Two pages when printed: the tour on page 1, the ten questions on page 2. Also: User guide · Methods guide
Page 1. A one-hour tour, in the order to learn it
Do this once with the app open, on a laptop, with Data, Sample datasets, Our class data (40 students) loaded. Each step is five to ten minutes. The point is not to learn statistics again; it is to know where everything is, so that when someone asks, you can say "that is under ..." without looking.
- The start screen and the three buttons. With no data loaded the screen offers the six analyses the chosen field uses most and the sample datasets. Above the table: Which test? (three questions that open the right dialog), Student (click it: student or researcher, and the field; student cards carry the formula, the worked numbers and a plain-English conclusion; researcher cards put the result first and fold the method away) and Ask (type a question in plain words, such as "did the two groups differ", and it names the method, says why, and opens it).
- Getting data in, with one messy file. Data, Open CSV or Excel file and Paste data accept what spreadsheets really export: semicolons, blank rows, "NA", a total row, dollar signs. The grey line under the table says how many rows and columns came in. Click a cell to edit it. Data, Set variable type if a number column was read as text. Filters keeps the rows you name, and every card says which filter was on.
- Which test? and Ask. Run both on the class data: "is there a difference in sleep hours between males and females" (Ask opens the Independent Samples T-Test with the variables picked). Then the Methods guide: the same choice by research question, with the steps.
- A t-test, an ANOVA and a regression on the class data, reading each card top to bottom: the grey line (dataset, variables, n), the StatCrunch-style block (the parameter in words, H0 and HA, one results table with Sample Mean, Std. Err., DF and T-Stat or Critical t), the table named as the course names it, the formula box, "show the numbers in the formula", the APA line, the blue decision box, the green or pink condition check, the graph. New in every test dialog: Perform, a hypothesis test or the confidence interval only (no test value to type).
- Check my work. On any card, check my work: the student types the numbers they got by hand and each one is marked as a match or not, within their rounding. This is the button to show someone who says "my answer is different".
- Copy for Word and the R script. Under every table, copy for Word gives an APA-style table that pastes into Word or Google Docs. Results, Script: every step, with the R code lists everything on the page with the choices made and the R call that does the same; download it as a .R file.
- The Learn pages. Learn, The Central Limit Theorem, What a confidence interval means and What a p-value is are the three to know for class. Each is a simulation with a sentence that changes as it runs.
- The course profile and the share link. Results, Course profile sets the defaults for a class (Welch or pooled, quartile rule, decimals, alpha, which menus show) and saves them as a file or a link; students open the link once. Results, Share link carries the data, the filter and the analyses to someone else; Save session keeps them as a file; Export results as HTML or Print or save as PDF is what students hand in.
Three things to say to every new user. Nothing leaves the browser: no account, no upload. The browser remembers your own data and results between visits on the same device. If a card is not what you meant, press edit on it, change one thing, and Run: the new result replaces the old.
Page 2. The ten questions students ask most, and where the answer is
| They ask | Say, and where to point |
|---|---|
| "Which test do I use?" | Press Which test? (three questions) or type the question into Ask. For the reasoning, the Methods guide, by research question. |
| "My file won't open" or "the numbers came in as text" | Open it again with Data, Open CSV or Excel file; the app reads semicolons, blank rows, NA and total rows. If a column is still text, Data, Set variable type. For a quick check, Data, Paste data straight from the spreadsheet. |
| "Where is the p-value, and what does it mean?" | The P-value column of the StatCrunch-style results table at the top of the card (for a t test it sits next to T-Stat; the course-named table below repeats it as p), then the blue box, which compares p with alpha and writes the conclusion in words. For the idea, Learn, What a p-value is. |
| "The test statistic is huge" or "it asks me for a test value" | A hypothesis test needs the mean the problem claims (H0) typed into Test value; the dialog refuses to run without it. A huge t means the test was run against 0 (the mean divided by its standard error): type the claimed mean, or if the task is an interval, set Perform to Confidence interval only. |
| "I only need the confidence interval, not a test" | In the dialog, set Perform to Confidence interval only: the test value, the hypothesis and alpha disappear; the card gives the interval, the margin of error and a sentence reading it. Type the level in Confidence level (percent): 95, 98, 90, anything. |
| "The problem gives sigma; do I use z or t?" | Use t unless the problem gives the population sigma; then T-Tests, One Sample Z-Test (sigma known). Every t card says so in its formula box. |
| "My answer by hand is different" | Press check my work on the card and type the numbers they got; each is marked within their rounding. Then open the formula box and show the numbers in the formula: the step where they differ is usually the standard error or the degrees of freedom. |
| "Which option is Welch, and which is pooled?" | In the Independent Samples T-Test dialog, the Welch's box (checked by default; the course profile can change the default). Welch when the two SDs differ, pooled when the course assumes equal variances. The card names which was used. |
| "How do I put this in my report?" | The Report it as (APA style) line on the card, copied as text; copy for Word under the table for an APA-style table; Results, Print or save as PDF or Export results as HTML to hand in the whole page. |
| "I closed the tab; is my work gone?" | No. The browser keeps your own data and the results on the same device; the app offers them back on the next visit. To keep them for good or move them, Results, Save session (a .sww.json file) and later Open a saved session. A sample dataset you did not change is not kept, since it is always one click away. |
| "Can I use it on my phone, offline, in Spanish?" | Yes to all three. Phone: the table sits above the results. Offline: open it once and it stays; on a phone, "Add to Home Screen" gives it an icon (the store app is coming later). Spanish: the Español button in the top links, or Results, Language / Idioma; the result cards stay in English so that everyone compares the same record. |
When the pink box appears (a condition is not met), do not reach for another test on their behalf. Read the box with them: it names the condition and what to do, which is usually a nonparametric alternative under Nonparametric or a different interval method. That reading is the learning.
If the app itself does something wrong, the troubleshooting section of the User guide has the three things to try, and the email on the home page reaches Safaa.