Figure Table Checker

Figure and Table Checker

Figures and tables are the most scrutinised elements of any research paper. Reviewers spot non-compliant formatting immediately. This skill covers universal standards, discipline-specific requirements, and the design principles that separate publication-ready visuals from rejected manuscripts.

---

Figure Compliance Checklist (Universal)

Apply to every figure before submission:

• [ ] Numbered sequentially in order of first mention in text (Figure 1, Figure 2…)
• [ ] Caption below the figure — always below, in complete sentence form
• [ ] Caption is self-contained — reader can understand the figure without reading the paper
• [ ] All axes labelled with units in parentheses: Time (s), Concentration (mM)
• [ ] Legend included where multiple data series are plotted
• [ ] Error bars defined — type stated explicitly in caption or methods (SD, SEM, or 95% CI)
• [ ] Statistical significance markers explained (e.g., p < .001, p < .01, p < .05)
• [ ] Sample sizes visible — either in legend, caption, or on the figure itself
• [ ] Font size legible at final print size — minimum 8pt for labels, 10pt preferred
• [ ] Colorblind-accessible palette used (see colour section below)
• [ ] Resolution sufficient — 300 dpi minimum for publication (600 dpi for line art)
• [ ] File format appropriate — check journal requirements (TIFF, EPS, PDF, PNG)
• [ ] No missing data without explanation
• [ ] Axes start at zero for bar charts (unless scale is justified and stated)
• [ ] Individual data points shown for small samples (n < 20) alongside summary statistics

---

Table Compliance Checklist (Universal)

Apply to every table before submission:

• [ ] Numbered sequentially (Table 1, Table 2…)
• [ ] Title/caption above the table — always above, in noun-phrase form (not a sentence)
• [ ] Column headers clear — units in header, not repeated in each cell
• [ ] No vertical lines — professional tables use horizontal rules only (booktabs principle)
• [ ] Minimal horizontal lines — top, below header, and bottom only (APA style)
• [ ] Consistent decimal places within each column
• [ ] Alignment appropriate — numbers right-aligned or decimal-aligned; text left-aligned
• [ ] Note(s) section present if abbreviations used or statistical thresholds need explanation
• [ ] Missing data represented consistently (em dash —, or stated as N/A with note)
• [ ] No duplication of data already presented in a figure (unless summary comparison needed)
• [ ] Spanning rows/columns used appropriately, not for decoration

---

APA 7th Table Format in Detail

Structure:

Table X
[Title in italics, above the table, title case]
[Blank line]
[Table with horizontal rules only]
[Blank line]
Note. [General note explaining abbreviations or scope.]
*p < .05. **p < .01. ***p < .001.

Three types of table notes (APA 7th):

1. General note — applies to entire table; starts with "Note."
2. Specific note — applies to a column, row, or cell; uses superscript lowercase letters (a, b, c)
3. Probability note — defines significance levels; uses asterisks (, , )

Example:

Table 2
Descriptive Statistics for Study Variables by Group

Variable        Control (n = 45)    Treatment (n = 47)
                M        SD          M         SD
----------------------------------------------------------
Anxiety        14.3      3.2        11.8        2.9
Depression     12.1      4.1         9.4        3.7
Wellbeing      22.7      5.3        26.4        4.8
----------------------------------------------------------

Note. M = mean; SD = standard deviation. All scores on validated scales.

---

APA 7th Figure Format in Detail

Caption format:

Figure X
[Italicised title, below figure, sentence case]
[Description. Error bars represent 95% CI. * p < .05.]

Example:

Figure 3
Mean Anxiety Scores by Group and Time Point
Error bars represent 95% confidence intervals. * p < .05 compared to control group.

---

Error Bar Decision Guide

Choosing the wrong error bar type is one of the most common figure errors.

| Error Bar Type | What It Shows | When to Use | |---|---|---| | Standard Deviation (SD) | Spread of the data — variability within the sample | Describing the distribution; showing how spread out the data are | | Standard Error of the Mean (SEM) | Precision of the mean estimate | Comparing means — smaller bars make groups look more distinct (can be misleading) | | 95% Confidence Interval (CI) | Range likely to contain the true population mean | Statistical inference — overlapping CIs do NOT directly indicate non-significance |

Recommendation hierarchy:

1. 95% CI is preferred for inferential graphs — directly communicates uncertainty
2. SD is preferred for descriptive graphs — shows data variability
3. SEM is least informative for most purposes — often used to make results look more precise than they are

Critical rule: ALWAYS state the error bar type in the caption. "Error bars represent ±SD" is mandatory. Unlabelled error bars are a common reason for rejection.

For small samples (n < 10): Consider showing individual data points (jitter plot, beeswarm plot) instead of or alongside summary statistics.

---

Color Accessibility

Approximately 8% of males and 0.5% of females have colour vision deficiency. Use accessible palettes.

Wong 8-Color Palette (Recommended)

Optimised for colourblind accessibility:

| Colour Name | Hex Code | RGB | |---|---|---| | Black | #000000 | 0, 0, 0 | | Orange | #E69F00 | 230, 159, 0 | | Sky Blue | #56B4E9 | 86, 180, 233 | | Bluish Green | #009E73 | 0, 158, 115 | | Yellow | #F0E442 | 240, 228, 66 | | Blue | #0072B2 | 0, 114, 178 | | Vermillion | #D55E00 | 213, 94, 0 | | Reddish Purple | #CC79A7 | 204, 121, 167 |

Okabe-Ito Palette (Alternative)

| Colour Name | Hex Code | |---|---| | Orange | #E69F00 | | Sky Blue | #56B4E9 | | Bluish Green | #009E73 | | Yellow | #F0E442 | | Blue | #0072B2 | | Vermillion | #D55E00 | | Reddish Purple | #CC79A7 | | Black | #000000 |

ColorBrewer Recommendations

For sequential data (light to dark of one colour):

• Blues, Greens, Oranges, Purples — all colourblind-safe sequential palettes

For diverging data (two extremes):

• RdYlBu, PuOr, BrBG — most are colourblind-safe

For qualitative data (distinct categories):

• Set2, Paired — generally accessible; avoid pure red/green combinations

Testing tool: Use the Coblis colour-blindness simulator (colblindor.com/coblis) to preview figures before submission.

Additional accessibility rules:

• Never rely on colour alone — use different line styles (solid, dashed, dotted) or point shapes (circle, triangle, square) alongside colour
• Minimum font size: 8pt at final print size (check after scaling)
• Sufficient contrast: Check with WebAIM contrast checker for text on figure backgrounds

---

Journal-Specific Figure Requirements

| Journal | Resolution | File Formats | Width Limits | Other | |---|---|---|---|---| | Nature / Nature family | 300 dpi (halftone); 600 dpi (line art) | TIFF, EPS, PDF | 89mm (1 col); 183mm (2 col) | Max file size 10 MB per figure | | Science | 300 dpi minimum | TIFF, EPS, PS | 5.5 cm (1 col); 12 cm (2 col) | Individual panels labelled A, B, C | | Cell / Cell Press | 300 dpi | TIFF, EPS | 85mm (1 col); 170mm (2 col) | No white space around figure | | APA Journals | 300 dpi minimum | TIFF or EPS | Varies | Captions in manuscript text, not in file | | IEEE | 300 dpi | EPS, TIFF | Column width | Fonts embedded; no bitmapped fonts | | PLOS ONE | 300 dpi (300–600 typical) | TIFF, EPS | 8.3 cm (1 col); 17.4 cm (2 col) | Alt text required | | Elsevier | 300 dpi (600 for sharp line art) | TIFF, EPS, PDF | Varies by journal | Separate figure files |

Check the target journal's "Instructions for Authors" first — these specifications override general guidelines.

---

CONSORT Flow Diagram

Required for all randomised controlled trials (RCTs). Four mandatory stages:

ENROLLMENT
├── Assessed for eligibility (n = XX)
│   ├── Excluded (n = XX)
│   │   ├── Did not meet inclusion criteria (n = XX)
│   │   ├── Declined to participate (n = XX)
│   │   └── Other reasons (n = XX)
│   └── Randomised (n = XX)
│
ALLOCATION
├── Allocated to intervention (n = XX)
│   └── Received allocated intervention (n = XX)
│   └── Did not receive (n = XX) [reasons]
└── Allocated to control (n = XX)
    └── Received allocated intervention (n = XX)
    └── Did not receive (n = XX) [reasons]
│
FOLLOW-UP
├── Lost to follow-up intervention (n = XX) [reasons]
├── Discontinued intervention (n = XX) [reasons]
└── Lost to follow-up control (n = XX) [reasons]
│
ANALYSIS
├── Analysed (intervention) (n = XX)
│   └── Excluded from analysis (n = XX) [reasons]
└── Analysed (control) (n = XX)
    └── Excluded from analysis (n = XX) [reasons]

Create using: PowerPoint, draw.io, consort-statement.org flowchart tool, or TikZ in LaTeX.

---

PRISMA 2020 Flow Diagram

Required for systematic reviews and meta-analyses. Four stages:

IDENTIFICATION
├── Records from databases (n = XX)
├── Records from registers (n = XX)
└── Records removed before screening:
    └── Duplicates (n = XX); ineligible (n = XX)

SCREENING
├── Records screened (n = XX)
└── Records excluded (n = XX)

ELIGIBILITY
├── Reports assessed for eligibility (n = XX)
└── Reports excluded (n = XX) [reason and n for each]

INCLUDED
└── Studies included in review (n = XX)
    └── Reports of included studies (n = XX)

PRISMA 2020 also has a separate section for studies identified through other methods (citation searching, etc.).

Official PRISMA flow diagram template: prisma-statement.org

---

Statistical Figure Requirements

Showing Individual Data Points

For small samples (n < 20 per group), consider:

• Strip/jitter plot — individual points scattered along the x-axis
• Beeswarm plot — points spread to avoid overlap
• Raincloud plot — combines half violin + box plot + jitter

Reporting requirement: n per group must be visible somewhere on the figure, in the caption, or in the legend.

Bar Chart vs. Other Chart Types

| Chart Type | Best For | Avoid When | |---|---|---| | Bar chart | Counts, frequency, totals | Showing continuous distributions | | Box plot | Showing median, IQR, outliers | Very small samples (n < 6) | | Violin plot | Distribution shape + summary | Very large files; overlapping distributions | | Line graph | Change over time; continuous x | Categorical x-axis | | Scatter plot | Two continuous variables; correlation | When showing group means | | Forest plot | Meta-analysis effect sizes | Single studies | | Heatmap | Correlation matrices; large datasets | Small datasets |

Avoid "dynamite plots" (bar + error bar) for continuous data where the distribution matters — they hide the data. Prefer box plots, violin plots, or jitter plots.

---

Common Figure Mistakes

| Mistake | Problem | Fix | |---|---|---| | No error bar label | Ambiguous (SD? SEM? CI?) | Add "Error bars represent ±SD" to caption | | Inaccessible colour palette | Red/green indistinguishable for 8% of readers | Switch to Wong or Okabe-Ito palette | | Axes not labelled with units | Reader cannot interpret scale | Add label with unit: "Reaction time (ms)" | | Bar chart with floating baseline | Non-zero y-axis distorts magnitude | Start y-axis at zero for bar charts | | Panel labels inconsistent | A, B, C vs. a, b, c vs. (A), (B), (C) | Pick one style and use consistently | | Low resolution image | Blurry at print size | Regenerate at 300–600 dpi | | Caption as single word | "Figure 1. Results." is uninformative | Write complete descriptive caption | | Too many significant figures | "Mean = 12.3456789" | Round to appropriate precision (2 d.p. typical) |

---

Tufte Data-Ink Ratio Principle

Edward Tufte's principle: maximise the ratio of data ink to total ink. Every mark on the figure should convey information.

Eliminate:

• Chartjunk (decorative grids, 3D effects, shadows)
• Redundant axes labels
• Unnecessary gridlines (if present, make light grey)
• Excessive tick marks
• Borders around entire plot area
• 3D bar and pie charts (always — they distort perception)

A good figure presents maximum information with minimum graphical elements.

---

Pre-Submission Figure and Table Checklist

• [ ] All figures and tables numbered sequentially
• [ ] Captions: figures captioned below, tables captioned above
• [ ] All axes labelled with units
• [ ] Error bars defined
• [ ] Colour palette is colourblind-accessible
• [ ] Font sizes readable at print size
• [ ] Resolution meets journal requirements
• [ ] File formats match journal requirements
• [ ] CONSORT / PRISMA flow diagrams included (if applicable)
• [ ] Individual data points shown for n < 20
• [ ] Table notes explain all abbreviations
• [ ] No vertical lines in tables
• [ ] No 3D chart effects
• [ ] All statistical significance markers explained

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: muhammad1438
• Source: muhammad1438/academic-writer-skills
• License: MIT

Install and usage instructions live in the source repository linked above.