# Supported analyses

> The research workflows SciGround supports: group comparisons, paired and multi-group designs, rank comparisons, several outcomes, bulk RNA-seq, equipment-failure missing data, fixed batches, replications, two planned decisions, and exploration followed by confirmation.

## Your research question. A supported analysis.

Find a workflow by the comparison, measurements and study structure you have. Each route states what it accepts and which claims it can support; the design determines eligibility.

## Choose the question your study asks.

### Groups and measurements

#### Compare two group means.

Treatment versus control in two independent groups. Student’s t-test runs from input matching to the supported claim and a shareable Record—or Welch’s t-test when you don’t want to assume equal variances, with its approximation shown separately.

#### Compare before and after.

The same animals or samples measured at two time points. Pairs are matched by ID, never by row order, and the differences are computed from the original values before a paired t-test.

#### Compare three or more groups.

Control plus several treatments in one study. A one-way ANOVA answers “is there any difference?” and pre-specified comparisons, with a correction for making several of them, answer “which planned comparisons differ?”.

#### Compare groups using ranks.

When the question is about ranks rather than means, the Mann–Whitney U test compares two independent groups and states exactly which hypothesis it tested.

#### Evaluate several measurements.

Test all your planned outcomes together with a pre-specified correction, so one lucky result can’t pass as a finding. Every outcome stays in the report, significant or not.

### Counts, batches and source differences

#### Analyze bulk RNA‑seq expression.

From a gene-level count table and sample sheet to differential-expression results through one fixed pipeline. Filtering, normalization, testing, and false-discovery control stay traceable—and are rechecked after a correction.

#### Account for measurements lost to equipment failure.

Keep the planned samples and failure reasons on record and analyze the values actually measured, with the sample count handled correctly.

#### Compare treatment and control across three fixed batches.

Estimate one treatment effect across three experiment days, each day counting equally, with every day-by-group cell visible.

#### Compare an original study with a replication.

Compare a published summary—for example, values read from a figure—with your replication’s individual data, keeping each source’s type and precision.

### Planned decisions and confirmation

#### Plan two opportunities to continue or stop.

Run a two-stage study with one pre-fixed interim decision while keeping the overall false-positive rate under control.

#### Explore, then confirm.

Explore freely, lock in one candidate, and test it on independent confirmation data.

## Start with how the observations relate.

| Your study | Look for |
| --- | --- |
| Different individuals in two groups | An independent-group comparison. |
| The same individuals measured twice | An ID-matched paired comparison. |
| Several groups or planned outcomes | The relevant group comparison and multiplicity rules. |
| Counts, batches, replication or staged decisions | The corresponding fixed workflow and its input conditions. |

This is a route-finding guide. Eligibility still depends on the complete design and inputs; a familiar test name is not enough.

## What accompanies a supported result

- Results tied to the locked design and approved input
- The supported claim, with assumptions and limits
- A Record that others can verify
- Rechecked results after an approved correction

## Limits and conditions

Each workflow has a defined input range and set of claims. See the boundaries workflow by workflow. [Read the limits](https://www.licklider.ai/sciground/analyses/limits/) ([Markdown](https://www.licklider.ai/sciground/analyses/limits.md)).

## Related

- [Integrations](https://www.licklider.ai/integrations/)
- [Research scope guide](https://www.licklider.ai/docs/sciground-scope/)
- [Analyze and verify](https://www.licklider.ai/sciground/analysis/)