Research Framework, Strategies, and Methodologies
Research Framework
Creswell
Philosophical Worldviews
Ontology
Epistemology
4 Worldviews
Post-positivism
Determination
Reductionism
Empirical observation and measurement
Theory verification
Constructivism
Understanding
Multiple participant meanings
Social and historical construction
Theory generation
Advocacy/Participatory
Political
Empowerment issue-oriented
Collaborative
Change-oriented
Pragmatism
Consequences of actions
Problem-centered
Pluralism
Real-world-practice oriented
Strategies
Surveys
Case studies
Experiment
Ethnographies
Action research
Grounded theory
Phenomenological analysis
Narratives
Approaches
Quantitative
Qualitative
Case study
Single case study
Multiple case study
Phenomenological analysis
Ethnography
Grounded theory
Mixed
Methods
Data collection
Data analysis
Data interpretation
Spearman's Correlations
Actual
Purchase of Actual # of Household
Variable Household
drinking water Members
gross income
1. Actual
Spearman's
Household —
rho
gross income
p-value —
2. Purchase of Spearman's
0.054 —
drinking water rho
p-value 0.526 —
3. Actual # of
Spearman's
Household 0.287 -0.003 —
rho
Members
p-value < .001 0.970 —
Pearson's Correlations
Actual
Purchase of Actual # of
Variable Household
drinking water Household Members
gross income
1. Actual
Pearson's
Household —
r
gross income
p-value —
2. Purchase of Pearson's
0.076 —
drinking water r
p-value 0.367 —
3. Actual # of
Pearson's
Household 0.151 0.007 —
r
Members
p-value 0.073 0.932 —
Pearson's r can range from -1 to 1. An r of -1 indicates a perfect negative linear relationship between
variables, an r of 0 indicates no linear relationship between variables, and an r of 1 indicates a perfect
positive linear relationship between variables.
Higher rho coefficients denote a stronger magnitude of the relationship between variables. Smaller rho
coefficients denote weaker relationships. Positive correlations denote a relationship that travels at the
same trajectory. As one value goes up, then the other value goes up.
A p-value less than 0.05 is typically considered to be statistically significant, in which case the null
hypothesis should be rejected. A p-value greater than 0.05 means that deviation from the null
hypothesis is not statistically significant, and the null hypothesis is not rejected.
Data Presentation
Type of Data
Quantitative/Numerical Data
Interval
Ratio
Qualitative/Categorical Data
Nominal (Sex, Status, Marital, Zip code)
Ordinal (Educational attainment, etc)
Methods of Data Presentation
Textual method
Tabular method
Graphical method
Statistical Test Assumptions
Parametric Test
T-Test
Anova (Analysis of Variance)
Level of measurement of the variables
Nominal
Ordinal
Interval (lacks zero score)
Ratio (with zero score)
Normality test
How to test for normality?
Shapiro-Wilk Test (Stat method) Descriptives>Statistics>Shapiro-Wilk
All variables should be normal
Hypotheses:
Ho: The data comes from a normal distribution
Decision rule:
Using the p-value, 5%-10% level of significance if social sciences, 1% level of
significance if medical research.
Homogeneity of variance test
How to test the homogeneity of variances?
Using the Levene’s test
The variances of the variable across groups should be equal
Linearity
How to test the linearity?
Using the graphical scattergraph
No significant outliers
Checking for the significant outlier
Boxplots
IQR method
If there’s an outlier,
1. There maybe a measurement error or data entry error. Correct the error if
possible.
2. That may not be part of the population you are studying (i.e., unusual
properties of conditions), you can legitimately remove the outlier.
3. That maybe natural part of your study.
If two or more assumptions were violated, then resort to Non-parametric test.
Test of Significant Differences
Two groups
Paired samples (Dependent samples)
Compares two means of one group
Paired samples T-Test
Wilcoxon signed-rank test – if DV is ordinal and if numerical but not normal
Independent samples
Two-sample t-test of significant difference between the means of two independent or unrelated
groups
Ho: There is no significant difference in the test variables between group 1 and group 2.
Tests to be performed:
Independent samples t-test – if continuous that both normal and homogenous
Welch’s test - If continuous and normal but not homogenous
Mann Whitney U-Test – if the DV is ordinal (rank), and if continuous (numeric) but non-
normal
Assumptions:
DV should be measured on a continuous scale
IV should consist of two categorical, independent groups
No relationship between observations
Random sample
DV should be approximately normally distributed for each group of the IV (using
Shapiro-Wilk test)
Homogeneity of variance
Test for outliers
Format:
Table X _________________________
Dependent Grouping Mean Test Statistics p-value Remark
Variable Variables
**Tested at α= 0.05 using Independent T-Test
Workshop Data 1
Table 1. Student performance between their Pre-test and Post-test
Dependent Grouping Mean Test Statistics p-value Remark
Variable Variables
Students Post-Test 42.00 453.00 <0.001* Significant
Pre-Test 30.09
*Tested at α = 0.05 using Wilcoxon signed-rank
Conclusion: There is a significant difference between the scores in pre-test and post-test of the
students in Math. The intervention material is effective.
Assumption Checks
Test of Normality (Shapiro-Wilk)
W p
Post test score - Pretest Score 0.924 0.026
Note. Significant results suggest a deviation from normality.
Workshop Data 2
Table 1. Comparison of the distance to the clinic of the Barangay A and Barangay B
Dependent Grouping Mean Test Statistics p-value Remark
Variable Variables
Distance to Brgy A 6.08 23.39 <0.001* Significant
the clinic (km) Brgy B 2.15
*Tested at α = 0.05 using Welch’s T-test
Conclusion: The households across the two barangays vary in proximity to the clinic.
Assumption Checks
Test of Normality (Shapiro-Wilk)
W p
Distance to Clinic (in km) Barangay A 0.985 0.316
Barangay B 0.990 0.689
Note. Significant results suggest a deviation from normality.
Test of Equality of Variances (Levene's)
F df1 df2 p
24.38
Distance to Clinic (in km) 1 195 < .001
9
Boxplots
Distance to Clinic (in km)
Workshop Data 3
Table 1. Level of Satisfaction between Student and Working Professionals
Dependent Grouping Median Test Statistics p-value Remark
Variable Variables
Level of Student 4.00 759.50 <0.001* Significant
Satisfaction W- 2.00
Professionals
*Tested at α = 0.05 using Mann-Whitney U Test
Conclusion: There is a significant difference between the level of satisfaction of the students and
working professionals.
Assumption Checks
Test of Normality (Shapiro-Wilk)
W p
Level of Satisfaction Student 0.798 < .001
Working Professional 0.813 < .001
Note. Significant results suggest a deviation from normality.
Test of Equality of Variances (Levene's)
F df1 df2 p
Level of Satisfaction 7.859 1 58 0.007
**Median is used because the DV is ordinal.
Two or more groups
No. of groups to compare Parametric Non-parametric
Two independent groups Independent T-Test Mann-Whitney U test
Paired groups Paired T-Test Wilcoxon signed-rank test
Two or more groups ANOVA Kruskal-Wallis
One-way ANOVA
Two or more independent groups (whether there is a significant difference)
Assumptions:
IV should be categorical and independent
DV should be continuous
The presence of outliers
Use the Boxplot
Normality
Distribution plots
Q-Q plots
Variances (homogeneity)
Levene’s Test
Hypotheses and Rejection Rule
Null hypothesis: There is no significant difference among the groups
Alt hypothesis: At least one of the groups is significantly different from the other groups
Post-Hoc Test
Tukey – good if the data met homogeneity and variances and the number of
participants are equal
Conditions Comparison Test Post-Hoc
All assumptions are met ANOVA Tukey/Scheffe
Homogeneity not met Welch’s Test Games-Howell
Normality not met or DV is Krustal-Wallis (Non- Dunn
ordinal parametric)
Format:
Main Variable Grouping Mean F p-value Remark
Variables
Performance A 3.50 12.868 0.002* Significant
Rating B 2.50
C 4.00
*Test at α = 0.05 using Kruskal-Wallis Test
Post-Hoc Test
Group p-value Remarks
A 0.018 Significant
B 0.252 Not significant
C <0.001 Significant
*Test at α = 0.05 using Dunn Test
Kruskal-Wallis Test
Kruskal-Wallis Test
d
Factor Statistic p
f
Compan
12.868 2 0.002
y
Dunn
Dunn's Post Hoc Comparisons - Campany
Compariso
z Wi Wj p pbonf pholm
n
0.03
A-B 2.371 17.050 8.050 0.018 0.053
5
0.25
A-C -1.146 17.050 21.400 0.252 0.755
2
0.00
B-C -3.517 8.050 21.400 < .001 0.001
1
Workshop
Table 1. Comparison among the variety of forest trees in Agusan del Norte Province.
Dependent Grouping Mean F p-value Remark
Variable Variables
Native Trees Banaba 39.556 (c) < .001* Significant
Katmon 143.556 (ab)
Malabulak 104.333 (ab) 6.280
Molave 80.556 (bc)
Narra 176.444 (a)
*Test at α = 0.05 using One Way ANOVA
Groups p-value Remarks
Banaba Katmon 0.011 Significant
Malabulak 0.220 Not significant
Molave 0.655 Not significant
Narra < .001 Significant
Katmon Malabulak 0.691 Not significant
Molave 0.244 Not significant
Narra 0.810 Not significant
Malabulak Molave 0.932 Not significant
Narra 0.138 Not significant
Molave Narra 0.022 Significant
Assumption Checks
Test for Equality of Variances (Levene's)
F df1 df2 p
1.810 4.000 40.000 0.146
Q-Q Plot
ANOVA - No. of Trees in the area
Cases Sum of Squares df Mean Square F p
6.28
Native Tree 102565.556 4 25641.389 < .001
0
Residuals 163318.889 40 4082.972
Note. Type III Sum of Squares
Post Hoc Tests
Standard
Post Hoc Comparisons - Native Tree
Mean Difference SE t ptukey
Banaba Katmon -104.000 30.122 -3.453 0.011
Malabulak -64.778 30.122 -2.151 0.220
Molave -41.000 30.122 -1.361 0.655
Narra -136.889 30.122 -4.545 < .001
Katmon Malabulak 39.222 30.122 1.302 0.691
Molave 63.000 30.122 2.092 0.244
Narra -32.889 30.122 -1.092 0.810
Malabulak Molave 23.778 30.122 0.789 0.932
Narra -72.111 30.122 -2.394 0.138
Molave Narra -95.889 30.122 -3.183 0.022
Note. P-value adjusted for comparing a family of 5
Descriptives
Descriptives - No. of Trees in the area
Native Tree N Mean SD SE Coefficient of variation
Banaba 9 39.556 36.446 12.149 0.921
Katmon 9 143.556 70.497 23.499 0.491
Malabulak 9 104.333 74.821 24.940 0.717
Molave 9 80.556 74.191 24.730 0.921
Narra 9 176.444 54.902 18.301 0.311
Measures of Association
Correlation analysis is a statistical technique that can show whether and how strongly pairs of variables
are related.
Different measures
Pearson’s R
Measures the strength and direction of the linear relationship between two quantitative
variables.
The sample correlation coefficient is denoted by 𝑟.
Do not use the Pearson’s correlation analysis if the relationship is explained better by a
different curve or pattern that is not linear
Unusual data points (called outliers) can mislead the association, especially if the data
set is small.
An observed relationship between two variables does not automatically imply that
there is some cause-and effect relationship between the two variables.
Assumptions:
The two variables considered should be measured at the interval or ratio level.
There is a linear relationship between the two variables (e.g. use scatterplot to
check the linearity)
There should be no significant outliers.
The variables should be approximately normally distributed (Desc. Shapiro-Wilk)
Sample problem:
Table 1. Test of significant association between the number of children in the family and the age of the
mother at her wedding.
Independent Dependent Correlation p-value Relationship Significance
Variable Variable Coefficient*
Mother’s age Number of -0.043 0.913 Very weak and Not significant
children negative
*Legend: a tested using Pearson’s correlation test; [-1.0,-0.8) or (0.8,1.0) very strong relationship; [-0.8, -
weak relationship; [-0.2, 0) or (0,0.2] very weak relationship; * significant at 𝛼=0.05
0.6) or (0.6,0.8] strong relationship; [-0.6, -0.4) or (0.4,0.6] moderate relationship; [-0.4, -0.2) or (0.2,0.4]
Interpretation:
The Correlation table shows that Pearson’s Correlation coefficient is not significant, suggesting that there
exists no correlation between the number of children in the family and the age of the mother at her
wedding (𝑝=0.913).
Note: there is no need to discuss the strength of the relationship because no significant association exists
between the variables.
Point Biserial Point
✓ Measures the strength and direction of the relationship between one continuous
✓The sample correlation coefficient is denoted by 𝑟𝑝𝑏
variable and one dichotomous (without natural ordering) variable
Sample problem using Point Biserial Correlation (one continuous & dichotomous (w/o order or
nominal level only)
Hypotheses:
H0 : There is no significant association between the college entrance exam score and the location of the
high school where the student graduated.
H1 : There is a significant association between the college entrance exam score and the location of the
high school where the student graduated.
Table 2. Test of significant association between the college entrance exam score and the location of
the high school where the student graduated.
Variable 1 Variable 1 Correlation p-value Relationship Significance
Coefficient*
College Location 0.336* 0.017 Very weak Significant
entrance exam relationship
score
*Legend: a tested using Point Biserial Correlation; [-1.0,-0.8) or (0.8,1.0) very strong relationship; [-0.8, -
weak relationship; [-0.2, 0) or (0,0.2] very weak relationship; * significant at 𝛼=0.05
0.6) or (0.6,0.8] strong relationship; [-0.6, -0.4) or (0.4,0.6] moderate relationship; [-0.4, -0.2) or (0.2,0.4]
Interpretation of the results:
The correlation table shows that the Point Biserial Correlation coefficient is significant at =0.05 level,
suggesting that there exists a significant association or relationship between the college entrance exam
score and the location of the high school where the student graduated.
Spearman’s Rank Order Correlation
Measures the strength and direction of the monotonic relationship between two ranked
variables.
The sample correlation coefficient is denoted by 𝑟𝑆
The nonparametric equivalent of Pearson’s correlation
Does not assume that the underlying relationship between X and Y is linear
No assumptions of normality are made regarding the distributions of X and Y
Variables measured in at least ordinal scale
Table 3. Test of significant association between the college entrance exam score and the location of
the high school where the student graduated.
Variable 1 Variable 2 Correlation p-value Relationship Significance
Coefficient*
Job GPA -0.720* 0.011 Strong and Significant
Performance negative
*Legend: a tested using Spearman Rank-Order correlation test;; [-1.0,-0.8) or (0.8,1.0) very strong
0.4, -0.2) or (0.2,0.4] weak relationship; [-0.2, 0) or (0,0.2] very weak relationship; * significant at 𝛼=0.05
relationship; [-0.8, -0.6) or (0.6,0.8] strong relationship; [-0.6, -0.4) or (0.4,0.6] moderate relationship; [-
Interpretation of the results:
The correlation table shows that the Spearman Rank-Order Correlation coefficient is significant at =0.05
level, suggesting that there exists a significant strong and negative association between the college GPA
and the job performance of the employee.
Chi-Square Test
used to determine if a significant relationship exists between two categorical variables
from a single population
Two categorical variables
Two or more categories (groups) for each variables
Independence of observations
No relationship between subjects in each category
The categorical variables are not “paired” in any way (for example, pre-test/post-
test observations)
Data is in the form of observed frequency counts.
Relatively large sample size
𝐻0: There is no significant association between the variables.
Hypotheses:
𝐻1: There is a significant association between the variables.
Table 4. Test of significant association between the preferred learning medium and the sex of the
𝝓a*
respondents.
Variable 1 Variable 1 p-value Relationship Significance
Preferred Sex 0.277* 0.006 Weak Significant
Learning
Medium
*Legend: a tested using Chi-Square Test; [-1.0,-0.8) or (0.8,1.0) very strong relationship; [-0.8, -0.6) or
relationship; [-0.2, 0) or (0,0.2] very weak relationship; * significant at 𝛼=0.05
(0.6,0.8] strong relationship; [-0.6, -0.4) or (0.4,0.6] moderate relationship; [-0.4, -0.2) or (0.2,0.4] weak
Interpretation of the results
The Chi-Squared Tests table shows that the 𝜒2 test is significant at =0.05 level, suggesting that the
(𝜒2(1)=7.695, 𝑝=0.006,with 𝜙=0.277).
opinion about the preferred learning medium is affected by whether the student is male or female
