.
PROCEDURE FOR OBTAINING MEDIAN
SCORES FOR EACH GROUP
¡ Open data set (Mann-WhitneyUTest.sav)
¡ Click analyze-Compare Means – Means
¡ Click on continuous variable (e.g total self-esteem) and move it into
the Dependent List box
¡ Click on the categorical variable (e.g. sex) and move it into the
Independent List Box
¡ Click on the Options button. Click on Median in the statistics
section and move into the Cell Statistics box. Click on Mean and
Standard Deviation and remove from the Cell Statistics box.
Click on continue
¡ Click on OK
 CONCLUSION: A Mann-Whitney U Test
revealed no significant difference in the self
esteem levels of males ( Md = 35, n=184) and
females ( Md = 34.5 n= 254), U = 21594, z= -
1.23, p = .22, r =.06
EFFECT SIZE
¡ EFFECT SIZE
¡ r=z/square root of N
¡ Measure of Effect Size
¡ EXAMPLE: according to Cohen
(1988)
¡ Z = -1.23; N = 436
.10 – small effect
¡ What is the value of r?
.30 – medium effect
ANSWER: .06 .50 – large effect
MANN-WHITNEY U TEST

CONCLUSION: A Mann-Whitney U Test

revealed no significant difference in the self
esteem levels of males ( Md = 35, n=184) and
females ( Md = 34.5 n= 254), U = 21594, z= -
1.23, p = .22, r =.06
(3) WILCOXON SIGNED RANK TEST

¡ Also known as Wilcoxon matched pairs signed rank tests

¡ It is designed for use with repeated measures, that is, when subjects
are measured on two occasions or under two different conditions.

¡ It converts scores to ranks and compares them at Time 1 and at

Time 2.
EXAMPLE
¡ Research Question
¡ Is there a change in the scores on the Fear of Statistics Test from Time
1 to Time 2?
¡ What is needed
¡ One group of subjects measured on the same continuous scale or
criterion on two different occasions. The variable involved are
scores at Time 1 or Condition 1, and scores at Time 2 or
Condition 2.
¡ Parametric Alternative: Paired sample t-test
PROCEDURE FOR WILCOXON SIGNED RANK TEST
¡ Open data set (WilcoxonTest.sav)
¡ Click analyze-Nonparametric Tests – 2 Related Samples
¡ Click on the variables that represent the scores at Time 1 and at
Time 2 ( e.g. fost 1, fost 2). Move these into the Test Pairs List
box.
¡ Tick Wilcoxon box in the Test Type section
¡ Click on Ok
¡ Click on the Options button Choose Quartiles (this will provide
the median scores for each time point)
¡ Click on Continue and then on OK
INTERPRETATION OF OUTPUT
¡ If the significance level is
equal to or less than .05, it
can be concluded that the
difference between the two
scores is statistically
significant.
It can be concluded that
the two sets of scores are
significantly different.
EFFECT SIZE
¡ EFFECT SIZE
COMPUTE FOR THE EFFECT ¡ Measure of Effect Size
according to Cohen
SIZE?
(1988)
¡ r=z/square root of N .10 – small effect
¡ r=.76 .30 – medium effect
.50 – large effect
INTERPRETATION OF OUTPUT

¡ A Wilcoxon Signed Rank Test revealed a statistically

significant reduction in fear of statistics following
participation in the training program, z = -4.18, p<.001,
with a large effect size ( r = .54).
The median score on the Fear of Statistics Scale
decreased from pre-program ( Md = 40) to post-
program (Md = 38)
(4) KRUSKAL-WALLIS TEST

¡ Also referred as Kruskal-Wallis H Test

¡ An alternative to a one-way between-groups analysis of
variance.
¡ It compares the scores on some continuous variable for
three or more groups.
¡ Scores are converted to ranks and the mean rank for each
group is compared.
¡ It is a between groups analysis, hence different people
must be in each of the different groups
EXAMPLE
¡ Research Question
¡ Is there a difference in optimism levels across marital?
¡ What is needed
¡ One categorical independent variable with three or more
categories
¡ One continuous dependent variable
¡ Parametric Alternative: One-way between-groups analysis of
variance
PROCEDURE FOR KRUSKAL-WALLIS TEST
¡ Open data set (KruskalwallisTest.sav)
¡ Click analyze-Nonparametric Tests – Legacy Dialogs - K
Independent Samples
¡ Click on the continuous, dependent variable (e.g. total optimism) and move
it into the Test Variable List box
¡ Click on categorical, independent variable (e.g. marital status) and move it
into the Grouping Variable box
¡ Click on the Define Range button. Type in the first value of your
categorical variable (e.g.1) in the minimum box. Type the largest value for
your categorical variable (e.g. 8) in the maximum box. Click continue
¡ In the Test Type section – tick Kruskal-Wallis H
¡ Click on Continue and then on OK
¡ The rank table will show
which groups have highest
overall ranking on the
dependent variable.
In this example, Widowed
has the highest optimism
level.
¡ Test of statistics shows that
optimism of respondents
when grouped according to
marital status is significantly
difference (p=.021).
(5) FRIEDMAN TEST

¡ A non-parametric alternative to the one-way

repeated measures analysis of variance.
It is used when you take the same sample of
subjects or cases and measure them at three or
more points in time or under three different
conditions.
EXAMPLE
¡ Research Question
¡ Is there a change in Fear of Statistics scores across three time periods
( pre-intervention, post-intervention and a follow-up)?
¡ What is needed
¡ One sample of subjects, measured on the same scale or measured
at three different time periods, or under three different
conditions.
¡ Parametric Alternative: Repeated measures (within-subjects)
analysis of variance.
PROCEDURE FOR FRIEDMAN TEST
¡ Open data set (FriedmanTest.sav)
¡ Click analyze-Nonparametric Tests – Legacy Dialogs - K
Related Samples
¡ Click on the variables that represent the three measurements (e.g.
fost1, fost2, fost3).
¡ In the Test Type section, check that the Friedman option is
selected
¡ Click on the Statistics button. Tick Quartiles. Click on Continue
¡ Click on OK.
 RESULTS
The results of this test suggest
that there are significant
differences in the Fear of
Statistics scores across the three
time periods. This is indicated by a
sig level of .000. Comparing the
mean ranks for the three sets of
scores, it appears that there was a
decrease in Fear of Statistics
scores over time.
 RESULTS
The results of the Friedman Test
indicated that there was a statistically
significant difference in Fear of
Statistics Test scores across the three
time points ( pre-intervention, post-
intervention, 3month follow up), X2
(2, n=30) = 41.57, p<.005. Inspection
of the median values showed a
decrease in fear of statistics from pre
intervention (MD= 40) to post-
intervention (Md=38) and a further
decrease at follow-up (MD = 35.5)
ACTIVITY
¡ Use file staffsurvey.sav
¡ Prepare your interpretation of the following problems:
b. Use the Mann-Whitney U Test to compare the staff
satisfaction scores (totsatis) for permanent and casual staff
(employstatus).
c. Conduct a Kruskal-Wallis Test to compare staff
satisfaction scores (totsatis) across each of the length of
service categories (use the servicegp3 variable).
Note: Upload your output in the Schoology and
gdrive.