DATA PROCESSING AND
STATISTICAL TREATMENT
Mark C. Maravillas, MaEM
ANALYZING THE DATA
MARK C. MARAVILLAS
Analysis is the resolution of
information with the use of
statistical method (Punsalan and
Uriarte, 1989).
It is now understood that a
classroom teacher-researcher needs
to distinguish what kind of analysis
he or she ought to employ.
For everyone’s sake there are three
(3) analyses, namely (a) descriptive
analysis, (b) correlation analysis,
and (c)inferential analysis.
Scales Measurement of Data
1. Nominal types of data
2. Ordinal types of data
3. Interval types of data
4. Ratio types of data
1. Nominal data connotes name and has no
numerical value
Example: sex, civil status, religious
affiliation
2. Ordinal data are typically ranked
Example: scales like strongly agree, agree,
disagree, strongly disagree; excellent, very
satisfactory, poor, very poor
2. Interval refers to numerical data
where zero is not significant
Example: temperature
3. Ratio data is a numerical level of
measurement where zero is significant .
Example: Scores in the test, height,
weight.
Name what data are these?
Descriptive Analysis analyzes the
present situation with the use of the
following statistical tools: Percentage,
Mean, Median and Mode.
Example:
Specific Question: What is the profile of
the respondents in terms of:
Sex, Length of Service and Educational
Qualification?
With this situation, we need to employ mode and
percentage. After tallying by encoding resulting in
frequencies, we have to run the SPSS.
Example:
Specific Question: What will be the
general academic performance in
Statistics of the two groups of learners
exposed and not exposed to XYZ
Intervention?
To deal with this question, we need to use mean
since the academic performance is measured
through grades or scores. These are the so-called
ratio data. Then, we will run the SPSS.
Example:
Specific Question: What is the teachers’
perceived level of impact of action
research on teaching?
This time the data to be analyzed are ordinal, so
w e n e e d m e di a n t o d e s c r i b e s u c h d a t a .
Summary of the measures of central tendency and the data requirement
for analysis (Almeda, Capistrano and Sarte, 2010)
Measures of Central Tendency Data Requirement
Mean At least interval scale and
values that are close to each
other
Median At least ordinal scale
Mode Even if nominal only
That’s all for descriptive analysis
Cor r el ati on Analy sis at t em pt s t o see t he
r el at i o nshi p tw o v ari a bl es . I n t hi s an al ys i s, w e
w i l l de t er mi n e w h eth er t h er e i s s uch ki nd of
c or r e l ati on w hi ch i s ei t her degr ee of
r el at i on shi p or si g ni f i c ant r el a t i on shi p.
I f w e u se s i gni f i ca nt r el at i ons hi p w e have t o
j um p t o i nf eren ti al ana l ysi s f o r w e w i l l consi der
l eve l s of si gni f i can ce, the 1% and 5% l evel s o f
s i gni f i ca nce. Th e co mmo n cor r el a t i on t ol l s u sed
f or cl a ssr o om acti on re sea r che s w i t h r est r i ct i on
t o d egr ee o f re l ati onsh i p ar e Sp ear m an r ho and
Pe arson ’s r .
When to use these tools appropriately?
For Spearman rho, parameter should be
non-parametric. This entails the use of
ordinal data since these data fall under
non-parametric. Therefore, Spearman rho
is used when data belong to non-
parametric.
Example:
If a classroom teacher-researcher wants
to correlate the perception on
motivational disciplinary exercise and the
perception on XYZ Principle, then he will
correlate 2 ordinal types of data.
Example:
If a classroom teacher-researcher wants
to correlate the age and the perception
on XYZ Principle, then he will correlate
ratio and ordinal types of data.
Example:
If a classroom teacher-researcher wants
to correlate temperature and the
perception on XYZ Principle, then he will
correlate interval and ordinal types of
data.
Al l of t hese i nst ances ar e appr opr i at e f or
Spear man r ho .
The 3 exampl es may be si mpl i f i ed by
showi ng t he i ll ust r at i ons bel ow:
Variables to be Correlated Appropriate Statistical Tool
Ordinal vs. Ordinal Spearman rho
Ordinal vs. Ratio Spearman rho
Ordinal vs. Interval Spearman rho
For Pearson’s r, on the other hand,
parametric test is used. The analyses are
parametric when the variables to be
correlated are either both ratios and both
intervals, or the two (2) variables are
ratio and interval or vice versa. Ratio and
interval data pose parametric tests.
Example:
A classroom teacher-researcher wishes to
correlate two variables like age and academic
performance (grades) of his learners. In this
situation, two (2) variables are apparent.
They are age and grades of the learners. Age
and grades are both ratio data that fall under
parametric test. Thus, this shows that we are
correctly guided to apply Pearson’s r.
Al l of t hese i nst ances ar e appr opr i at e f or
Pear son’ s r.
The 3 exampl es may be si mpl i f i ed by
showi ng t he i ll ust r at i ons bel ow:
Variables to be Correlated Appropriate Statistical Tool
Ratio vs. Ratio Pearson’s r
Ratio vs. Interval Pearson’s r
Interval vs. Interval Pearson’s r
That’s all for
Correlational
Analysis
Inferential Analysis infers the difference of the two
or more variables to be correlated to get the
significant level. But we will limit our analysis to
two groups of comparison and to variables for
correlation.
To determine whether the analysis is inferential or
not, a classroom teacher-researcher has to spot if
the phrasal expressions like “Significant Difference,
Significant Relationship are explicitly stated in the
specific questions and in the hypothesis. This is the
moment when a classroom-teacher researcher
concludes that his or her action research proposal
needs inferential analysis.
Pairing of variables, test and
appropriate tool
Pairing of Variables Test Appropriate Tool
Comparison
Before and After Non-Parametric* Wilcoxon
Two Independent Groups Non-Parametric* Mann-Whitney U
Two Independent Groups Parametric** Independent-Samples T-Test
Before and After Parametric** Paired-Samples T-Test
Correlation Non-Parametric* Spearman rho
Parametric** Pearson’s r
*requires ordinal data
**requires ratio or interval data
Example:
Hypothesis : There will be no significant
difference on the perceptions of learners
on COVID 19 vaccination before and after
the discussion of vaccine efficacy by the
RHU.
Now, the data to be analyzed are non-parametric because
these are ordinal data. So Wilcoxon is to be used for analysis.
Example:
Hypothesis : There will be no significant
difference on the perceptions of learners on
COVID 19 vaccination when data are grouped
according to gender .
Still, the data to be analyzed are non-parametric; hence, Mann-
Whitney U is to be applied for analysis. It should be the Mann-
Whitney U test to be applied for because there are two
independent groups to utilize the ordinal type of data.
Example:
Hypothesis : There will be no significant
difference on the academic performance of
learners on the use of XYZ Intervention before
and after the discussion of frequency
distribution table.
Notice that there is a change of data. We may now use
academic performance (grades) and this qualifies for a ratio
data. A ratio is parametric, and we have two groups which are
before (pre) and after (post). The appropriate tool for this is T-
Test for Paired Samples.
That’s all for
Inferential
Analysis
Interpreting Analysis Results
Interpreting degree of relationship between two variables
using Spearman rho and Pearson’s r (Muijs, 2004)
Values Descriptive Interpretation
<0.+/-1 Weak
<0.+/-3 Modest
<0.+/-5 Moderate
<0.+/-8 Strong
≥=+/-0.8 Very Strong
Interpreting significant relationship and difference using
Pearson’s r, Spearman rho, Wilcoxon, Mann-Whitney U,
Paired-Samples T-Test, and Independent-Samples T-Test.
Values Descriptive Interpretation
.000 to .010** Significant at 1% level
.011 to .050* Significant at 5% level
Above .50 Not Significant
Thank you!