50 ChatGPT Prompts for Quantitative Data Analysis

Quantitative data analysis demands both statistical rigour and clear interpretive thinking: choosing the right method, running it correctly and translating the output into findings that actually mean something. The right ChatGPT prompts for quantitative data analysis help analysts, researchers and students select methods, check assumptions, interpret outputs, diagnose problems and explain results to any audience.

Important: always verify statistical outputs and interpretations independently. AI can make arithmetic errors or suggest methods that do not fit your data without flagging the limitation, so run the actual analysis in statistical software.

Below are 50 prompts across five stages: study design and sampling, data preparation and exploration, statistical tests and modelling, survey and experimental data, and visualisation and reporting. They work in ChatGPT or any other AI chatbot.

Statistical advice is only as good as the context you provide. Include:

• Your research question: what you want to know and why
• Your variables: names, types (continuous, categorical, ordinal) and roles
• Your design and sample: sample size, how data was collected and any grouping or repeated measures
• Your tools: R, Python, SPSS, Stata or Excel, so instructions and code match

Model choice matters too. GPT-6 Astra handles multi-step statistical reasoning well, and Claude Sonnet 5 tends to be careful about assumptions and caveats when interpreting results.

Good analysis starts before data collection. GPT-5.6 Sol is good at structured analysis plans, and the AI article summarizer condenses papers when you are reviewing how others designed similar studies.

1. Statistical method selector

My research question is [question]. My outcome variable is [variable and type], my predictors are [variables and types], my sample size is [n] and the design is [independent groups / repeated measures / observational]. Recommend the most appropriate statistical method, explain why, list its assumptions and suggest an alternative if those assumptions are not met.

2. Hypothesis test framework builder

Help me turn this research question into testable hypotheses: [question]. Write the null and alternative hypotheses, say whether the test should be one- or two-tailed, and define what result would support each.

3. Power analysis and sample size

I plan to use [test] to detect an effect of about [expected effect size] with alpha of [0.05] and power of [0.80]. Explain how to estimate the required sample size, walk me through the calculation and tell me which tool or package I should use to confirm it.

4. Sampling strategy

I want to study [population] to answer [question]. Compare random, stratified, cluster and convenience sampling for my situation, and explain the bias risks of each.

5. Operationalising variables

Help me turn these concepts into measurable variables: [concepts]. For each, suggest how to measure it, the level of measurement and possible validity concerns.

6. Research design comparison

Compare an experimental, quasi-experimental and observational design for studying [question]. Explain what each can and cannot tell me about causation.

7. Confounder identification

I am studying the effect of [exposure] on [outcome]. List likely confounders, mediators and colliders, and explain how to handle each in the design or analysis.

8. Measurement scale choice

I need to measure [construct] in a survey. Compare using a single item, a Likert scale and a validated multi-item scale, and explain how each choice affects the analysis later.

9. Pre-registration outline

Help me draft a pre-registration for my study on [topic]: hypotheses, design, sample size and justification, variables, the planned analysis, exclusion criteria and how I will handle missing data.

10. Analysis plan

Write a step-by-step analysis plan for this study: [describe design, variables and questions]. Include data cleaning, descriptive statistics, the main tests, checks for assumptions and planned robustness analyses.

Exploration reveals problems before they distort your results. Gemini 3 Pro can work with long descriptive outputs in one go, and Claude Haiku 4.5 is quick for recoding and data dictionary tasks.

11. Exploratory data analysis guide

Here is a description of my dataset: [variables, types, n]. Give me a step-by-step exploratory analysis plan — distributions, summary statistics, relationships and data quality checks — and the code for each step in [R / Python].

12. Missing data strategy

My dataset has missing values in [variables] at roughly [percentages]. Explain how to assess whether data is missing completely at random, at random or not at random, and compare deletion, single imputation and multiple imputation for my case.

13. Outlier diagnosis

I found possible outliers in [variable]: [describe values]. Explain how to decide whether they are errors, genuine extreme values or influential points, and what to do in each case.

14. Assumption checks

List the assumptions of [test or model] and explain how to check each one with plots and formal tests, what violations look like, and what to do if an assumption fails.

15. Transformation choice

My variable [variable] is heavily skewed. Compare log, square-root and other transformations, explain when to use each, and how transforming changes the way I interpret results.

16. Descriptive statistics table

Design a descriptive statistics table for my variables: [list]. Suggest which statistics to report for each variable type and how to present them for publication.

17. Variable recoding

Help me recode these variables for analysis: [describe]. Suggest sensible categories, reverse-coding where needed and the code to do it in [tool].

18. Scale reliability

My scale has [number] items measuring [construct]. Explain how to assess its reliability with Cronbach’s alpha or omega, how to interpret the result and what to do if reliability is low.

19. Data dictionary

Create a data dictionary for these variables: [list]. Include the variable name, description, type, units, allowed values and how missing values are coded.

20. Correlation matrix interpretation

Here is my correlation matrix: [paste]. Highlight the strongest relationships, flag possible multicollinearity and explain what these correlations can and cannot tell me.

Choosing and interpreting tests correctly is where most mistakes happen. DeepSeek V4 Pro is useful for working through statistical reasoning step by step, and the AI math calculator helps you check formulas by hand.

21. Statistical output interpreter

Here is my statistical output from [software]: [paste]. Explain what each number means, whether the result is statistically and practically significant, how to report it correctly and what I should not conclude from it.

22. Regression analysis coach

I am running a [linear / multiple] regression with [outcome] and predictors [list]. Walk me through model building, checking assumptions, interpreting coefficients and R-squared, and spotting overfitting.

23. T-test or ANOVA?

I am comparing [outcome] across [number] groups. Tell me whether to use a t-test, one-way ANOVA or another test, explain the post-hoc tests needed and how to report the results.

24. Chi-square test

Here is my contingency table: [paste]. Explain whether a chi-square test or Fisher’s exact test is appropriate, how to interpret the result and which effect size to report.

25. Non-parametric alternatives

My data violates the normality assumption for [test]. Suggest the right non-parametric alternative, explain what it tests differently and how to interpret and report it.

26. Logistic regression interpretation

Here is my logistic regression output: [paste]. Explain the odds ratios in plain language, how to judge model fit and common mistakes in interpreting logistic results.

27. Interaction effects

I found a significant interaction between [variable A] and [variable B] on [outcome]. Explain what it means, how to probe it with simple slopes and how to plot it clearly.

28. Mixed and multilevel models

My data has [students within schools / repeated measures within people]. Explain when I need a mixed or multilevel model, how to choose random effects and how to interpret the output.

29. Time series basics

I have [frequency] data on [variable] over [period]. Explain how to check for trend, seasonality and autocorrelation, and which modelling approaches suit my goal of [forecasting / explaining change].

30. Effect sizes and confidence intervals

For my [test] result of [paste], calculate or explain the appropriate effect size and confidence interval, and help me describe the practical meaning beyond the p-value.

Survey and experimental data come with their own pitfalls. The AI summarizer helps you get a first overview of long lists of open-ended responses before you code them.

31. Survey data analyst

I have survey data with [number] respondents and these question types: [describe]. Recommend an analysis approach for each question type, how to handle incomplete responses and the key comparisons worth making.

32. Likert scale analysis

Explain how to analyse Likert-scale data from my survey on [topic]: when to treat items as ordinal or interval, how to combine items into a scale, and how to present results visually.

33. Survey weighting

My survey sample over-represents [group] compared with the population. Explain how weighting works, which method suits my situation and how weighting affects standard errors.

34. A/B test analysis

Here are my A/B test results: [variant, sample size, conversions]. Tell me whether the difference is statistically significant, the confidence interval for the lift, and whether I ran the test long enough to trust it.

35. Cross-tabulation

Help me create cross-tabulations of [variable] by [demographic groups]. Suggest which comparisons are meaningful, which statistics to report and how to avoid over-interpreting small cells.

36. Quantifying open-ended responses

Here are open-ended responses from my survey: [paste]. Propose a coding scheme, code each response and count the frequency of each theme, noting any responses that fit more than one code.

37. Factor analysis

I want to check whether my [number] survey items measure [number] underlying factors. Explain when to use exploratory or confirmatory factor analysis, how to decide the number of factors and how to interpret loadings.

38. Repeated measures and longitudinal data

I measured [outcome] at [number] time points for the same participants. Compare repeated-measures ANOVA and mixed models for my data, and explain how to handle participants with missing time points.

39. Attrition analysis

[Percentage] of participants dropped out of my study. Explain how to test whether dropouts differ from completers, how attrition might bias my results and how to address it.

40. Multiple comparisons correction

I am running [number] tests on the same dataset. Explain the risk of false positives, compare Bonferroni, Holm and false discovery rate corrections, and recommend one for my situation.

Results only matter if others can understand and trust them. Claude Sonnet 4.6 is dependable for analysis code, the AI PDF summarizer helps you compare your reporting with published papers, and the AI grammar checker polishes the final write-up.

41. Data visualisation advisor

I want to show [finding] to [audience]. Recommend the best chart type, what to put on each axis, how to show uncertainty and common design mistakes to avoid, with code in [R / Python].

42. Statistical error diagnostician

Review my analysis for common errors: [describe design, tests used and conclusions]. Check for p-hacking risks, violated assumptions, confusing correlation with causation, pseudoreplication and overgeneralisation.

43. Non-technical findings communicator

Explain these results to [non-technical audience] in plain language: [paste findings]. Lead with what it means for them, give one relatable example, and state the uncertainty honestly without jargon.

44. Results section writer

Help me write the results section for these analyses in [APA / journal] style: [paste outputs]. Report test statistics, degrees of freedom, p-values, effect sizes and confidence intervals correctly.

45. Limitations section

Based on my study design — [describe] — help me write an honest limitations section covering sampling, measurement, design and generalisability, and suggest how future research could address each.

46. Analysis code generator

Write [R / Python] code to run [analysis] on a dataset with these columns: [list]. Include assumption checks, the main model, a summary table and a plot, with comments explaining each step.

47. Reproducibility checklist

Create a reproducibility checklist for my analysis project: data versioning, code organisation, random seeds, package versions, documentation and what to share with others.

48. Peer review of an analysis

Act as a critical peer reviewer. Here is my methods and results section: [paste]. List the questions a reviewer would ask, weaknesses in the analysis and suggestions to strengthen it.

49. Publication-ready tables

Format these results into a publication-ready table: [paste]. Suggest the column layout, the number of decimal places, how to mark significance and a clear table note.

50. Executive summary of findings

Turn my analysis into a one-page executive summary for [decision-makers]: the question, key findings with the most important numbers, how confident we can be and the recommended action.

Chat Smith gives you ChatGPT alongside Claude, Gemini, Grok and DeepSeek in one app, so you can plan an analysis with one model and double-check the interpretation with another. Save the prompts you use most as templates for every new dataset.

Use Gemini 3.5 Flash for quick questions while you work. AI can sharpen your statistical thinking, but the final analysis, checks and conclusions should always come from your own verified work.