Typical uses
- Results and baseline-characteristics tables from journal articles
- Supplementary data tables published as PDF
- Survey crosstabs and frequency tables from reports
- Statistical tables in government and agency publications
- Instrument or lab output embedded in a PDF report
How to extract a table from a research PDF
- Download the publisher's PDF. The version from the journal site usually has a clean text layer; a scan from a library copy may need OCR.
- Drop it on the converter on this page and keep Tables only, so prose paragraphs are ignored.
- Choose Split to get each table on its own sheet. That is usually right for papers, where Table 1, Table 2 and so on have different columns.
- Convert, download and rename the sheets to match the table numbers in the paper.
Cleaning research tables for analysis
Published tables are made for reading, not analysis. Common fixes:
- Combined cells. "45.2 (12.1)" (mean and SD) or "1.23 (0.98–1.54)" (estimate and CI) arrive as text in one cell. Split them with Data → Text to Columns using "(" and "–" as delimiters, or with
TEXTBEFORE/TEXTAFTER. - Significance markers. Asterisks and superscript letters (e.g. "0.031*") keep the cell as text. Remove them with Find & Replace before converting to numbers.
- P-values like "<0.001". These are not numbers. Keep them as text, or record a separate flag column.
- Multi-level headers. Grouped headers ("Treatment → n, %") become two header rows. Combine them into single column names before importing into a stats package.
- Footnote rows under the table come through as rows; delete them.
Exporting to statistical software
Once the sheet has one header row and one observation per row, use File → Save As → CSV (UTF-8). Then:
- R:
read.csv("table1.csv")orreadxl::read_excel("tables.xlsx", sheet = 1) - Python:
pandas.read_excel("tables.xlsx", sheet_name=None)loads every sheet at once - SPSS / Stata: import the CSV or Excel file directly
Accuracy and citation
Always check extracted values against the PDF, especially for tables you will re-analyse or cite. Pay attention to decimal alignment and negative signs, and record the source table and page in your notes.
For older papers that only exist as scanned images, see scanned PDF to Excel. OCR currently reads English text.