Help & user manual

New to Chavery?

The welcome tour takes two minutes and shows what Chavery does and where to start. You can take it again whenever you like.

The user manual

Sign-in, planning, the Master Chart, choosing a statistical test, writing checks, references and building the final document — the manual covers each in a couple of pages, using the same screen and button names you see in the app.

What has been added since the manual was printed

The PDF above was written before the features below. Each says where to find it and what Chavery does — and does not do — with your data.

Statistical tests, checked on sample data

Every test on the Analyse screen was run on a worked sample study and compared with an independent calculation (scipy, statsmodels, scikit-learn or the published formula): one-sample, paired, independent and Welch t-tests, Wilcoxon, Mann-Whitney, one-way and repeated-measures ANOVA, Kruskal-Wallis, Friedman, Pearson and Spearman correlation, chi-square, Fisher's exact, McNemar, linear and logistic regression, Cronbach's alpha, ICC, Cohen's kappa, diagnostic accuracy, ROC, Tukey post hoc and structural equation models. Each agreed with its reference, and are re-checked whenever anything changes.

Checking a test's assumptions before you run it

On Analyse, once you have chosen a test and its columns, Chavery shows what your data say about that test's assumptions before you run it: for each group, the Shapiro-Wilk result, the skewness and kurtosis, and a normal quantile plot — because a p-value alone is a poor judge of normality. If your data argue against the test you chose, Chavery shows the usual alternative and you decide: use it, or keep your test and write why it is still appropriate. A test the evidence argues against is not saved without that reason. The reason is stored with the analysis and printed in your results chapter beside the checks, so an examiner can judge it. Nothing is decided for you, and nothing in your data is changed.

Yes/No answers, blanks and spellings

• A yes/no answer against a number (for example menstrual irregularity against BMI): when the test chooser asks what kind of values you are relating, pick One is a yes/no and the other is a number to compare the number between the yes and no groups, or pick A yes/no outcome under prediction to run logistic regression. “Yes”, “Y”, “Positive”, “1” and their opposites are all read as yes and no. If a column has two answers that are neither, Chavery names both and asks you to correct it instead of guessing which is the outcome.
• A blank cell is missing, not a category. It is left out of the counts; it never becomes an extra row in a table.
• “Yes”, “yes” and “YES ” are one answer. Spellings that differ only by capitals or spaces are counted together, and in a yes/no column the yes words are joined and the no words are joined. Anywhere else “Y” is never merged with “Yes” — only you can say they mean the same. Every merge is listed with the result, and your Master Chart is never changed.

Sample size

The 18 calculators on the Sample size card each return the smallest sample at which the study reaches the power you asked for, and each is checked by working out the power of the real test at that size by a second route. The designs analysed by a t-test are sized with the exact t distribution, not the normal approximation, which recruits one to three too few in a small study; the textbook figure is shown beside the answer, together with the formula and its assumptions. If you leave alpha out, each method uses its own (0.05, but one-sided 0.025 for non-inferiority); on screen the box is filled in for you. Impossible inputs — a zero difference, a power of 1.2 — are refused with the reason, not answered.

Structural equation modelling

On Analyse, under an objective, choose Build a structural equation model: name each construct and the columns that measure it, say what predicts what, and optionally free a pair of items that share more than the construct. You get fit indices, loadings, paths, a path diagram and a written sentence; ask for suggestions and Chavery lists which covariances the data are asking for but adds none. Save it, confirm the objective, and the fit table, loadings, paths and diagram appear in the Results chapter. Part of the ₹18,000 plan.

Images, figures and the CONSORT diagram

• Insert image on the Write toolbar puts a picture at the cursor (or paste one with Ctrl+V; in the Cards view too). Chavery keeps a copy of at most 72 KB for the screen.
• When you build the Word or PDF file, Print-quality images asks for the original of each image, one at a time, beside the small copy it replaces. Each is resized to 300 dpi at the width it prints, used for that build, and not kept. You can print the small copy instead; it prints soft.
• Graphs and diagrams from your analyses — including the ROC curve, the repeated-measures line plot, the diagnostic accuracy chart and the CONSORT flow of participants — reach the built document. Anything left out is named before the build.

References and citation numbers

Cite a reference on the Write screen puts a numbered marker at the cursor. Numbers follow first appearance across the whole document, so citing something early renumbers what follows; Chavery can find numbers that a later citation moved and fix them on request. When an assistant you connected proposes text, its references arrive with it, so cited works reach the reference list.

An AI assistant you connect

Suggested chapter text and references wait under Waiting for you. Open one with Read it, compare it with what you wrote, and accept or leave it. Nothing is in your document until you accept, the old text can be copied first, and the numbers, tests and sentences in your Results always come from a formula or a written rule, never from the assistant.

Guides and students

• A guide opens on Supervision. If you supervise students, or an administrator has given you the Chavery Guide product (so you can invite your first student), signing in takes you to your guide dashboard, even if you also have your own research. The sidebar shows Supervision, Learn, Examples, Help and Settings.
• The page opens on one thing to do. Do this next names it — a methodology decision first, because the student can do nothing else until it is decided, then the chapter that has waited longest — and has the only blue button. Under it, Waiting for you lists everything that is waiting. With nothing waiting it says so, and names anyone who has not worked for three weeks. Reviewing a student, the blue button is the decision you opened the page for; open a chapter from the list and that chapter's Approve chapter is blue, open the student from their tile and Approve methodology is.
• One tile per student shows their project, its stage, how many sections you have approved, what is waiting on you, and when they last worked on it. The tile says Needs you (a section or the methodology is waiting for your decision), With the student (you sent something back or left a comment they have not resolved), Quiet (nothing waiting either way and no change for 21 days or more), On track or Completed. The edge is amber for Needs you and Quiet — the two to look at — grey for With the student, and green for On track and Completed. Tiles that need you come first. A tile never shows the student's text; open it to see what they have assigned you.
• Your own research is one switch away: use Working in at the top of the sidebar and choose My research. There, a My guide tile shows your own guide, whether anything from them needs you, and when they last replied.
• As a student the same My guide tile is on your dashboard. With no guide yet it offers Assign a guide.

Assistants: your rates

An approved assistant signs in to Assistance work. Under Your rates they set what they usually charge for an article and for a dissertation, in whole rupees before GST. That rate fills in their proposal for each student, which they can change before sending. It is a default, not a price: every charge is still proposed by the assistant, approved by an administrator and accepted by the student, and the student never sees the assistant's rate. Chavery keeps 15% of the assistance charge and pays the assistant the other 85% once the student has paid.

IQAC Coordinator, Research Dean and Research Coordinator

Each opens on their own dashboard, with only that dashboard, Help and Settings in the sidebar. All three are read-only and show counts, dates, names and statuses — never a researcher's objectives, data, results or written chapters. Every figure has its definition under What these figures count.

• IQAC Coordinator — Quality & compliance. Is the process being followed? Projects with a required milestone overdue, a submission date passed, no guide, no milestone plan or no activity for 30 days or more; sections approved and waiting for a guide; all by department, with the projects that need attention named. Download summary (CSV) saves the department table for a report.
• Research Dean — Research output. How much research is there, and where? Projects by stage and department, active researchers, completion, submissions already late or due in the next 90 days, each guide's workload, milestones due in the next 30 days, and how many are already overdue.
• Research Coordinator — University. The colleges in the university, its roster, and the submissions waiting for the coordinator's decision, with a count of each at the top.
• An institution administrator gives someone the IQAC Coordinator or Research Dean role; an administrator can open both dashboards too.

Making an article from your dissertation

In the project, open More about this project on its Overview and choose Turn into an article. Tick the chapters the article keeps, and drag them (or use the arrows) into the order it should read. Then write, optionally ask a connected assistant to rewrite, approve, Build the article as Word, and continue to Publish to choose a journal. Only objectives you have confirmed appear in its results.

Inside a project: the tabs and what to do next

• Every screen in a project has the same header: the project's name, ← All projects, and the stages of the research in the order you do them — Overview, Literature, Collect, Analyse, Write, Submit (and Guide feedback once you have a guide). The tab you are on is underlined, a stage that is finished has a tick, and the stage your project has reached has a dot. They are the same names as the journey on your home screen.
• Under the tabs, one line says what to do in this stage — Next in Analyse: run the analysis for each objective — or that the stage is complete and what comes next. Chavery works it out from what your project holds, by a written rule, so the same project always gets the same line.
• The Overview opens on Do this next, with the one blue button. When the thing to do is a card on that page — your sample size, your dates, your guide — the button takes you to it and opens it. Your plan, sample size, dates and supervision follow; More about this project holds the research blueprint, the gap board, writing progress and, for a dissertation, Turn into an article; archiving or deleting is folded away at the bottom.
• Literature covers searching, your library, the gap board and the reference list; Write covers the editor and Writing progress, section by section; Submit builds the Word or PDF file.

Reading the screens: what to press, what the colours mean

• One blue button on a screen is the next step; every other button is plain. On your home screen it sits under Do this next. When nothing is waiting for you there is no blue button.
• Amber means it is your move. It is read from your seat: a chapter your guide sent back is amber on your screen and grey on theirs, and a chapter waiting for your guide is amber on theirs and grey on yours.
• Green is done or approved. Red is a fault or a danger: a problem the writing checker found, or an action that deletes. Being sent back for changes is never red; it is the review doing its job. Grey is everything else, including work that is waiting on someone else.
• The menu is in groups: My research (Projects, New project, Documents), Quick tools, Learn & help, Your roles (such as Supervision or Assistance work) and Account. Quick tools need no project: Check my writing and Format a reference — they used to be called Write and References, which are now only the names of a project's own screens. Statistics are in Chavery Analyse and inside a project; there is no separate “Run a test” tool. New project was called Plan, and Plans is now Pricing.

Get Chavery on your phone

Scan the code with your phone's camera to open Chavery at app.chavery.in.

Chavery opens like an app, works offline, and can notify you when your guide responds.

• Android / Chrome: after opening the link, choose “Install” or “Add to Home screen” from the browser menu.
• iPhone / Safari: open the Share menu and choose “Add to Home Screen”.

Still stuck?

Write to support@chavery.in with your project name and what you were trying to do — we read every message. You can also browse Learn for short guides on methodology and statistics, or Examples for a worked dissertation end to end.