TL;DR: Chapter 4 is a workflow, not a write-up sprint. Clean and code, run descriptives, run the inferential test, present table first and narrative second, then interpret without overclaiming. Every section maps to one numbered sub-problem, in the same order. The statistical treatment was already decided in Chapter 3, so analysis should hold no surprises. This guide works a frequency table, a weighted mean and a Pearson correlation all the way through the arithmetic on an illustrative dataset, then shows how to report each in APA form.
Most guides to data analysis stop at "use the appropriate statistical treatment," which is exactly the sentence that is no help at 2am. This one runs the numbers. Everything below uses a small constructed dataset so that you can see each figure appear, check the arithmetic yourself, and recognise the same shape in your own data.
Three things should already be settled before you touch Chapter 4: the sub-problems, the statistical treatment named in Chapter 3, and the level of significance. If any of those is still open, fix it there first. Analysis that improvises its own method is the version panels take apart.
Which subject is this for? Practical Research 1 and 2 are applied subjects of the 2016 K to 12 Senior High School curriculum, and that structure is being replaced. DepEd Order No. 017, s. 2026 institutionalises the Strengthened Senior High School Curriculum for Grade 11 from SY 2026-2027 and Grade 12 from SY 2027-2028, and its Annex A lists Research 1 and Research 2 as 80-hour Academic Track electives in the Field Experience cluster rather than applied subjects everyone takes. Grade 12 learners not enrolled in pilot schools continue under the 2016 curriculum for SY 2026-2027. The arithmetic below is unchanged either way. See our guide to the Strengthened SHS Curriculum.
What Chapter 4 Contains
One section per specific sub-problem, in the same order as your statement of the problem, and each section built the same way:
- The table. Numbered, titled, and readable on its own.
- The narrative. Two to four sentences reading the table, pointing at the highest and lowest values and anything notable.
- The interpretation. What the numbers mean for your respondents, connected back to the literature where it applies.
What does not belong: conclusions, recommendations, and any statistic you did not announce in Chapter 3.
Step 1: Clean and Code Before You Compute Anything
The dataset used from here on is illustrative, constructed for this guide. It is not real data from any school.
Illustrative study: the relationship between daily smartphone screen time and self-reported sleep quality among Grade 11 students in one public senior high school in Cebu City. Of 195 questionnaires distributed, 184 were returned and 178 were usable.
Cleaning is unglamorous and non-optional:
- Discard unusable responses, and record how many and why. Six were discarded here for incomplete Part III responses.
- Reverse-score the reverse-worded items. On a 4-point scale, 4 becomes 1, 3 becomes 2, 2 becomes 3, 1 becomes 4. Skipping this quietly corrupts every scale mean and the reliability coefficient. See building a research questionnaire for why the reverse item is there in the first place.
- Code categorical variables consistently. Sex as 1 and 2, strand as 1 to 4. These codes are labels, not quantities, and must never enter a correlation. Our guide to independent and dependent variables covers why.
- Range-check every column. A 7 in a column that only allows 1 to 4 is an encoding error, not a finding.
- Fix your n and keep it fixed. Every table from here on totals 178.
Step 2: The Frequency and Percentage Table
The first sub-problem is descriptive, so the treatment is frequency and percentage. Percentage is frequency divided by total, times 100.
Table 1. Profile of respondents by strand (illustrative, n = 178)
| Strand | Frequency | Percentage |
|---|---|---|
| STEM | 58 | 32.6 |
| HUMSS | 51 | 28.7 |
| ABM | 41 | 23.0 |
| TVL | 28 | 15.7 |
| Total | 178 | 100.0 |
The narrative under it does one job: read the table without repeating every row. "Most respondents were from the STEM strand (32.6 percent), followed by HUMSS (28.7 percent). TVL students were the smallest group at 15.7 percent."
Two habits to build. Percentages should sum to 100.0, and when rounding pushes them to 99.9 or 100.1 you add a note rather than silently editing a number. And the total must equal your usable-response count, which is the figure a panel cross-checks against your description of the respondents of the study.
Step 3: The Weighted Mean, Computed by Hand
The second descriptive sub-problem asks about the level of a Likert-measured variable, so the treatment is the weighted mean.
Take one item: "I feel rested when I wake up on a school day," on a 4-point scale where 4 is Always and 1 is Never.
Table 2. Responses to item 7 (illustrative, n = 178)
| Response | Scale value | Frequency | Product |
|---|---|---|---|
| Always | 4 | 32 | 128 |
| Often | 3 | 54 | 162 |
| Sometimes | 2 | 71 | 142 |
| Never | 1 | 21 | 21 |
| Total | 178 | 453 |
Weighted mean = 453 / 178 = 2.5449, rounded to 2.54
Now read it against the interpretation range table fixed in Chapter 3. On the common 4-point convention (3.26 to 4.00 Always, 2.51 to 3.25 Often, 1.76 to 2.50 Sometimes, 1.00 to 1.75 Never), 2.54 falls in the Often band.
Look closely at that result, because it teaches something. 2.54 sits 0.03 above the boundary. Had four more respondents answered Sometimes instead of Often, the verbal interpretation would flip to a different band while the underlying data barely moved. This is why a mean is always reported with its verbal interpretation and never in place of it, and why a defensible Chapter 5 does not hang a conclusion on a value sitting on a boundary. Saying so explicitly is a sign of a careful researcher, not a weak one.
The overall scale mean is computed the same way across all five items, and that scale mean, not any single item, is what enters the correlation.
Step 4: The Correlation, Worked Step by Step
The relational sub-problem asks whether daily screen time relates to sleep quality. Both are continuous, so the treatment is Pearson r, as planned in Chapter 3 alongside the research hypothesis.
The eight cases below are an illustrative demonstration dataset, small enough to work by hand. A real study would use all 178 cases; eight is far too few for a defensible finding.
| Case | Screen time, hours (x) | Sleep quality mean (y) |
|---|---|---|
| 1 | 3.0 | 3.4 |
| 2 | 4.5 | 2.6 |
| 3 | 5.0 | 3.2 |
| 4 | 6.0 | 2.8 |
| 5 | 6.5 | 3.0 |
| 6 | 7.5 | 2.2 |
| 7 | 8.0 | 2.6 |
| 8 | 9.5 | 2.0 |
The computational formula is:
r = [n(Σxy) - (Σx)(Σy)] / √{[n(Σx²) - (Σx)²][n(Σy²) - (Σy)²]}
Compute the five sums first:
n = 8
Σx = 50.0 Σy = 21.8
Σxy = 130.5 Σx² = 343.0 Σy² = 61.0
Numerator:
8(130.5) - (50.0)(21.8) = 1044 - 1090 = -46
Denominator:
8(343.0) - 50.0² = 2744 - 2500 = 244
8(61.0) - 21.8² = 488 - 475.24 = 12.76
√(244 × 12.76) = √3113.44 = 55.798
Result:
r = -46 / 55.798 = -0.8244, rounded to -0.82
Then test it. With df = n - 2 = 6, the t-statistic is t = r√[(n - 2)/(1 - r²)] = -3.57, which for a two-tailed test gives p = .012. Since .012 is less than the .05 level fixed in Chapter 3, the null hypothesis of no significant relationship is rejected.
How to word that, exactly:
A significant negative relationship was found between daily smartphone screen time and self-reported sleep quality, r(6) = -0.82, p = .012. Respondents reporting longer daily screen time tended to report lower sleep quality.
How not to word it. Not "screen time causes poor sleep," because a correlational design cannot support that. Not "the hypothesis is proven," because tests reject or fail to reject. And not "there is a very strong effect," because the correlation coefficient describes the strength of a linear association, not the size or importance of an effect. Which designs can support which claims is set out in our guide to quantitative research designs, linked in the sources below.
Step 5: Reporting It in APA Form
| Element | Convention | Example |
|---|---|---|
| Statistical symbols | Italicised | r, p, M, SD, t, n |
| Values that cannot exceed 1 | No leading zero | p = .012, not 0.012 |
| Values that can exceed 1 | Keep the leading zero | M = 2.54 |
| Degrees of freedom | In parentheses after the symbol | t(6) = -3.57 |
| p-values | Report exact, unless very small | p = .012; p < .001 |
| Tables | Numbered, titled, referred to in the text | "as shown in Table 2" |
Departments vary on how strictly APA is enforced, and some Philippine manuals impose their own table format. Follow yours where it differs. What is universal is that a reader should be able to reconstruct what you did from what you reported.
If Your Data Is Qualitative Instead
None of the above applies. Qualitative analysis produces themes, not coefficients, and the most common route in student work is thematic analysis, set out in Braun and Clarke's widely used 2006 account in Qualitative Research in Psychology.
The working sequence is: transcribe the interviews verbatim; read the transcripts repeatedly before coding anything; code meaningful segments; group codes into candidate themes; review the themes against the whole dataset; define and name the final themes; then write each theme up with verbatim excerpts as evidence.
Two cautions. Do not report percentages of participants per theme as if they were survey results; with eight purposively selected participants, "62.5 percent" is a misleading way to say "five people." And do not paraphrase away the quotations. The excerpts are the evidence, in the same way the correlation coefficient is the evidence in the quantitative case. Our guide to qualitative research designs covers how the design shapes what counts as an adequate analysis.
Common Mistakes That Cost Marks
- Analysing data the sub-problems never asked for. If it has no sub-problem, it has no place in Chapter 4.
- Changing the statistical treatment after seeing the results. The treatment was declared in Chapter 3.
- Reporting a mean with no verbal interpretation, or interpreting it against a range table invented after the fact.
- Forgetting to reverse-score. It quietly corrupts every downstream number.
- Correlating coded categorical variables. Strand coded 1 to 4 is not a quantity.
- Inconsistent n across tables. The fastest way for a panel to find something else wrong.
- Writing conclusions inside Chapter 4. Description and interpretation here, judgement in Chapter 5.
- Claiming causation from correlation. Still the most common single error in student research papers.
A Short Checklist Before You Write Chapter 4
- Confirm your usable n, and use it in every table.
- Reverse-score, range-check and code the dataset, then save an untouched copy of the raw file.
- Lay out one section per sub-problem, in order.
- Run only the treatments named in Chapter 3.
- Draft each section as table, narrative, interpretation, in that order.
- Read every interpretation sentence and ask whether the design can support it. Delete the ones that cannot.
Ready to find the right school? Browse schools or compare programs and reviews across hundreds of Philippine schools on SchoolFinderPH.
Disclaimer: departmental research manuals differ on table formats, interpretation ranges and citation style. The arithmetic in this guide is standard, but the presentation rules are your school's to set. Confirm them before you format anything.
Sources
- DepEd — K to 12 Senior High School Applied Track, Practical Research 2 Curriculum Guide — competencies CS_RS12-IId-g-2 and CS_RS12-IId-g-3 on presenting and interpreting data and using statistical techniques limited to bivariate analysis
- DepEd — K to 12 Senior High School Applied Track, Practical Research 1 Curriculum Guide
- APA Style — Journal Article Reporting Standards, quantitative research
- APA Style — Numbers and statistics guidelines
- APA Style — Tables and figures
- Braun, V. and Clarke, V. (2006), "Using thematic analysis in psychology," Qualitative Research in Psychology, 3(2), 77-101 — the six-phase thematic analysis sequence summarised above
- SchoolFinderPH — Quantitative Research Designs
- SchoolFinderPH — Thesis Chapters 1 to 5
