What our statistics help covers ¶
Most statistics problems start at the design stage. Bring your research questions early, before you collect data.
Statistics help at Copperplate is one-to-one consulting for researchers who need to analyze their own data correctly. A statistician helps you choose a suitable test, plan your sample size, run and check the analysis with you, interpret the output, and write the results in APA style. You stay the author and the analyst of record.
This page is for people with a live research project: a master's thesis, a PhD chapter or a journal manuscript. If you want to learn statistics for a course, our statistics tutoring is built around teaching, with practice problems and explanations at your pace. If your question is about one program, see our dedicated pages for SPSS help, Stata help and R help.
We work across psychology, education, nursing and public health, biostatistics, business, economics and the social sciences. Your consultant is matched to your field, so the conventions of your discipline are familiar from the first message.
Choosing the right statistical analysis test ¶
The right test depends on your research question, how your variables are measured and how your data were collected. We start with three questions: what are you comparing or predicting, what type is each variable (continuous, ordinal, categorical), and are observations independent or repeated?
| Research question | Typical test | Common alternative |
|---|---|---|
| Do two independent groups differ on a continuous outcome? | Independent-samples t test | Welch's t test or Mann–Whitney U |
| Do three or more groups differ? | One-way analysis of variance (ANOVA) | Welch's ANOVA or Kruskal–Wallis |
| Do two factors interact? | Two-way (factorial) ANOVA | Regression with an interaction term |
| Has the same group changed over time? | Paired t test or repeated-measures ANOVA | Wilcoxon signed-rank or a mixed model |
| Are two categorical variables associated? | Chi-square test of independence | Fisher's exact test |
| Which predictors explain an outcome? | Multiple linear regression | Logistic regression for a binary outcome |
These are examples of statistical analysis choices. Real designs often need more: clustered data, covariates, missing values or a meta-analysis model. We explain why one approach fits your design better than another, so you can justify it in your methodology chapter. Our guide to one-way vs two-way ANOVA shows this reasoning in detail.
Statistics Help & Consulting
Send your file today and get a quote with a matched editor.
Power analysis and sample size ¶
Statistical power is the probability that a study detects an effect of a given size when that effect truly exists. Committees and ethics boards increasingly ask for an a priori power analysis, and reviewers question underpowered null results.
We calculate the sample size you need from four inputs: the planned test, the expected effect size, the alpha level and the target power. We base the expected effect on prior studies or a smallest effect of interest, and document every assumption.
Example
For an independent-samples t test with a medium effect (d = 0.50), two-tailed α = .05 and power of .80, you need 64 participants per group, 128 in total. A one-way ANOVA with three groups and f = 0.25 under the same settings needs 159 participants.
If your data are already collected, we can run a sensitivity analysis instead, which tells you the smallest effect your sample could reliably detect.
Running and checking your analysis together ¶
We work in the software you use: SPSS, Stata, R, jamovi, JASP or SAS. Sessions are collaborative. You see each step, and you keep the syntax or script so the analysis is reproducible.
- Data screening: coding errors, out-of-range values, duplicate cases and missing data patterns.
- Assumption checks: normality of residuals, homogeneity of variance (Levene's test), linearity, multicollinearity (VIF) and influential cases (Cook's distance).
- Model choice: robust or nonparametric alternatives when assumptions fail, and corrections for multiple comparisons.
- Verification: we re-run analyses you have already done and flag discrepancies in degrees of freedom, test choice or reported values.
A significant Levene's test does not mean your ANOVA is ruined. Welch's ANOVA with Games–Howell post hoc tests is a standard, defensible alternative.
Interpreting output and APA-style results reporting ¶
Statistical significance tells you an effect is unlikely to be zero. The effect size tells your reader whether it matters.
Software output is full of numbers you will never report. We go through each table with you, explain what every value means, and show you which ones belong in your results section.
APA 7 reporting follows consistent rules. Statistical symbols such as t, F, p and d are italicized. Degrees of freedom go in parentheses. Exact p values are given to two or three decimals, with p < .001 for anything smaller. Values that cannot exceed 1, such as p and r, take no leading zero. Every key test needs an effect size and, where possible, a confidence interval.
APA 7 results
Participants in the intervention group reported higher self-efficacy (M = 4.12, SD = 0.61) than those in the control group (M = 3.74, SD = 0.58), t(118) = 2.48, p = .015, d = 0.64, 95% CI [0.27, 1.01].
We also review your tables and figures against APA 7 table formatting, and comment on whether your discussion claims match what the results support.
How statistics consulting works ¶
- Send your materials: research questions, hypotheses, your data file (anonymized), any existing output and your university or journal guidelines.
- Receive a plan and quote: we confirm the analyses needed and the scope of support. See pricing for how quotes are built.
- Work through the analysis: live sessions or written walkthroughs, with annotated syntax or scripts.
- Get your results reviewed: your drafted results section comes back with Track Changes and margin comments explaining each correction.
- Ask follow-up questions: after delivery, you can ask your consultant about any comment, which helps when preparing for a defense or a reviewer response.
Turnaround runs from 24 hours to 7 days, depending on the size of the dataset and the number of analyses. Your files are handled confidentially, an NDA is available on request, and we delete files when you ask.
Statistics help and academic integrity ¶
We consult, explain and check. We do not produce work for you to submit as your own, and we do not complete graded homework, quizzes or exams. You run or approve every analysis and write your own results. That is also what lets you answer questions about your methods in a viva or peer review.
Universities differ on what outside statistical support is allowed, and some require you to acknowledge it. Please check your institution's rules and talk to your supervisor. Our academic integrity policy sets out exactly what we will and will not do.