Improving quantitative writing one sentence at a time

Abstract

Scientific writing, particularly quantitative writing, is difficult to master. To help undergraduate students write more clearly about data, we sought to deconstruct writing into discrete, specific elements. We focused on statements typically used to describe data found in the results sections of research articles (quantitative comparative statements, QC). In this paper, we define the essential components of a QC statement and the rules that govern those components. Clearly defined rules allowed us to quantify writing quality of QC statements (4C scoring). Using 4C scoring, we measured student writing gains in a post-test at the end of the term compared to a pre-test (37% improvement). In addition to overall score, 4C scoring provided insight into common writing mistakes by measuring presence/absence of each essential component. Student writing quality in lab reports improved when they practiced writing isolated QC statements. Although we observed a significant increase in writing quality in lab reports describing a simple experiment, we noted a decrease in writing quality when the complexity of the experimental system increased. Our data suggest a negative correlation of writing quality with complexity. We discuss how our data aligns with existing cognitive theories of writing and how science instructors might improve the scientific writing of their students.

Fig 1. 4C annotation of a quantitative comparative statement.

(A) Original quantitative comparative statement. (B) Identify and box the relational phrase with both magnitude and direction. (C) Circle what the relational phrase refers to (context). (D) Underline the comparison. (E) Fully 4C annotated quantitative comparative statement.

Fig 2. Quantitative comparative statements from results section of published research articles in major pan-discipline journals.

The mean (middle vertical line) ± SD are shown. Physical science papers are denoted in red, Biological sciences are in blue, and Social sciences are in green.

Fig 3. Using 4C scoring to measure quantitative writing before and after instruction.

(A) Mean 4C scores of quantitative comparative statements on an identical pre- and post- test. (B) Percent of statements that contain each of the essential components of a QC statement. (C) Percent difference between the pre-test and post-test broken down by essential components of QC statements. (***t-test, p < 0.001) Error bars in A represent Standard Error of the Mean (SEM).

Fig 4. Analysis of native writing samples (lab reports).

(A) Mean 4C scores of QC statements from lab reports (enzyme kinetics). (B) Mean 4C scores of QC statements from second lab reports (transcriptional regulation). (C) Percent difference between the two lab reports within a given year, broken down by essential components (*p < 0.05, ***p < 0.001) Error bars in A and B represent SEM.

Fig 5. Writing syntax is negatively impacted by complexity but can be improved with writing support.

(A) Writing syntax as a function of complexity measured by 4C scoring and reported as either unsupported (closed circles) or supported (open circles) by instructional intervention. Linear regression lines are shown (unsupported, R² = 0.9644, supported R² = 0.9471). (B) Students were stratified based on overall performance in the course. Statements from students within the group were averaged and reported. Error bars represent SEM.

Fig 6. Model describing the effect of complexity on writing ability.

(A) Simple linear model of the relationship between writing quality and complexity (cognitive load). (B) Model of the relationship between writing quality and complexity in which low complexity has minimal impact on writing quality but higher complexity negatively impacts writing quality.