TL;DR: Philippine student research typically draws from five quantitative designs: descriptive (what are current levels/perceptions), correlational (are two variables related), comparative (do groups differ), quasi-experimental (did an intervention cause a change, without random assignment), and experimental (did an intervention cause a change, with random assignment). Each pairs with specific statistical treatments: descriptive designs commonly use weighted mean and frequency/percentage, correlational designs use Pearson r, comparative designs use an independent or paired t-test for two groups or ANOVA for three or more, and designs involving categorical variables use chi-square. Matching the wrong test to your design is one of the most common Chapter 3 corrections advisers make.
Quantitative research design determines not just how you collect data, but which statistical test you're allowed to use to analyze it, and that choice is one of the first things a research adviser checks in Chapter 3. This guide covers the five designs Philippine senior high and college students most often work with, what question each one answers, and which statistical treatment, weighted mean, t-test, ANOVA, Pearson r, or chi-square, actually fits each. If you're not sure quantitative is the right track for your study at all, see qualitative vs quantitative research first; if you've settled on qualitative instead, see qualitative research designs.
What Is a Quantitative Research Design?
A quantitative research design is your plan for collecting and analyzing numerical data to answer a "how much," "how strongly related," or "is there a significant difference" question, as opposed to the meaning- and experience-focused questions qualitative research handles. As one methodology resource puts it, "research design refers to your approach for answering your fundamental research questions," and in quantitative work that approach determines exactly which statistical test can validly answer each of your specific research questions.
The five designs covered here, descriptive, correlational, comparative, quasi-experimental, and experimental, sit on a spectrum from simply describing a variable, to relating two variables, to comparing groups, to actually testing cause and effect.
Disclaimer: which design and statistical treatment fits your specific study depends on your data and research questions. Confirm your final choice with your adviser, statistician, or research coordinator before running your analysis.
The 5 Quantitative Research Designs and Their Statistical Treatment
| Design | Core Question | Typical Statistical Treatment | Requires Groups/Intervention? |
|---|---|---|---|
| Descriptive | What are the current levels, characteristics, or perceptions? | Frequency, percentage, mean, weighted mean | No |
| Correlational | Is there a relationship between two variables, and how strong? | Pearson r (or Spearman rho for non-normal/ordinal data) | No |
| Comparative (ex post facto) | Do pre-existing groups differ on a variable? | t-test (2 groups) or ANOVA (3+ groups) | Groups exist already, no intervention |
| Quasi-Experimental | Did an intervention cause a change, without random assignment? | t-test or ANOVA on pre/post or group comparison scores | Yes, intervention, non-random groups |
| Experimental | Did an intervention cause a change, with random assignment? | t-test or ANOVA on pre/post or group comparison scores | Yes, intervention, random assignment |
Descriptive Design: When You Just Need to Describe What's There
Descriptive design fits when your research question asks what the current state of something is, levels, frequencies, characteristics, or perceptions, without testing a relationship or a cause. The researcher measures and reports, without manipulating anything or comparing groups against each other in a formal statistical sense.
Statistical treatment: frequency counts and percentages for categorical data (like "how many respondents are male vs. female"), and mean or weighted mean for scaled/rating data (like a 4-point Likert-scale survey on student satisfaction). A weighted mean is computed by multiplying each response option by its assigned weight, summing across responses, and dividing by the total number of respondents, then interpreted against a verbal range the researcher defines (for example, scores near the top of a 4-point scale labeled "Strongly Agree").
Student-scale example: A survey measuring senior high students' self-reported level of stress during exam week, reported as an overall weighted mean with a verbal interpretation, is a straightforward descriptive study.
Correlational Design: When You're Testing a Relationship
Correlational design fits when your question asks whether two variables are related, and how strongly, without claiming one causes the other. It requires measuring both variables in the same group of respondents and testing the strength and direction of their relationship.
Statistical treatment: Pearson Product-Moment Correlation (Pearson r) is the standard test for two continuous, normally distributed variables; Spearman rank correlation is the common alternative when data is ordinal or not normally distributed. Both produce a correlation coefficient (from -1 to +1) indicating direction and strength, plus a significance value (commonly tested at the 0.05 level).
Student-scale example: A study measuring whether there's a significant relationship between college students' weekly study hours and their GPA, using Pearson r to test the correlation, is a classic correlational design. Important caveat: even a strong, significant Pearson r result shows association, not proof that more study hours caused the higher GPA.
Comparative Design: When You're Comparing Groups That Already Exist
Comparative design, sometimes called ex post facto or causal-comparative design, fits when you want to compare two or more pre-existing groups on some outcome, without you introducing any intervention. The "cause" (if any) already happened before the study began; you're just measuring whether groups defined by that pre-existing factor differ now.
Statistical treatment: an independent-samples t-test when comparing exactly two groups (for example, male vs. female students' anxiety scores), or one-way ANOVA (Analysis of Variance) when comparing three or more groups (for example, anxiety scores across four different year levels). ANOVA tells you whether at least one group differs significantly; a post-hoc test is then needed to identify which specific groups differ from each other.
Student-scale example: A study comparing the academic performance of students from public versus private senior high schools now attending the same college program uses a comparative design with a t-test, since the "public vs. private" grouping already existed before the study.
Quasi-Experimental Design: When You're Testing an Intervention Without Random Assignment
Quasi-experimental design fits when you want to test whether an intervention causes a change, but true random assignment of participants to groups isn't feasible, which describes most real classroom-based Philippine student research. Instead of randomizing individuals, researchers commonly use intact, already-existing groups (two sections of the same subject, for instance) as the experimental and comparison groups.
Statistical treatment: a paired-samples t-test for a single group's pretest-versus-posttest scores, or an independent-samples t-test/ANOVA comparing the experimental group's gain against a comparison group's gain. Chi-square applies instead when the outcome variable is categorical (passed/failed, for example) rather than a continuous score.
Student-scale example: A study testing whether a new reviewer format improves board-exam-style quiz scores, applied to one existing class section (experimental group) while another intact section continues with the usual method (comparison group), then comparing pretest-to-posttest gains between the two with a t-test, is a standard quasi-experimental design.
Experimental Design: When You Can Randomly Assign Participants
True experimental design fits the same "does this intervention cause a change" question as quasi-experimental design, but with one critical difference: participants are randomly assigned to the experimental or control group, which is what allows researchers to more confidently attribute any measured difference to the intervention itself rather than to pre-existing group differences. According to one methodology overview, this design "establishes procedures that allow the researcher to test a hypothesis and to systematically and scientifically study causal relationships among variables."
Statistical treatment: the same core tests as quasi-experimental design, t-test for two groups or ANOVA for three or more, applied to the difference in outcomes between randomly assigned groups. True random assignment is uncommon in Philippine classroom-based student research, since schools rarely allow researchers to randomly reshuffle students out of their assigned sections, which is exactly why the quasi-experimental design is far more common at the student level.
Which Statistical Treatment Fits Which Design?
| If your research question asks... | Use this design | Use this statistical treatment |
|---|---|---|
| What is the current level/perception of X? | Descriptive | Weighted mean, frequency, percentage |
| Is X related to Y, and how strongly? | Correlational | Pearson r (or Spearman rho) |
| Do two existing groups differ on X? | Comparative | Independent-samples t-test |
| Do three or more existing groups differ on X? | Comparative | One-way ANOVA |
| Did an intervention change X (no random assignment)? | Quasi-experimental | Paired t-test (pre/post) or t-test/ANOVA (group comparison) |
| Is there a relationship between two categorical variables? | Any (categorical data) | Chi-square |
| Did an intervention change X (random assignment)? | Experimental | t-test or ANOVA on group differences |
Most quantitative theses use more than one row of this table, since a single study typically has several sub-problems, description, relationship, and comparison, each needing its own matching treatment. Map each of your sub-problems from your statement of the problem to a specific test before you touch Chapter 3, rather than picking a design and hoping the statistics fit afterward.
Common Mistakes Students Make With Quantitative Design
- Claiming causation from a correlational design. A significant Pearson r shows a relationship exists, not that one variable caused the other; that claim needs an experimental or quasi-experimental design instead.
- Using a t-test when you have three or more groups. Running multiple t-tests instead of one ANOVA inflates your chance of a false-positive result; use ANOVA (with post-hoc tests) for 3+ groups.
- Skipping the verbal interpretation for a weighted mean. A raw mean like "3.42" means little to a reader without a defined interpretation range (for example, what range counts as "Agree" versus "Strongly Agree") stated in Chapter 3.
- Calling a study "experimental" when groups weren't randomly assigned. If you used two intact class sections, that's quasi-experimental, not experimental; the distinction matters to a defense panel.
- Choosing a statistical test before finalizing the research question. The test should follow from what you're asking, not the other way around; picking ANOVA because it "sounds more advanced" without matching data is a common early mistake.
Ready to find the right school? Browse schools or compare programs across hundreds of Philippine colleges and universities on SchoolFinderPH.
Sources
- Types of Quantitative Research Design — Grand Canyon University
- DepEd Practical Research 2 Curriculum Guide (SHS)
- SchoolFinderPH — Qualitative vs Quantitative Research
- SchoolFinderPH — Qualitative Research Designs
- SchoolFinderPH — Statement of the Problem Guide
- SchoolFinderPH — Thesis Chapters 1 to 5
