Lesson 2 – Kinds of Quantitative Research

Quantitative Research Designs

Refers to the overall strategy chosen by researchers to integrate all components of the study in a coherent, logical way.
Emphasizes objective measurements and the statistical, mathematical, or numerical analysis of data collected through polls, questionnaires, surveys, or manipulation of pre-existing statistical data using computational techniques.
Two grand categories:
- Experimental
- Non-Experimental

Non-Experimental Research Design

Researcher observes phenomena as they occur naturally; no external variables are introduced or controlled.
Data are collected without changing conditions or introducing treatments.
Goal: describe the person, object, or situation being studied – not establish cause–effect.

Types of Non-Experimental (Descriptive) Designs

1. Survey

Provides a quantitative or numeric description of trends, attitudes, or opinions of a population by studying a sample.
Example: “The determination of the different kinds of physical activities and how often high-school students do them during quarantine.”

2. Correlational

Identifies relationships between variables without asserting causality.
Two directions:
- Positive correlation – variables move in the same direction.
- Negative correlation – variables move in opposite directions.
Example: “Relationship between the amount of physical activity and student academic achievement.”

3. Ex-Post Facto

Latin for “after the established fact.”
Investigates possible causes of an already-occurring phenomenon, looking backward in time.
Example: “How do parents’ academic achievements affect children’s obesity?”

Experimental Research Design

Planned set of procedures to investigate relationships between variables and establish cause–effect.
Utilizes two sets of variables:
- Control group (constant; no treatment)
- Experimental group (receives treatment)
Includes:
- A hypothesis
- A manipulable independent variable
- Measurable dependent variables

When to Conduct Experimental Research

Time is vital in establishing cause–effect.
Invariable behavior expected between cause and effect.
Desire to understand the importance of the causal relationship.

Three Primary Experimental Frameworks

Pre-Experimental
True Experimental
Quasi-Experimental

1. Pre-Experimental Research Design

Simplest form; uses either a single group or multiple groups observed after a treatment.
Lacks random assignment and often lacks a separate control group.

Advantages

Cost-effective; simple; natural-environment friendly; suitable for beginners; minimal human intervention.

Disadvantages

Weak for causal inference; limited control; high threat to internal validity; integrity of results is hard to judge; absence of a true control group.

Specific Designs

1.1 One-Shot Case Study

Single group studied once after treatment.
No comparison/control.
Example: Measure a class at term’s end after a new instructional method.

1.2 One-Group Pretest–Posttest

Single group measured before and after treatment.
Assumes any change is due to the intervention.
Example: Pretest → teach with new method → Posttest.

1.3 Static Group Comparison

One treated group compared to one untreated group, both post-tested only.
Pre-existing differences are unknown.
Example: One class with new method vs. another class without, both given end-of-term posttests.

2. True Experimental Research Design

At least one independent variable manipulated, participants randomly assigned, and a dependent/outcome variable measured.
Considered the most accurate for hypothesis testing via statistical analysis.

Core Characteristics

Random selection & assignment of participants.
Presence of control and experimental groups.
Researcher controls/manipulates the independent variable.

Types of True Experimental Designs

2.1 Pretest–Posttest Control-Group

Random assignment → both groups pre-tested.
Experimental group receives treatment; control does not.
Both groups post-tested; differences attributed to treatment.
Matrix:
- Experimental: Pretest / Treatment / Posttest
- Control: Pretest / $\times$ (no treatment) / Posttest
Example: Role-playing as a teaching strategy and its effect on reading comprehension.

2.2 Posttest-Only Control-Group

Random assignment; no pretest.
Treatment applied only to experimental group; both post-tested.
Matrix:
- Experimental: $\times$ (no pretest) / Treatment / Posttest
- Control: $\times$ / $\times$ / Posttest
Example: Technology integration and student academic performance.

2.3 Solomon Four-Group

Combines the above two designs; participants randomly assigned to four groups:
- Experimental A: Pretest + Treatment + Posttest
- Control A: Pretest + No treatment + Posttest
- Experimental B: No pretest + Treatment + Posttest
- Control B: No pretest + No treatment + Posttest
Used to verify treatment effectiveness while controlling for pretesting effects.
Example: Intensive review program and students’ mathematics grades.

3. Quasi-Experimental Research Design

“Quasi” = resembles experimental but lacks random assignment.
Independent variable manipulated, but participants are not fully randomized.
Common in natural/field settings where random assignment is infeasible (e.g., schools, clinics).

Key Characteristics

Manipulation of independent variable.
No random selection, no random assignment, and/or limited control.

Common Quasi-Experimental Designs

3.1 Nonequivalent Groups Design

Two (or more) existing groups that appear similar; only one experiences treatment.
Lack of random assignment makes pre-existing differences a confound.
Examples:
- New fraction-teaching method in one third-grade class vs. another class.
- After-school program at one school vs. none at a similar school.

3.2 Pretest–Posttest Design

Dependent variable measured before and after treatment within a single group.
Mirrors within-subjects structure; still vulnerable to history/maturation.
Example: Elementary students’ attitudes toward drugs → antidrug program → re-measure attitudes.

3.3 Interrupted Time-Series Design

Multiple measurements before and after treatment over time.
Detects trends and immediate/long-term effects.
Example: Weekly productivity tracked for a year; midway, shift length reduced from $10$ hrs to $8$ hrs.

3.4 Combined (Hybrid) Designs

Merges nonequivalent-groups with pretest–posttest.
Treatment group: Pretest → Treatment → Posttest.
Control group: Pretest → No treatment → Posttest.
Causal inference strengthened by group comparison + temporal comparison.
Example: School A gets antidrug program; School B does not. Both pre- and post-tested.

Cross-Cutting Themes & Significance

Objective measurement and statistical analysis underpin all quantitative designs.
Control vs. naturalism trade-off: Greater control (true experiments) improves internal validity but may reduce ecological validity; non-experimental and quasi designs increase real-world relevance but sacrifice some causal certainty.
Ethical considerations:
- Random assignment may withhold potentially beneficial treatment; researchers must weigh risks/benefits.
- Manipulation must respect participant autonomy and consent.
Practical implications:
- Policy makers rely on experimental/quasi-experimental evidence to judge program effectiveness.
- Educators use survey and correlational data to tailor instruction.
Philosophical reflection: The concluding quote – “The best way to predict the future is to design it today” – underscores the proactive nature of experimental design: by systematically manipulating variables now, researchers can shape and foresee future outcomes.

Quick Reference Formulae & Symbols

Correlation strength often expressed with Pearson’s $r$ , ranging $-1 \le r \le 1$ .
Experimental effect size frequently reported as Cohen’s $d$ : $d = \dfrac{\bar{X}<em>{\text{exp}} - \bar{X}</em>{\text{ctrl}}}{s_{\text{pooled}}}$ .
Threats to internal validity (history, maturation, testing, instrumentation, selection, mortality) are particularly relevant to pre-experimental and quasi designs.

Study Tips

Map each research question to the strongest feasible design given ethical and logistical constraints.
For exams, be ready to:
- Classify a scenario into survey, correlational, ex-post facto, pre-experimental, true, or quasi design.
- Identify whether random assignment, manipulation, and/or control groups are present.
- Justify strengths and weaknesses of each design type.
Remember: Random assignment + manipulation + control group = True Experiment; removing any one element shifts the design down the hierarchy of causal certainty.

Inspirational Note

“The best way to predict the future is to design it today.” – Anonymous. Let your research design be that act of shaping tomorrow.