Interpreting Data in Multimodal Academic and Technical Texts

Fundamentals of Multimodal Academic Texts and Forms of Data Representation

A multimodal text is defined as a communicative work that integrates two or more distinct semiotic modes, combining written words, numerical values, graphical images, data charts, schematic diagrams, and deliberate spatial layouts. Within academic and technical literature, expository prose and graphical displays function in strict synergy: the running text articulates the contextual framework, theoretical reasoning, and analytical explanations, while visual data displays furnish concrete empirical evidence to substantiate those assertions. The integration of textual discourse with quantitative representations ensures that complex information is both transparently demonstrated and rigorously defended.

Selecting an appropriate graphical representation requires matching the structural properties of the dataset to the functional advantages of a specific visual medium. No single display form is universally superior; each format possesses distinct analytical capabilities tailored to particular types of data relationships. A table is uniquely suited for presenting exact numerical values and organizing multi-variable categorizations. A bar chart is designed for discrete categorical comparisons across distinct groups. A line graph excels at illustrating continuous trends, rates of change, and directional shifts over chronological intervals. A pie chart is dedicated to displaying proportional distributions that constitute parts of a single whole. Specialized figures and diagrams are employed to depict multi-stage processes, operational sequences, organizational structures, and theoretical relationships. Mismatching the data type with its visual form produces cognitive friction and misleads the reader; for instance, plotting independent categorical entities along a continuous line graph erroneously implies a temporal or sequential continuity that does not exist.

Anatomy of Graphical Displays and Systematic Reading Protocols

Every structural component of a chart performs an indispensable communicative function that clarifies the scope and meaning of the underlying data. The title establishes the topical domain, the geographic or institutional location, and the precise time frame under investigation. Axis labels explicitly identify the dependent and independent variables being measured, while the axis scale defines the numeric range and interval increments, such as progression in units of ten ($0, 10, 20, 30\dots$). The legend decodes visual signifiers, explaining the meaning of varied colors, shading patterns, or graphical symbols. Unit designations indicate the exact physical or financial metrics applied, such as percentages ($\%$), temporal durations (hours), currencies ($₱$), or masses ($kg$). Finally, the source attribution documents the institutional or empirical provenance of the information, enabling verification and establishing academic accountability.

Neglecting or misinterpreting individual graphical elements fundamentally alters data interpretation. For instance, a graphical bar labeled solely with the numerical value $250$ remains entirely ambiguous without context: it could represent $250$ enrolled students, an expenditure of $₱250$, or a physical yield of $250\,kg$, just as it could reflect metrics collected in $2019$ or $2024$, confined to a single academic institution or aggregated across an entire nation. The formulation of titles similarly dictates interpretative clarity. A rudimentary title such as "Sales" fails to convey analytical parameters, whereas an exhaustive title such as "Monthly Sales of Store A in Pesos, January–June 2024" specifies what is measured, the precise entity evaluated, the currency unit employed, and the exact chronological window. To extract data methodically without introducing cognitive bias, readers must adhere to an invariant reading order: begin with the Title, proceed to the Axis Labels and Units, evaluate the Scale increments, decipher the Legend, analyze the internal Data points, and verify the Source note.

Structural Analysis of Tables, Bar Charts, Line Graphs, and Diagrams

Tables present organized tabular data where exact numeric precision is paramount. A standard academic table comprises a descriptive title, column headings identifying primary variables, comparison groups, row labels denoting categorical entries, individual data cells, explicit units of measurement, summary rows (such as totals), and an explanatory source note. In Table 1, titled "Preferred Study Mode of Grade 11 Students ($n = 200$)," three study modalities are quantified: Online resources account for $80$ students ($40\%$), Printed modules represent $70$ students ($35\%$), and Group study accounts for $50$ students ($25\%$), yielding a total sample ($n$) of $200$ students ($100\%$). To extract values accurately, the reader locates the intersection where a specific row meets a target column. A necessary component of reading tables is performing internal arithmetic logic checks to confirm data validity: summing the discrete frequencies ($80 + 70 + 50 = 200$) and confirming that the proportional distribution equals unity ($40\% + 35\% + 25\% = 100\%$). Discrepancies in these arithmetic totals serve as immediate warnings of editorial errors or omitted categories. Furthermore, the explicitly stated sample size ($n = 200$) provides the metric baseline required to assess the empirical reliability of the survey.

Bar charts compare discrete, non-continuous categories through vertical or horizontal rectangular bars whose lengths correspond directly to their numerical values. In Figure 1, titled "Average Daily Screen Time by Strand (hours)," the y-axis delineates daily duration on a scale from $0$ to $8\,\text{hours}$, while the x-axis categorizes four distinct academic strands. Humanities and Social Sciences (HUMSS) students recorded the highest average screen time at $7.0\,\text{hours}$ per day, followed by Technical-Vocational-Livelihood (TVL) at $6.5\,\text{hours}$, Accountancy, Business, and Management (ABM) at $6.2\,\text{hours}$, and Science, Technology, Engineering, and Mathematics (STEM) at the lowest average of $5.5\,\text{hours}$. Interpreting these metrics requires relative proportional calculation: the absolute disparity between the highest and lowest cohorts is $1.5\,\text{hours}$ ($7.0 - 5.5 = 1.5$), demonstrating that HUMSS screen time exceeds STEM screen time by approximately $27\%$ ($1.5 \div 5.5 \approx 0.2727$).

Line graphs map continuous numeric values over sequential intervals, typically tracking chronological movement by linking discrete data points with straight line segments. The gradient or slope of the line carries explicit qualitative meaning: a steep upward trajectory indicates rapid expansion, a horizontal line signifies stability or a plateau, and a downward trajectory reflects an active decline. In Figure 2, titled "Students Using the School's E-Learning Platform, 2020–2024," user engagement is plotted chronologically: $120$ users in $2020$, $250$ users in $2021$, $410$ users in $2022$, $430$ users in $2023$, and $435$ users in $2024$. The trajectory demonstrates a sharp initial surge between $2020$ and $2022$, adding $290$ users across two years ($410 - 120 = 290$), before transitioning into an extended leveling off between $2022$ and $2024$, adding only $25$ users over the subsequent two-year period. In multi-line graphs displaying two or more distinct cohorts simultaneously, an unambiguous legend is strictly required to prevent incorrect group attribution.

Pie charts depict proportional distributions of a singular whole and remain valid only when all component slices sum to exactly $100\%$ of a unified category. In Figure 3, titled "Main Source of News Among Students (\%)," the distribution reveals that Social media constitutes the dominant source at $55\%$ ($0.55$), followed by Television at $20\%$ ($0.20$), Online news sites at $15\%$ ($0.15$), and Print newspapers at $10\%$ ($0.10$). In formal academic prose, the term "majority" is reserved exclusively for categories that exceed $50\%$; when the largest segment fails to surpass this threshold, it must instead be designated as "the largest proportion" or "the largest share." Pie charts become illegible and analytically ineffective when partitioned into more than five or six slices, under which conditions bar charts or tables should be selected. Complementing quantitative charts, process diagrams visualize procedural sequences and relational structures rather than absolute magnitudes. In Figure 4, titled "Water Filtration Process," directional flow arrows guide the reader through five sequential stages: raw water enters the screening phase to eliminate coarse debris, undergoes sedimentation to allow suspended solids to settle, passes through mechanical filtration, is subjected to chemical disinfection, and ultimately yields purified, potable water.

Cognitive Tiers in Data Interpretation: The Three Levels of Reading

Comprehensive comprehension of multimodal texts requires advancing through three increasingly complex cognitive tiers: reading the data, reading between the data, and reading beyond the data. Level 1, reading the data, consists of direct informational extraction. It focuses exclusively on descriptive accuracy, identifying verbatim metrics, explicit variable names, exact units of measurement, and temporal dates directly recorded on the visual display without subjective manipulation. In the context of Figure 2, a Level 1 reading consists of extracting the absolute fact that in the year $2022$, platform enrollment reached exactly $410$ users.

Level 2, reading between the data, shifts the focus from isolated data points to mathematical relationships, proportional comparisons, and ongoing patterns. This level involves computing rate differences, comparing cohort extremes, and tracking directional momentum across time intervals. Applied to Figure 2, a Level 2 reading observes that while cumulative user enrollment grew continuously across the entire five-year span, the rate of increase moderated substantially after $2022$, shifting from exponential growth to marginal annual gains.

Level 3, reading beyond the data, involves inductive reasoning, synthesizing empirical evidence with logical inference, contextual factors, and broader implications. Because Level 3 moves beyond direct visual proof into reasoned extrapolation, it requires epistemic hedging words such as may, suggests, indicates, and is likely to. In Figure 2, a Level 3 interpretation reasons that the dramatic post-2022 plateau suggests that the e-learning platform has saturated its addressable institutional demographic, meaning that virtually all students with digital accessibility have already been onboarded into the active user base.

Academic Lexicon for Describing Trends, Magnitudes, and Comparative Metrics

Communicating empirical data within academic and technical contexts requires precise, objective vocabulary that replaces vague colloquial expressions with standardized verbs, modifiers, and proportional terms. Informal assertions such as "the graph went way up and then kinda stopped" undermine analytical authority and fail to describe the magnitude of the change. Academic register requires precise terminology: "the number of users rose sharply between $2020$ and $2022$ and subsequently leveled off through $2024$."

To articulate upward movement, academic writers employ graduated verbs such as rose, increased, climbed, or surged, reserving surged for rapid, high-magnitude ascents. Conversely, downward shifts are communicated through verbs such as declined, fell, dropped, or plummeted, with plummeted indicating an exceptionally steep collapse. Trajectories that cease fluctuating are captured through phrases denoting stability, such as remained stable, leveled off, or stayed constant. For example, an analysis of weekly academic habits might state that study durations declined from $15\,\text{hours}$ to $10\,\text{hours}$ in Week 2, remained stable at $10\,\text{hours}$ across Weeks 2 through 4, and surged to a peak of $20\,\text{hours}$ in Week 5.

Proportional distributions and comparative evaluations require careful alignment between numerical percentages and fractional phrasing: an exact metric of $25\%$ is translated as a quarter, $50\%$ as half, and $75\%$ as three quarters. Extreme metrics should be documented using specialized boundary phrases, such as peaked at to denote maximum values and reached a low of to establish minimum baselines. When evaluating relationships between independent cohorts, analytical prose utilizes comparative markers such as higher than, lower than, the largest share, the second choice, or twice as much as.

Consider the academic performance metrics in Table 2, titled "Grade 11 Quarterly Average in Effective Communication (Class A vs. Class B)." In the 1st Quarter, Class A scored $80$ while Class B scored $78$. In the 2nd Quarter, Class A achieved $83$ against Class B's $79$. In the 3rd Quarter, Class A reached $85$ while Class B recorded $84$. In the 4th Quarter, Class A closed at $88$ while Class B attained $87$. Comparative analysis demonstrates that Class A maintained higher academic scores than Class B across all four quarters. The performance disparity was widest during the 2nd Quarter with a $4$-point gap ($83 - 79 = 4$) and narrowest during the 3rd and 4th Quarters with a $1$-point margin ($85 - 84 = 1$; $88 - 87 = 1$). Concurrently, trend analysis confirms that both sections improved steadily throughout the academic year: Class A improved overall by $8$ points ($88 - 80 = 8$), while Class B climbed by $9$ points ($87 - 78 = 9$). When introducing empirical findings, personal subjective declarations such as "I think the graph shows" must be rejected in favor of formal attribution frames, including "The data indicate that…" or "As shown in Table 2…"

The Read-Interpret-Conclude Analytical Framework and Valid Inference Formulation

The Read-Interpret-Conclude method provides a systematic approach for evaluating multimodal evidence without introducing analytical errors. The initial phase, Read, establishes the basic parameters of the display, including the sample size, the categorical variables evaluated, and the descriptive metrics reported. The subsequent phase, Interpret, identifies structural relationships, contrasting the primary group against subordinate groups and calculating internal differences. The culminating phase, Conclude, synthesizes these empirical relationships into an overarching finding, while explicitly identifying the analytical limitations of the visual.

Applying this three-phase framework to Table 1 demonstrates its practical use. During the Read stage, the analyst notes that $200$ Grade 11 students were surveyed regarding three study methods, with results tabulated as absolute counts and relative percentages. In the Interpret stage, the analyst notes that online resources represent the primary preference at $40\%$ ($80$ students), printed modules represent the secondary choice at $35\%$ ($70$ students), and collaborative group study represents the smallest share at $25\%$ ($50$ students). In the Conclude stage, the analyst reasons that Grade 11 students lean heavily toward individualized, independent learning modalities, as online and print self-study combine to account for $75\%$ of student choices ($40\% + 35\% = 75\%$). However, the analyst also acknowledges the limitations of the data: the survey does not reveal the underlying pedagogical motivations for these preferences, nor does it measure whether self-directed study produces superior academic achievement compared to collaborative group work.

Maintaining validity requires distinguishing direct conclusions from logical inferences. A conclusion is an assertion that rests directly on empirical data, whereas an inference is an explanatory hypothesis that integrates recorded evidence with logical deduction. Valid conclusions must remain supported, limited, and cautious. Data analysts must guard against common interpretive errors: overgeneralizing findings beyond the target demographic, confusing statistical correlation with direct causation, ignoring chronological windows, neglecting sample size constraints, and making unsubstantiated predictive forecasts.

These distinctions are evident when examining Figure 3 on student news consumption. The assertion that "most surveyed students get their news from social media" constitutes a valid conclusion, given that social media accounts for $55\%$ of the sample. Conversely, claiming that "newspapers are no longer useful" represents an invalid conclusion, because the underlying metric quantifies student consumption frequency rather than the utility, accuracy, or societal value of print journalism. Similarly, asserting that "all Filipinos get news from social media" represents an invalid overgeneralization, because the survey sample was confined to a specific cohort of students rather than the national population. In the same vein, demonstrating a statistical correlation between variables—such as an association between extended sleep duration and higher academic grades—does not prove that sleep directly causes higher performance, as unmeasured confounding variables, such as disciplined study routines, may drive the observed outcome.

Data Integrity: Scale Distortion, Truncated Axes, and Visual Traps

Visual data displays can be manipulated to mislead viewers while maintaining technical numerical accuracy. A common technique for visual distortion is the truncated axis, where an axis baseline starts at an arbitrary non-zero value rather than at zero. Altering the baseline changes the physical proportions of graphical elements, distorting the perceived differences between categories.

Consider a performance evaluation comparing Class A ($95$ average) and Class B ($100$ average). When plotted on an honest scale where the y-axis starts at $0$ and extends to $100$, the physical heights of the two bars appear virtually identical, visually reflecting that the actual difference is only $5$ points, or an approximate $5\%$ variance. However, if the y-axis is truncated to begin at $90$, the visible height of Class A's bar represents only $5$ units ($95 - 90 = 5$), while Class B's bar represents $10$ units ($100 - 90 = 10$). This alteration produces a $2:1$ physical height ratio, creating the false visual impression that Class B performed twice as well as Class A. While axis truncation is acceptable in specialized line graphs tracking slight fluctuations within naturally bounded, non-zero systems—such as physiological human body temperatures or meteorological indices—such adjustments must be prominently labeled. In bar charts, where bar length directly represents value, the baseline must start at zero to avoid visual distortion.

Beyond truncated axes, readers must remain alert to several recurring visual traps and misrepresentations. Cherry-picking occurs when an author presents only favorable subsets of a timeline while suppressing adjacent declines, such as highlighting sales gains from March through May while omitting steep losses in January and February. Uneven axis intervals mislead readers by varying scale steps without visual warning, such as transitioning irregularly from $0$ to $10$, $20$, $50$, and $100$. Three-dimensional projections and decorative volumetric styling skew spatial geometry, making foreground categories appear disproportionately larger than background wedges or bars. Inadequate or undisclosed sample sizes compromise data reliability; asserting that "$9\,\text{out of}\,10$ individuals prefer a product" creates a false sense of consensus if the study surveyed only $10$ participants ($n = 10$). Inconsistent mathematical distributions, such as pie charts whose components sum to an erroneous total ($50\% + 40\% + 30\% = 120\%$), indicate severe analytical flaws or hidden variables. Finally, mismatched titles mislead audiences by asserting broad thematic conclusions that the underlying data do not support.

Evaluating the fairness and technical integrity of multimodal visuals requires reviewing a standardized verification checklist: verify whether the title matches the data; ensure that axis units, categorical labels, and scales are clearly identified and uniform; verify that all percentage shares sum to $100\%$; ensure that the chronological timeframe is presented without omissions; confirm that the absolute sample size ($n$) is explicitly documented; and verify that the visual presentation is free from geometric, volumetric, and baseline scale distortions.

Assessing Source Credibility, Intellectual Integrity, and Audience Alignment

Evaluating data displays requires determining the credibility and institutional integrity of the publishing entity. Critical evaluation centers on four core considerations: identifying the author or organization and publication date; assessing whether the source has a history of rigorous, impartial analysis; verifying whether the empirical metrics are current; and evaluating whether the publishing entity has financial, ideological, or institutional conflicts of interest. Data provided by public statistical bureaus, accredited universities, and peer-reviewed journals generally offer higher credibility than unsourced digital infographics or private corporate reports. For example, a commercial beverage manufacturer publishing internal research claiming its product is the "healthiest choice" presents a clear conflict of interest that requires skepticism, whereas an independent public health agency evaluating dietary metrics provides a more reliable foundation. Similarly, currency is critical: technical metrics collected in $2005$ cannot accurately describe technological ecosystems in $2025$.

Maintaining intellectual integrity requires presenting data transparently, attributing source provenance, and never fabricating, altering, or selectively omitting empirical results. Using misleading visual formats or concealing sources constitutes academic dishonesty. In media and information literacy, visually sophisticated charts are often used to lend false authority to disinformation. Readers and writers must evaluate visual evidence with the same scrutiny applied to written claims.

Selecting a data representation also depends on the target audience and communicative purpose. A general audience seeking quick awareness benefits from simplified, high-contrast bar charts or accessible infographics. Researchers and technical specialists require data density, detailed tables, explicit sample sizes, standardized units, and methodology notes. Technicians and operational trainees require flowcharts or process diagrams showing procedural sequences and functional components. Organizational decision-makers and business managers rely on clean line graphs that clearly show operational performance and institutional trends. The author must answer three design questions: Who is the reader? What core insight must they understand? Which visual format presents this evidence most clearly and honestly? Matching the visual medium to the audience's needs is both an aesthetic and an ethical responsibility to ensure honest communication.

Principles of Academic Data Commentary Construction

An academic data commentary provides a structured written analysis of a visual display. A complete commentary consists of four core elements: introducing the visual display and identifying its topical focus; reporting the most important numerical values; interpreting the primary trends or comparative differences; and concluding with an inference while providing a formal citation.

A commentary based on Figure 2 demonstrates this four-part structure: "Figure 2 shows the number of students using the school's e-learning platform from $2020$ to $2024$. Usage rose sharply from $120$ users in $2020$ to $410$ in $2022$, then leveled off at $435$ by $2024$. This pattern suggests that the platform was quickly adopted but may have reached its maximum audience. (Source: Hypothetical school data, $2024$.)" In this commentary, the opening sentence establishes the visual reference, topic, and timeframe. The second sentence provides specific quantitative values and uses precise verbs (rose sharply, leveled off). The third sentence offers an inference using appropriate hedging language (suggests, may have). The passage concludes with an explicit parenthetical source attribution.

Effective commentaries balance quantitative reporting with analytical conciseness, typically spanning three to five sentences. Authors should report key values rather than transcribing every figure in the chart. Commentaries must remain objective, avoiding colloquialisms, ungrounded speculation, and subjective assertions. Sentence starters such as "The data indicate that…", "As illustrated in…", "This trend suggests that…", and "However, the evidence does not demonstrate that…" help maintain academic tone.

Comparing weak and strong commentaries for Figure 3 highlights these principles. A weak commentary reads: "Lots of students use social media for news. It's kind of a big deal. I think newspapers are dying." This draft fails because it uses imprecise modifiers (lots), informal language (kind of a big deal), an unhedged and unsupported opinion (I think newspapers are dying), and omits source citations. In contrast, a strong commentary states: "As shown in Figure 3, social media is the main news source for the majority of surveyed students ($55\%$), while newspapers account for only $10\%$. This suggests that print news reaches a small share of this group. (Source: Hypothetical student survey, $2025$.)" The strong version relies on exact metrics ($55\%$, $10\%$), uses appropriate vocabulary (majority, account for), limits its inference to the surveyed group using hedging language (suggests), and provides an explicit source citation.

Academic writers use standardized citation conventions to document data sources. Below a figure, formal attribution appears as a caption: "Source: Philippine Statistics Authority ($2023$)." Within running prose, sources are integrated through narrative attribution: "According to the Department of Education ($2024$)…" In formal bibliographies or reference lists, the entry includes the corporate or individual author, publication year, report title, and publishing entity or uniform resource locator (URL). When an author presents original empirical data collected personally, academic transparency requires acknowledging its provenance: "Source: Author's survey, $2025$." Providing clear citations credits the original researchers, allows readers to verify the underlying data, and establishes the writer's academic credibility.