Notes on the Kinds of Quantitative Research

Experimental Quantitative Research

- Involves the collection of data before and after the administration of a treatment to the research
sample (S3)
- Establishes cause-and-effect relationship or causality (S3)
- Like quasi-experimental is used to establish the cause-and-effect relationship of two or more
variables, this design provides a more conclusive result because it uses random assignment of
subjects and experimental manipulations. For example, a comparison of the effects of various
blended learning to the reading comprehension of elementary pupils (S1)
- This research utilizes scientific method to test cause-and-effect relationships under conditions
controlled by the researcher. In this case an effort is made to determine and impose control over all
other variables except one. An independent variable is manipulated to determine the effects on the
dependent variables. For instance, a teacher would like to know if a new teaching strategy is
effective or not, so he/she teaches one section using the new strategy and teachers another
comparable section without the new strategy, then an achievement test was given to the two
sections. The manipulated independent variable is the new teaching strategy which is being tested
if it influences the dependent variable which is the achievement of the students. Notice that the
sections are comparable with one another meaning all other variables are controlled by the
teacher. (S2)

Non-Experimental Quantitative Research

- data are collected only once (S3)
- establishes associations between variables (S3)
- (descriptive, correlational, ex post facto/causal-comparative, survey)

Descriptive Design
- This design is used to describe a particular phenomenon by observing at it occurs in nature. There
is no experimental manipulation, and the research does not start with a hypothesis. The goal of
descriptive research is only to describe the person or object of the study. An example of
descriptive research design is “the determination of the different kinds of physical activities and
how often high school students do it during the quarantine period.” (S1)
- This design is concerned with describing the nature, characteristics, and components of the
population or a phenomenon. There is not manipulation of variables or search for cause and effect
related to the phenomenon. This design attempts to find general attributes of the presently existing
situation and determine the frequency with which it occurs. Descriptive research is used if, for
example, you want to know how many hours senior high school students spend in social media,
the number of malnourished students who failed the achievement test, and how healthy is the food
served during recess in the public schools. (S2)

Correlational Design
- Identifies the relationship between variables. Data is collected by observation since it does not
consider the cause and effect for example, the relationship between the amount of physical activity
done and student academic achievement (S1)
- It is the systematic investigation of the nature of relationships, or associations between and among
variables without necessarily investigating into causal reasons underlying them. It is also
concerned with the extent of relationships that exists between or among the variables. For
example, if pre-board examination results can be used to predict performance in the Licensure
Examination for Teachers (LET), then the higher the pre-board grade, the higher most likely be the
score in the LET. Correlational research is employed if you like to know, for example, if the

1
 following factors are related to each other: sex and mathematical ability, marriage and cancer
recovery, occupation and life span (S2)

Causal-Comparative design or ex post facto (after the fact) design

- Is used to investigate a possible relationship between previous events and present conditions. The
term ‘ex post facto’ means after the fact, looks at the possible causes of an already occurring
phenomenon. Just like the first two, there is no experimental manipulation in this design. An
example of this is "how does the parent’s academic achievement affect the children obesity?" (S1)
- It is also known as ex post facto (after the fact) research. This kind of research derives conclusion
from observations and manifestations that already occurred in in the past and now compared to
some dependent variables. It discusses why and how a phenomenon occurs. For example, a
researcher is interested in how weight influences stress-coping level of adults. Here, the subjects
would be separated into different groups (underweight, normal weight, overweight) and their stress-
coping levels measured. This is an ex post facto design because a pre-existing characteristic
(weight) was used to form the groups (S2)

Quasi-experimental Design
- Is used to establish the cause-and-effect relationship of variables. Although it resembles the
experimental design, the quasi-experimental has lesser validity due to the absence of random
selection and assignment of subjects. Here, the independent variable is identified but not
manipulated. The researcher does not modify pre-existing groups of subjects. The group exposed
to treatment (experimental) is compared to the group unexposed to treatment (control): example,
the effects of unemployment on attitude towards following safety protocol in ECQ declared areas
(S1)

Evaluation Research
- This kind of research aims to assess the effects, impacts or outcomes of practices, policies, or
programs. Assessing the implementation of nursing care in a hospital and determining the impact
of a new treatment procedure for patients are examples of evaluation research.

Survey Research
- Survey research is used to gather information from groups of people by selecting and studying
samples chosen from a population. It may be done in various ways like face-to-face, phone, mail,
and online. Survey research may be cross-sectional if the information is collected from a sample in
just single point, its time just like the child-rearing practices of single parents, and population
control practices of unmarried couples. Survey research is considered longitudinal if the researcher
collects information on the same subjects over a period sometimes lasting many years in order to
study the changes through the years. Longitudinal survey is utilized, for example, to determine the
growth of rice yield in the country and the rate of promotion of doctorate degree holders five years
after earning the degree (S2)

Source 3
Science in Today's World-Research in Daily Life 2 (SIBS Publication, 2018)
Page 58

Quantitative Research Designs

2
How are appropriate research designs for a problem chosen?

Research design refers to the detailed outline of the overall strategy that integrates in a coherent
and logical manner the various components of a scientific study. An appropriate research design ensures
the effective conduct of an investigation to address the research problem or objective. The strategy
involves selecting research subjects and gathering, measuring, and analyzing data.

The research objectives or statements of the problem must be the primary considerations in
choosing an appropriate research design. A problem seeking for a cause-and-effect relationship, or
causality, requires an experimental research design. A cause-and-effect relationship means that a
researcher-initiated change in one variable creates an automatic change on another variable. This kind of
relationship can be established or confirmed only by performing an actual experiment. Meanwhile, studies
seeking for the description or prediction of a variable, as well as the degree of relationship between two or
more variables, can be addressed by nonexperimental research designs, such as correlation and
regression. Note, however, that a considerable association between two variables does not necessarily
imply a causal relation.

A study that calls for a simple association yet utilizes an experimental research design is excessive
and likely to be a waste of effort and resources. This is because performing an experimental study is too
rigid for a fairly simple task, which can make use of simple correlational statistics, like the Pearson product-
moment correlation.

Experimental Methods

What are the unique characteristics of the experimental research method?

The distinguishing characteristics of experimental studies include direct control of an intervention,
regulation of extraneous variables, and limitation of the systematic difference between or among study
groups. In this method, the researcher manipulates the independent variable and studies its effects on the
dependent variable. The researcher also regulates the extraneous variables in the study to limit their impact
on the dependent variable. If extraneous variables are not managed well, the researcher cannot attribute
the changes in the dependent variable solely to the manipulation of or changes in the independent variable.

3
Therefore, it is important to identify and control or maintain the extraneous variables constant throughout
the conduct of the experiment to elicit valid and reliable results. Any variation on the extraneous variables
at any time within the research process and any incomparability of the study groups implies failure of the
whole investigation procedure.

1. Direct control of the intervention or treatment (independent variable). Direct control of the
independent variable means that the researcher manipulates the independent variable by deciding
on the time, amount, or manner of administration of the intervention or treatment given to one or
more groups of research participants.

2. Regulation of extraneous and confounding variables. In an experiment, all the conditions

deemed to affect the dependent variables, as suggested by the literature review, are kept the same
in all study groups. The only factor that should be different in these groups is the presence or
degree of intervention or treatment. The successful control of extraneous and confounding
variables is necessary to establish causality or a causal relationship between the independent and
the dependent variables. As such, it is viewed that the only research method that can establish a
causal relationship between variables is the experimental method.

3. Elimination of systematic differences among the elements in the study groups. It is critical to
maintain comparability or parallelism of the elements in the study groups. For instance, variables
like weight, age, and lineage of test animals in the comparable groups must be almost, if not
completely, the same. Similarly, the physical composition, the chemical contents, and the moisture
of the soil must also be the same when experimenting on the growth of plants. Meanwhile, in social
and in educational sciences, characteristics like gender, achievement, attitudes, and economic
background of the participants may be maintained in comparable groups. This "no systematic
differences" in the study groups can be achieved if participants or subjects are randomly assigned
in forming the groups. If randomization cannot be done, like in the case of block class sections, in
which students cannot simply be moved from one group to another, the matching of participants or
subjects in the study groups can be achieved using some of the aforementioned variables to
ensure parallelism. The matching variables must be proven to affect the dependent variables
based on the literature review.

Threats to the Validity of an Experimental Research

The validity of an experimental study is more difficult to establish compared to that of any
nonexperimental research. An experiment is said to be valid if the threats to its internal validity and external
validity have been addressed. Internal validity is attained if the observed changes in the dependent
variables are direct effects of the changes done to the independent variables. In contrast, external validity is
achieved if the results of an experiment can be generalized or applied to any similar conditions outside the
experimental setting. External validity is easier to establish in scientific investigations than in social science
researches. The different threats to the internal validity of science research are as follows:

1. Selection. This refers to the selection of subjects in the experimental and the control groups. The
subjects in the study groups must have comparable characteristics so that the groups can be
considered parallel or comparable. The assignment of subjects to study groups must be done with
caution. If the research subjects are animals, these characteristics may include age, sex, and
lineage, or pedigree. The source, or breeder, of these subjects must also be considered. If the
required number of research subjects is too many that animals from one breeder are not enough,
animals from a specific breeder must be equally and randomly distributed to all research groups. If

4
 the research subjects are insects, it is best to breed them so that their physical characteristics can
be controlled.

2. History. This refers to some events that occur in between measurements of the changes that are
observed in the dependent variable and that are not part of the experiment and may affect the
dependent variable. An example of these are diseases acquired by the research subjects during
the experiment. For instance, if the experiment is about the effectiveness of an insecticide, insect
subjects with diseases may cause unfavorable effects on the result of the experiment. One way to
address this is to keep the duration of the experiment short enough to generate results. Time
frames used in similar studies can be used as reference.

3. Instrumentation. This refers to the accuracy and functionality of the mechanical device in
measuring the change in the dependent variable. The device must be the standard instrument
used for a particular purpose, and it must be in good and functional condition. If observation is
used to detect the changes, there must be an observation guide, which should be valid for the
purpose of the study, and the observers must be briefed and trained for consistency of observation.

4. Maturation. This refers to any biological processes or changes that can interfere with the effect of
the independent variable on the dependent variable. In between measurements of the changes in
the dependent variable, the subjects may have become more mature or perhaps have developed
resistance against the treatment. In such cases, the effect of the independent variable on the
dependent variable may be hindered. Therefore, to limit the impact of this threat on the validity of
the experiment, it is recommended that the experimental duration should not be too long. Similarly,
studies involving the same threat may be used as reference.

5. Mortality. This refers to the actual death of the research subjects during the course of the
experiment. This mortality may be caused by treatment overdose or by disease acquired even
before the conduct of the experiment. A large percentage of subject mortality will affect the
consistency of the research results. It is suggested to first determine the lethal dose of the
treatment before the start of experimentation. The lethal dose is the amount of treatment
preparation that can cause death to a test subject. The minimum lethal dose (LDos) means that 5%
of the test subjects will die if the subjects are given the preparation. Another way to address this
threat is through acclimatization, a process in which the test organisms are allowed to adjust to
their new environment to minimize stress.

Quick Quiz 4-1

1. In an experiment on the effect of exposure to cigarette smoke on the ability of mice to finish a
maze, what possible threat(s) to validity can affect the results? Justify your answer.
2. What measure(s) can possibly limit the effect of the identified threat(s) to validity in item 12
Elaborate your answer.

Research investigations that employ experimentation can use a specific type of experimental research
design. Experts categorize experimental designs into pre-experimental designs, true experimental designs,
and quasi-experimental designs. In pure sciences like biology, chemistry, and physics, true experimental
designs are used. Meanwhile, social sciences mainly utilize quasi-experimental designs.

Experimental designs are often utilized in the fields of science, technology, engineering, and mathematics.
The following symbols are commonly used in such experimental designs:

5
Pre-Experimental Designs

Pre-experimental designs often implement inadequate control of extraneous variables, making these
designs inferior to other experimental designs. Because of poor internal validity, these cannot adequately
establish cause-and-effect relationships. Studies using pre-experimental designs are done as preliminary
research before the actual experimentation in order to refine procedures and address unperceived
problems during the conceptualization of the study.

The following are some common pre-experimental designs:

1. One-group post-observation-only design. Researchers who use the one-group post- observation-only
design identify a group of research subjects or participants, administer the treatment or intervention, and
make post-observation or measurement of the dependent variable. In symbols, this design can be
illustrated as:

This design is the weakest because it provides no comparison among pre-experimental designs
because of the absence of a comparison or control group. Simply, there is no pre- observation data to
compare with the post-observation data. Pre-observation involves the measurement of the dependent
variable before the treatment is given. Because of this, it cannot be established that changes in the
dependent variable are caused by the treatment or intervention.

2. One-group pre-observation and post-observation design. Primarily, the difference of the one-group
pre-observation and post-observation design from the one-group post- observation-only design is the
presence of pre-observation. Once a group of research subjects or participants is identified, the pre-
observation of the dependent variable is done, the treatment or intervention is administered, and lastly the
post-observation is conducted. This design is illustrated below.

In this design, only one comparison can be done. The pre-observation data are compared to the
post-observation data to identify possible changes caused by the treatment given. If the extraneous
variables are completely controlled, the change in the dependent variable, as represented by the difference
of the pre- and post-observation measurements, can be attributed to the treatment given. However, if the
extraneous variables are not successfully controlled, then the changes in the pre- and post-observation
data cannot be entirely attributed to the given treatment.

6
3. Nonequivalent-groups post-observation-only design. This design has an experimental group
that can be compared with the control group. One group of research subjects or participants is given the
treatment-the experimental group (A)-and another group acts as the comparison group that is not given any
treatment-the control group (B). This design is illustrated below,

The distinguishing feature of this design is the nonequivalent groups. The participants in the two
groups may vary in characteristics, which may differentially affect the dependent variable. For instance, in
animal research subjects, differences in health status, in pedigree, in environment of origin, and in other
similar characteristics may affect their reaction to any experimental treatment. In the case of human
research participants, differences in academic background, in knowledge and skills, in interest, in
personality, and in other similar characteristics may also affect the research results. If these variations in
characteristics exist initially in the two research groups and no action is done to control them, it is difficult to
attribute the difference in the post-observation data to the treatment or intervention given in the
experimental group.

True Experimental Designs

The distinguishing feature of the true, or pure, experimental designs from the other designs is the
random selection and assignment of participants to the research groups. Theoretically, randomization
ensures parallelism, or equivalence, of the research groups. Parallelism is necessary to establish a cause-
and-effect relationship, or causality, between independent and dependent variables. This process is
deemed useful in controlling the effect of extraneous variables on the dependent variable, increasing the
validity of the experiment.

There are three types that fall under true experimental design: the randomized pre-and-post-
observation control-group design, the randomized pre-and-post-observation comparison-group design, and
the Solomon four-group design. However, the Solomon four-group design is beyond the scope of this book
because it involves complex analysis using four experimental groups.

1. Randomized pre-and-post-observation control-group design. In this design, participants are

randomly selected and assigned to two research groups: the experimental group and the control
group. The pre-observation is conducted in both groups. After the pre- observation, the treatment is
administered in the experimental group, while nothing other than food and water, in the case
involving animal subjects, is given to the control group for the duration of the experiment. In
symbols,

This type of true experimental design has sufficient features to establish causality. The
random selection and assignment of the research subjects to the two groups assures that they are
equivalent. This means that extraneous variables are controlled or that the effects on the
dependent variable are minimized, if not totally removed. This process is enhanced by determining
and comparing the pre-observation data of the dependent variable using a statistical procedure like

7
 the t-test for independent samples. If these data provide no significant statistical difference, then it
is safe to conclude that the two groups are initially equivalent. If there exists a significant difference
in the pre-observation measurements, then the characteristics of the subjects in the two groups
must be reevaluated to attain the desired group equivalence.
Another essential comparison that can be done using this design is the post-
observation data. Once the pre-observation measures are established to be the same, the
comparison of the post-observation data will test the causality between the independent
and the dependent variables. That is, if significant differences exist in the post-observation
measurements, then changes in the dependent variables can be attributed to the treatment
given. If the difference between the pretest and the posttest observations is statistically
significant, it may be concluded that the intervention made a difference or is effective. To
ascertain this, mean gains from pre-observation to post-observation between groups are
tested for a statistical difference. However, if the statistical comparison suggests no
significant difference, then it cannot be concluded that the independent variable has a
causal relationship with the dependent variable.

2. Randomized pre-and-post-observation comparison-group design. In this design, participants

are randomly assigned to the experimental and the comparison groups. The experimental group
receives a newly developed or formulated treatment or intervention, while the comparison group
receives an established effective treatment or intervention. For example, in a study about the
development of an insecticide from a plant extract, the established effective treatment may be a
commercially available insecticide. In this experiment, researchers may administer to the subjects
in the comparison group the solvent used to extract the active component of the plant. This is to
ensure that insect mortality is due to the active components of the plant and not due to the solvent
used. A pure experimental design is represented by the following diagram.

The comparative analyses that can be drawn out from this design are almost the same as
those from the previous design. The equivalence of the research groups after the random
assignment of the subjects in terms of the dependent variable can be verified by comparing the
means of the pre-observation data. If there is no significant difference in the said means or
averages, then the two groups are parallel in terms of the dependent variable. Otherwise, the two
groups must be made equivalent first before the treatments can be administered.

This design is often employed to know the effectiveness of the developed treatment or
intervention compared to a proven effective treatment of intervention. If the post- observation mean
data in both groups are statistically the same, perhaps after some replications, then the two
treatments are equally effective to almost the same degree. If otherwise, then it can be said that
one treatment is more effective than the other.

In medical or pharmaceutical research, this design is used to know if a newly developed

medicine is as effective as a commercially available one. Usually, a third group of participants is
formed to be another comparable group. This third group is given a placebo, which is a substance
or treatment with no active ingredients meant to have health effects. This substance invokes
psychological effects similar to taking an actual effective substance or treatment. If the disease is
cured in the placebo group, it may mean that changes in the health status of the participants
cannot be entirely attributed to the newly developed and commercially available treatments.

8
 Quasi-Experimental Designs

Experiments that do not employ random assignment fall under quasi-experimental

designs. In social sciences, these experiments include intact classes, or block sections, like those
usually found in secondary schools.

Quasi-experimental designs closely approximate the rigors of experimental designs. They

employ control or comparable groups with pre- and post-observation of the dependent variables.
There are many quasi-experimental designs, but basic ones, particularly the nonequivalent pre
and-post-observation control-group and the nonequivalent pre-and-post-observation comparable
group designs, will be discussed in this book. These two are the most common designs used in
scientific and social science investigations.

1. Nonequivalent pre-and-post-observation control-group design. This is similar to the

first true experimental design previously discussed-the randomized pre-and-post-
observation control-group design-except for the randomization process. It is commonly
used in educational research but very seldom applied in scientific investigations. In this
setup, the selection of students to form a group is not done as doing so will disrupt the
students class schedules. In such cases, two block sections, or intact student groups, are
matched for the purpose of the experiment. The case is the same when performing an
investigation involving two communities or families. The researcher cannot uproot a family
to be moved to another community just for the purpose of equating the families in the two
communities. In symbols, this design is described as follows.

The main issue in using this experimental design is the comparability of the
research groups as randomization is not implemented. To address this concern, the
participants in the two groups must initially be compared in terms of a few variables found
to affect the dependent variable based on the review of related literature. If there is no
significant difference in the mean scores of the said variables between the groups, then it
can be safely argued that the two groups are equivalent in terms of these variables, and
accordingly, the experiment can proceed. If there is a significant difference, then the
participants can be matched in terms of these variables. Using the matching technique, the
researcher identifies in both groups’ pairs of participants with almost the same
characteristics in terms of the matching variable. The data of these pairs will be part of the
research data and those of participants without a match will not be considered in the data
analysis. After the matched pairs have been established, the experiment can proceed.
The analysis for the effectiveness of the treatment or intervention in this design is
the same as the process in using the randomized pre-and-post-observation control-group
design.

2. Nonequivalent pre-and-post-observation comparable-group design. This design is

almost similar to the nonequivalent pre-and-post-observation control-group design, except
that instead of a control group, the other group is given another kind of treatment. For
instance, in an education or a social science research, a treatment or intervention to be
given to participants must be as effective as the one to be tested. For example, if the

9
 researcher wants to test the effectiveness of a newly developed module in science, the
comparable group must be given a module that is proven to be effective for the purpose
similar to that of the previous module. The diagram of this research design is as follows.

Quick Search
Search for an experimental study and identify a weakness or limitation in the research design used.
Additionally, discuss a measure or another research design that can overcome the identified weakness or
limitation.

Nonexperimental Methods

Unlike experimental research methods, nonexperimental quantitative research methods are

seldom used in pure scientific investigations. Often, these designs are employed in social science research,
particularly in the fields of education and psychology. They are characterized by non-manipulation of
variables to establish causal relationships. This method is used to describe the profile of research
participants, compare groups of human respondents, and correlate characteristics of research participants.

The nature of nonexperimental quantitative research is descriptive. This research method

highlights the description of the general condition of a group of research participants at the time the study is
being conducted. The description of such general condition is intended to explore the causes of an
identified phenomenon and to test hypotheses pertaining to the causes. Statistical measures such as
frequency, percentage, mean, or average, standard deviation, and variability of the participants' quantitative
data are used for the description process.

Nonexperimental quantitative research is characterized by three types of research designs: causal-

comparative, or ex post facto, correlational, and survey designs. The causal-comparative design explores
the causal relationship between two variables without having direct control of the intervention as it occurs
already, hence an ex post facto investigation of the effects of a preexisting phenomenon. Meanwhile, the
correlational design investigates relationships in general. It must be noted that unlike an experimental
design, the causal-comparative and the correlational designs do not establish causal relationships. They
merely explore possible causal relationships. As such, studies employing causal-comparative or
correlational designs are conducted before doing an experimental research. Experiments are not done to
determine the nature of the causal relationship between two variables if previous causal-comparative or
correlational studies have not detected any relationship between the variables. Not all relationships suggest
causality.

Causal-Comparative or Ex Post Facto Design

10
 The causal-comparative design is a natural experiment because it compares two research groups
that receive naturally- occurring treatments or interventions. The term ex post facto in Latin means "after
the fact." What makes this design different from the experimental design is the fact that the researcher does
not manipulate the independent variable to effect changes in the dependent variable.

Ex post facto, or causal comparative, research is a design used to explore the "cause for, or
consequences of, existing differences in groups of individuals" (Fraenkel and Wallen, 2009).

In this design, all the researcher has to do is investigate the manifested effects to prove or disprove
a hypothesis. For example, in testing the hypothesis that smoking results in lung cancer in humans,
experimental research cannot be implemented because forcing healthy individuals to smoke and finding out
if they develop lung cancer is unethical and unacceptable. Therefore, to carry out the testing of the
hypothesis, the causal-comparative design is implemented. creating two groups composed of smoking and
nonsmoking participants, and their lung conditions are monitored for a specific period. If some smokers
develop lung cancer, while nonsmokers remain healthy after a given time, then there is enough reason for
the researcher to conclude that smoking causes lung cancer.

The major limitation of the causal-comparative design is the lack of control over the independent,
the extraneous, or the confounding variables. This is because it is difficult to establish similar
characteristics of human participants in the two comparable groups. To avoid such limitation, random
selection, matching, or some statistical equating procedure such as analysis of covariance (ANCOVA)
could be done. This statistical equating procedure is beyond the scope of this material.

Steps in Causal-Comparative Design*

1. Formulate the problem.

- Identify and define the particular phenomena of interest.
- Consider possible causes for or consequences of these phenomena.
2. Select the sample.
3. Identify the instruments to be used.
4. Collect data.
5. Compare results.
*Adapted from Fraenkel and Wallen (2009).

MISCONCEPTION ALERT!
It was thought that causal comparative design could establish a greater extent of causality between
two variables. However, according to Burke (2001), the degree of causality established between the
independent and dependent variables by the causal-comparative and the correlational designs is just the
same. The only difference between the two designs is that in the causal-comparative design, the
independent variable is categorical data, while in the correlational design, the independent variable can be
interval or ratio data.

Correlational Design

There are two types of correlational research design: the exploratory design and the prediction
design. Unlike the casual-comparative design, these two do not establish cause-and-effect relationships
between two variables. However, these two designs differ in purpose and statistical procedures used.

11
 1. Exploratory correlational design. The exploratory correlational design explores the degree of
relationship between two or more variables measured in the interval or ratio scale. This
relationship suggests a present association between variables and not past or future
relationships. This is because the data analyzed are collected from research participants at
one time just before they are studied. For instance, if the relationship or association between
students' mathematical abilities and their achievements in Physics is to be tested, a
mathematical ability test and a physics achievement test are given to students within a two to
three-day interval. The results of these assessments will then be correlated using a statistical
procedure that measures association or relationship. In this design, the participants are studied
as a single group and not classified into categories.

2. Prediction correlational design. The prediction design is employed to predict a specific

outcome through another variable. This design typically involves two types of variables- the
predictor variable, which is the variable used to predict, and the criterion variable, which is the
variable being predicted. In a prediction study, the criterion variable is measured within the
duration of the study, while the predictor variable has already been assessed before. For
example, prediction correlation design is used in a study involving the high school admission
test scores and students' academic achievement in terms of grade point average (GPA).
Students' GPA is computed after tenth grade, while their admission test scores have been
assessed before their entrance to the same high school. After collecting these data, a
regression analysis a statistical procedure used for prediction-must be implemented to yield
the prediction equation. For a study where there is only one criterion variable and one predictor
variable, simple regression analysis can be done. Meanwhile, for a research with one criterion
variable and two or more predictor variables-for example, predicting achievement in a subject
using students' attitude toward the subject and the teacher's teaching competence-multiple
regression analysis can be used.

Survey Research Design

The term survey is frequently misconstrued as a mere procedure in gathering data, while in fact, it
is a "legitimate approach to research that has clear benefits in helping to describe and explore variables
and constructs of interest" (Ponto, 2015). Often, this design is frequently used in social and psychological
research to describe and explore human behavior (Singleton & Straits, 2009). Survey research is a
quantitative research design wherein the researcher collects information from a sample or from a
population (census) to describe the characteristics of the population. Characteristics described in a survey
include opinions about, perceptions of, attitudes and behaviors toward, and even judgments about a
particular product. In survey research, the researcher collects quantitative data using a questionnaire or by
conducting interviews.

There are two general types of survey design: the cross-sectional and the longitudinal survey,

12
 Cross-sectional survey, or point-in-time survey, refers to the collection of information "from a
sample that has been drawn from a predetermined population" (Fraenkel & Wallen, 2009). The collection is
only done in a single time frame. Cross-sectional design is useful in examining current attitudes, beliefs,
opinions, or practices; in facilitating the comparison of two or more groups of respondents; in identifying the
needs of the community; in evaluating program effectiveness; and in conducting in large-scale assessment
such as a national survey (Creswell, 2012). An example of a cross-sectional survey is the study
"Employment Profile of Women with Disabilities in San Remigio and Mandaue City, Cebu" (Mina, 2017).
The study surveyed at one point in time a sample of PWD women from San Remigio and Mandaue City in
the Province of Cebu, and the rate and the quality of their employment were recorded. It should be noted
that when the entire population is surveyed, the procedure is referred to as census.

Meanwhile, longitudinal survey refers to the collection of information over a period and not only
once. The purpose of this survey is to study the change over a specified period. There are three types of
longitudinal survey:

1. Trend study. This is done at different points in time on different samples from a population
whose members may change (Fraenkel & Wallen, 2009). For example, in the study entitled
Poverty and Agriculture in the Philippines: Trends in Income Poverty and Distribution (Reyes,
Tabuga, Asis, & Datu, 2012), poverty profile and poverty condition involving the agriculture
sector from 1985 to 2009 were recorded, and the trend was analyzed. In the study, the
population of interest was consistent; however, the survey samples selected for the study from
1985 to 2009 were not precisely the same individuals.

2. Cohort study. The term cohort is defined as a group of individuals or subjects sharing similar
characteristics or experiences. This group is the subject of a cohort study. A cohort study is
defined by Fraenkel and Wallen (2009) "as a design in survey research in which a particular
population is studied over time by taking different random samples at various points in time.
The population conceptually remains to be the same, but the individuals change." For example,
in a study entitled "College and Career Readiness of US High School Graduates" ("The
College and Career." 2016), a cohort of high school graduates from the 50 states in the US
was studied from 2014 to 2015. The Scholastic Assessment Test (SAT) or American College
Testing (ACT) scores of high school graduates of class 2014 were recorded. Subsequently, the
SAT and ACT scores high school graduates of class 2015 in the same states were also
recorded. The SAT/ACT scores from both periods, 2014 and 2015, were then compared.

3. Panel study. This type of longitudinal design studies over a period the same sample of
respondents from the same population. It is considered the most difficult to carry out due to
threats of mortality. Nonetheless, panel study is the best survey design to truly understand
changes over a specified time as it repeatedly tracks exactly the same respondents. For
example, in the study entitled "A Panel Study on the Effects of Social Media Use and Internet
Connectedness on Academic Performance and Social Support" (Leung, 2015), a group of 718
students participated for two years. On the first year, their social media use, their internet

13
 connectedness, and their academic performance were recorded. The same data were
gathered on the second year from the same sample of students. The results were then
compared, and the relationship between these variables was determined.

RESEARCH IN HISTORY

A Brief History of Survey Research

Survey research is considered a "young field relative to many scientific domains" (Groves, 2001).
The term survey was derived from Anglo-French word surveer, which means "to look over" (Merriam-
Webster Dictionary). Moreover, Kerlinger (1973) defined survey as social scientific research that focuses
on people's beliefs, opinions, attitudes, motivations, and behavior. Survey research was invented by
lazarsfeld, Gallup, and Cantril sometime in the 1930s (Converse, 2017).

There were three significant phases in the development of survey research (Mathiyazhagan and
Nandan, 2010):
Phase 1 (1890-1930s). The mere collection of data was considered a survey. The focus of survey
research during this period was business and politics.
Phase 2 (1935-1945). Descriptive studies based on survey data were known as survey methods in
the second phase, which continued from 1930 to World War II.
Phase 3 (1945 and beyond). The explanation part of the data was considered more important in
the survey methods. It was also in 1968 that a panel study, a type of survey research, was first conducted.
It was entitled The Panel Study of Income Dynamics.

Modern survey research, propelled by scientific leaps for the past 200 years, now employs the use
of random selection, statistical control (PRA, N.D.), data comparison between groups, predicting trends,
and extracting significant effects of interventions on both cohort and panel studies (Fraenkel and Wallen,
20121.

References:
Converse, J.M. (2017). Survey Research in the United States: Roots and Emergence (1890-1960). New
York: Routledge. Fraenkel, J. R., & Wallen, N. E. (2009). How to Design and Evaluate Research in
Education, New York: McGraw-Hill Education

14