From statistics basics to using statory
New to statistics? No problem. Four basics and two deep dives take you from preparing data to reading results.
What is statistics?
Statistics is not scary — it is how you summarize more than you can read, so you can decide with evidence instead of gut feeling.
Statistics is summarizing
Imagine running a cafe. A month of sales records is thousands of rows — you cannot read them all. "About 300 cups a day, americano sells best": reducing a large record to a few numbers is the first face of statistics, called descriptive statistics.
Statistics is comparison and relationship
After summarizing come questions: "Do weekends really sell more than weekdays?" (comparison) or "Do iced drinks sell more when it gets hotter?" (relationship). The second face, inferential statistics, checks whether such differences and relationships are real or just chance.
Chance vs. real — significance
If weekends sold 5 more cups, it could be chance. If 50 more, chance is unlikely. Statistics computes "how unlikely this is to be chance" — that is significance, and the significance badge on every statory result is exactly this verdict.
Sample and population
You cannot ask every customer, so you ask some (a sample) and estimate the whole. Large enough samples give stable estimates — this is why "how many respondents do I need?" matters.
In statory
You do not need to memorize any of this. Upload data and statory runs the summaries (basic battery) and the comparisons/relationships (designed statistics) automatically, attaching a significance badge and a plain-language explanation to each result.
What data do I need?
Half of analysis is the shape of your data — one row per person (case), one column per question (variable). One such table is all you need.
The basic shape: rows = people, columns = variables
For a survey, each row is one respondent and each column is one question, with variable names (e.g., gender, satisfaction) in the first row. Excel (xlsx), CSV, and SPSS (sav) files work as-is. For interviews, one document is one case.
DATAMAP — the manual for your data
Variable names alone cannot tell whether "1" means male or female. A DATAMAP is a companion sheet (variable info + value labels); upload it with your data and statory merges it automatically so every table and chart shows human-readable labels. Designing a survey? The design assistant generates a DATAMAP draft for you.
Naming multi-item scales
If "job satisfaction" is asked with 5 items, name them Q1_1, Q1_2 … Q1_5 — same prefix, numbered. statory detects the set automatically, verifies reliability, and builds a composite (mean) variable for analysis.
Do not worry about missing values and outliers
Blanks (missing) and odd values (outliers) are fine. Automatic diagnosis checks them at upload; the auto path handles them invisibly, and the data prep screen (/prep) lets you adjust them precisely if you prefer.
Quantitative vs. qualitative analysis
statory stands on three kinds of questions — ones numbers answer, ones words answer, and ones that connect the two.
Quantitative — questions numbers answer
"Does satisfaction differ by gender?" (difference), "Are stress and burnout related?" (relationship), "What drives turnover intention?" (influence). Survey scores and sales figures are the material; results come as statistics with significance.
Qualitative — questions words answer
"What do users talk about?" (themes), "How can opinions be categorized?" (coding), "Who connects with whom?" (networks). Interview transcripts and open-ended answers are the material, and every result carries verbatim excerpts as evidence.
Mixed — questions that connect the two
When numbers and words work together — "Do satisfied and dissatisfied groups talk differently?" statory places group voices right next to the quantitative result.
Special & strategic — decisions without data files
Decision analyses like "which alternative should we choose?" (AHP·TOPSIS) or "what is our fallback if negotiation fails?" (game theory) can be completed with no data table — just your judgment structure.
In statory
The analysis picker is organized exactly along these three branches (quantitative · qualitative · special/strategic), and once data is connected, each method shows upfront whether it qualifies.
How to use statory
Upload your data and press [Proceed automatically] — that is it. Want control? [I will do it myself]. No data yet? "No data yet?" has you covered.
Auto path — from upload to report
Upload a file on the home page and the recognition card tells you the data type (quantitative · qualitative · mixed) and quality. Press [Proceed automatically] to receive research questions and an analysis design; one confirmation ([Run all as designed]) completes preprocessing, all analyses, and key findings automatically.
Direct path — three branches you choose
Press [I will do it myself]: quantitative goes through data prep (/prep) to the analysis picker, qualitative to the six-purpose selection, mixed to two side-by-side lanes. /core7 (the essential seven) and /analysis (all 73 methods) are always open.
No data yet? Start by creating it
"No data yet?" on the home page offers four doors — ① create with a survey (design assistant: variable table · guaranteed analyses · DATAMAP draft · sample-size estimate) ② create with interviews (design guide: case counts · questioning · what analysis follows) ③ enter data on screen (pairwise, decision-table, or game-payoff grids — complete with no file) ④ this guide. When collection is done, upload at the front door.
Upload data — we guide you automatically
Not sure which analysis to run? Upload your data and we will detect its type and recommend a suitable analysis path.
Data Analysis Guide
Home → Data Analysis card → attach a file or type text → choose an analysis
If you already have Excel, CSV, or SAV data, start from the Data Analysis card. You can also type a request without a file and continue on the upload screen.
Good input examples
- "In this survey data, compare satisfaction by gender"
- "Analyze the relationship between yearly sales and ad spending"
- "Find main themes in my interview transcripts"
- "See positive vs negative tone in customer feedback"
Recommended path
- Attach a file so we can check data type (numbers vs text) and quality first.
- If you do not know the method, choose one below by your goal.
- You must upload data before starting an analysis.
Deep dive — the direct path
The screens you meet after pressing [I will do it myself], and how to decide at each fork.
/prep — the preprocessing home
Missing/outlier diagnosis with auto-fix, AI label shortening, new variables (compute · recode · binning · case filters), and save/restore. Refine, then continue via [To analysis selection] at the bottom. Rule of thumb: skip it if your data needs no touch-up.
The picker — three branches and qualification badges
/analysis is organized into quantitative · qualitative · special/strategic. With data connected, every method card shows its qualification — unusable methods dim with a one-line reason (e.g., "needs 2 numeric variables"). Rule of thumb: a dimmed card means change the data or pick another method.
Three doors into special & strategic
① Pick directly from the picker (a wizard guides the inputs) ② if a structure (alternatives × criteria matrix, etc.) is detected in uploaded data, the recognition card proposes it ③ qualitative results that reveal a decision structure offer a link. Whichever door — confirming the values is always yours.
/core7 — the essential seven
Frequency, descriptives, reliability, t-test, ANOVA, correlation, regression — in order, fast. Upload → quick tidy → run the seven. Rule of thumb: the shortest path when you need standard paper tables now.
Deep dive — reading your results
statory results form a pyramid — read top-down, descending only as far as you need.
Level 1: key findings
The most important conclusions from the entire run come first. Short on time? Stop here.
Level 2: one-line conclusions with significance badges
Each research question gets a one-sentence conclusion and a significance badge (significant / not significant) — the verdict on "is this unlikely to be chance?". Click only the questions you care about to descend.
Level 3: details — tables, charts, APA sentences
Expanding reveals statistical tables, charts, publication-format (APA) sentences, and AI interpretation. AI interpretation is advisory; the tables are always the source of truth for numbers.
Composite variables and the α (alpha) notation
A label like "Community satisfaction (mean of Q1_1–Q1_16, α=.91)" means a composite variable averaging several items. α (Cronbach's alpha) is the reliability of those items measuring the same thing; statory uses the composite only when α ≥ 0.6, otherwise it analyzes the original items and records the reason in the prep report.
Data overview and verbatim evidence
With many variables, frequencies/descriptives are isolated into a compact table (one row per variable) that expands on click. Qualitative results always carry verbatim excerpts as evidence — statory never ships a summary without them.
Measurement & structural model path diagrams
Factor analysis, confirmatory factor analysis (CFA), and structural equation modeling (SEM) results come with a path diagram linking latent factors, items, and paths. Path diagrams keep improving — publication-quality auto-layout and export are on the way. Numbers (loadings, path coefficients, fit indices) are always sourced from the tables.
Which analysis should I use?
Upload your data and we will recommend an analysis path. If you already know what you need, pick a category below.
Prefer to choose yourself? — pick a category below
※ Upload your data first to start an analysis.