How to Cite Data Analysis in Academic Papers

If you ran a statistical analysis for a thesis, research paper, or class assignment, you need to cite it. Some authors cite the data source but forget to cite the analytical tool: the software and method that produced the results.

This guide shows one made-up example and its citation in four example layouts, labeled APA, MLA, Chicago, and BibTeX.

Worked example

This example uses made-up scores. It asks if scores differ between a prep group and a no-prep group. NumPy makes the two sets of scores. The script builds in a gap: the prep scores are drawn with a mean of 72 and the no-prep scores with a mean of 66, a 6 point gap. The method is Welch's two-sample t-test, a way to compare two groups when their score spread may differ. The code and output show each step. Readers can run the same test.

import numpy as np
from scipy import stats

# Simulated data, seed 42. Two small sets of scores on a made-up scale.
rng = np.random.default_rng(42)
prep = rng.normal(loc=72, scale=8, size=12)
no_prep = rng.normal(loc=66, scale=9, size=11)

m1, m0 = prep.mean(), no_prep.mean()
s1, s0 = prep.std(ddof=1), no_prep.std(ddof=1)
t, p = stats.ttest_ind(prep, no_prep, equal_var=False)
se2 = s1**2 / len(prep) + s0**2 / len(no_prep)
df = se2**2 / (
    (s1**2 / len(prep))**2 / (len(prep) - 1)
    + (s0**2 / len(no_prep))**2 / (len(no_prep) - 1)
)
ci_low, ci_high = stats.t.interval(
    0.95, df, loc=m1 - m0, scale=np.sqrt(se2)
)
sp = np.sqrt(
    ((len(prep) - 1) * s1**2 + (len(no_prep) - 1) * s0**2)
    / (len(prep) + len(no_prep) - 2)
)
d = (m1 - m0) / sp

print("Question: In this simulated sample, do prep scores differ "
      "from no-prep scores?")
print("Input: simulated scores, seed 42; "
      "prep n={}, no-prep n={}".format(len(prep), len(no_prep)))
print("Method: Welch two-sample t-test; Cohen's d uses pooled SD")
print("Prep: M={:.2f}, SD={:.2f}".format(m1, s1))
print("No prep: M={:.2f}, SD={:.2f}".format(m0, s0))
print("Welch: t({:.2f})={:.2f}, p={:.4f}, mean difference={:.2f}, "
      "95% CI [{:.2f}, {:.2f}], d={:.2f}".format(
          df, t, p, m1 - m0, ci_low, ci_high, d))
print("Result: the script built in a 6 point gap. With 12 and 11 scores "
      "the run found 3.72 with an interval from -2.59 to 10.04, so a small "
      "made-up sample cannot rule out no difference. The result is inconclusive "
      "for this small sample, not evidence of no gap.")
print("Limit: the small made-up sample does not support a claim about "
      "real students.")
print('\nAPA: MCP Analytics Team. (2026). Simulated score comparison: '
      'Prep and no prep (seed 42) [Statistical report]. MCP Analytics.')
print('MLA: MCP Analytics Team. "Simulated Score Comparison: '
      'Prep and No Prep (Seed 42)." MCP Analytics, 2026.')
print('Chicago: MCP Analytics Team. "Simulated Score Comparison: '
      'Prep and No Prep (Seed 42)." Statistical report. 2026.')
print("BibTeX: @misc{mcp2026sim, author={{MCP Analytics Team}}, "
      "title={Simulated score comparison: Prep and no prep (seed 42)}, "
      "year={2026}, note={Statistical report, simulated data}}")

Printed output (copied from a run with python3 -I):

Question: In this simulated sample, do prep scores differ from no-prep scores?
Input: simulated scores, seed 42; prep n=12, no-prep n=11
Method: Welch two-sample t-test; Cohen's d uses pooled SD
Prep: M=70.87, SD=7.67
No prep: M=67.14, SD=6.90
Welch: t(21.00)=1.23, p=0.2337, mean difference=3.72, 95% CI [-2.59, 10.04], d=0.51
Result: the script built in a 6 point gap. With 12 and 11 scores the run found 3.72 with an interval from -2.59 to 10.04, so a small made-up sample cannot rule out no difference. The result is inconclusive for this small sample, not evidence of no gap.
Limit: the small made-up sample does not support a claim about real students.

APA: MCP Analytics Team. (2026). Simulated score comparison: Prep and no prep (seed 42) [Statistical report]. MCP Analytics.
MLA: MCP Analytics Team. "Simulated Score Comparison: Prep and No Prep (Seed 42)." MCP Analytics, 2026.
Chicago: MCP Analytics Team. "Simulated Score Comparison: Prep and No Prep (Seed 42)." Statistical report. 2026.
BibTeX: @misc{mcp2026sim, author={{MCP Analytics Team}}, title={Simulated score comparison: Prep and no prep (seed 42)}, year={2026}, note={Statistical report, simulated data}}

Result: The script built in a 6 point gap. With 12 and 11 scores the run found 3.72 with an interval from -2.59 to 10.04, so a small made-up sample cannot rule out no difference. The result is inconclusive for this small sample, not evidence of no gap. This small sample says nothing about real students. Cohen's d here uses the pooled SD.

Example layouts for citing it

The four citation lines in the output above are example layouts that this page made up. They are not checked against any style guide, and this page does not say they follow MLA, Chicago, APA, or any other style. Check each one against the style guide you use.

Sources

• MLA Style Center, "How to Cite an Online Work": https://style.mla.org/works-cited/citations-by-format/online-works/
• Chicago Manual of Style, "Chicago-Style Citation Quick Guide": https://www.chicagomanualofstyle.org/tools_citationguide.html
• BibTeX @misc entry type, as described at bibtex.org: https://www.bibtex.org/EntryTypes/Misc/

These guides exist at these addresses. Check the rules for the journal, school or publisher that will get the paper.

What to cite in a statistical analysis

A complete methodology citation includes three things:

1. The statistical method: name the test and its variant (e.g., "Welch's two-sample t-test," not just "t-test")
2. The software: what tool ran the analysis (R, SPSS, Python, MCP Analytics)
3. The specific report or output, if your analysis generated a shareable report, cite it with a URL so reviewers can verify

Many reports give the test statistic, degrees of freedom, p-value, and effect size. The citation goes in your methods section alongside these numbers. Check what your journal asks for.

Example layout labeled APA

Some fields use APA. Here is an example layout for citing a statistical analysis:

In your methods section, write something like this (numbers from the made-up example above):

A Welch two-sample t-test compared simulated scores for a prep group (M = 70.87, SD = 7.67, n = 12) and a no-prep group (M = 67.14, SD = 6.90, n = 11), t(21.00) = 1.23, p = .2337, mean difference = 3.72, 95% CI [-2.59, 10.04], d = 0.51. The data are made up.

Example layout labeled MLA

Some fields use MLA. Here is an example layout:

Example layout labeled Chicago

Some fields and university style guides use Chicago. Here is an example layout:

Example layout labeled BibTeX

For LaTeX users, an example layout for a .bib file:

@misc{mcp2026sim,
  author={{MCP Analytics Team}},
  title={Simulated score comparison:
         Prep and no prep (seed 42)},
  year={2026},
  note={Statistical report, simulated data}
}

Example layouts for reporting common tests

Here are example layouts for reporting common statistical tests. The numbers are made up and show layout only. Check the exact format against the style guide you use.

Why citation matters

Citing your analytical tools isn't just academic formality. It enables:

• Reproducibility: reviewers can verify your results by following the same method
• Transparency: readers know exactly which test variant was used (e.g., Welch's vs Student's t-test)
• Auditability, if your report URL is cited, the methodology card shows every diagnostic test that was run
• Academic integrity: some journals ask authors to name the tools they used, so check yours

One-click citations with MCP Analytics

Every report generated by MCP Analytics includes a methodology section with a "Cite This Analysis" button. Click it, choose a layout (labeled APA, MLA, Chicago, or BibTeX), and copy the citation. Check it against your style guide.

The methodology section also shows:

• What statistical method was used and what it does, step by step
• Sample size, data exclusions, and retention rate
• Key test statistics (t-value, p-value, effect size)
• Assumptions that were checked (normality, variance equality)
• Limitations of the method

This means your report supplies the details you need to write a methods section and a citation. Check the citation format your style guide or journal requires.