Descriptive Statistics and Statistical Tests in SciDAVis

SciDAVis can summarize any table with one click: mean, standard deviation, variance, min, max and more appear in a new statistics table. What it cannot do is run hypothesis tests — there is no built-in t-test, ANOVA or normality test in SciDAVis 2.9.x. This guide shows you exactly what the statistics commands produce, how to read the results table, and honest workarounds for the tests that are missing. Menu paths and output columns were verified against the SciDAVis Handbook and the current SciDAVis source code.

Getting descriptive statistics for your columns

1. Open the table and select the column(s) you want to summarize (click a column header; Ctrl-click for several). You can select multiple columns in one table — one results row is created per column. Columns from different tables cannot be combined into a single statistics table.
2. Choose Analysis → Statistics on Columns.
3. SciDAVis creates a new table named “Column Statistics of <your table>” with one row per selected column.

The sibling command Analysis → Statistics on Rows does the same thing across rows instead of columns (useful when each row is one experimental run). Both commands live in the Analysis menu, which appears when a table window is active.

Reading the results table

The column-statistics table has 11 columns. Here is what each one means, with the exact headings SciDAVis uses:

The row-statistics variant is similar but simpler: Row | Cols | Mean | StandardDev | Variance | Sum | Max | Min | N — it has no iMax/iMin columns.

What the table does not give you: standard error

There is no SE (standard error of the mean) column. That is easy to derive yourself, since SE = SD / √N and all three ingredients are in the table:

1. In the statistics table, add a new column (e.g. SE).
2. Right-click the new column header and use Set Column Values with the formula StandardDev / sqrt(N) — SciDAVis evaluates it row by row, giving one SE per summarized column.

This two-step pattern (statistics table → derived column) is the general recipe for anything beyond the built-in summaries.

The honest part: no t-test, no ANOVA, no normality test

Let us be direct, because this is where many tutorials mislead: SciDAVis 2.9.x has no built-in t-test, ANOVA, chi-square test or normality (Shapiro–Wilk) test — not in the Analysis menu, not in any dialog. We checked the current source code: the Analysis menu contains Statistics on Columns/Rows, FFT, correlation, convolution and fitting, and nothing else in the hypothesis-testing family. An older QtiPlot-era Handbook describes the same set. Do not go hunting for a hidden menu; it does not exist.

Practical alternatives, in order of convenience:

1. SciDAVis Python scripting. SciDAVis ships a Python console (Scripting menu). If SciPy is installed in your Python environment, you can run scipy.stats.ttest_ind, scipy.stats.f_oneway or scipy.stats.shapiro directly on your table data without leaving the program — see our Python scripting tutorial.
2. Export and use a statistics package. File → Export ASCII writes your table as CSV, which drops straight into R, JASP or Excel. This is the most robust option for ANOVA with post-hoc tests, which SciPy alone handles less comfortably.
3. Manual computation for simple cases. A one-sample t statistic needs only mean, SD and N — all in the statistics table — so for quick checks you can compute it in a spare column with Set Column Values and compare against a t-table.

Example workflow: reporting “mean ± SD (n = …)”

The most common use of these statistics is the methods line of a paper. Here is the full loop:

1. Select your measurement columns → Analysis → Statistics on Columns.
2. Read off Mean, StandardDev and N for each group.
3. Report as “12.4 ± 1.1 (n = 30)” — and note whether ± is SD or SE, since the two are often confused. SciDAVis gives you SD; derive SE as shown above if that is what your field expects.
4. If groups need comparing, take the CSV export route (option 2 above) and cite the actual software used.

FAQ

Is StandardDev the sample or the population standard deviation?

SciDAVis computes it with the GSL statistics routines, which use the sample formula (dividing by N−1). For the large N typical in lab data the difference is negligible, but cite it correctly if it matters for your field.

Why was my column skipped by Statistics on Columns?

Either the column is not numeric (text or date columns are ignored), or every cell is empty/invalid. Check the column type via Table → Column Options. Also make sure the column header was actually selected before opening the Analysis menu.

Do empty cells count as zero?

No. Invalid and empty cells are skipped entirely, and N reports how many valid values went into the calculation. This is the behavior you want for ragged datasets, but double-check that a genuinely-missing value isn’t hiding a data-entry mistake.

Can I get statistics for only some rows of a column?

Not directly — Statistics on Columns always uses the whole column. The workaround: copy the rows of interest into a new table (select rows → copy → paste) and run the statistics there. Alternatively use Analysis → Statistics on Rows if your layout is transposed.

Where is the t-test in SciDAVis?

There isn’t one, in any 2.9.x release we verified. Use the Python scripting console with SciPy, or export to CSV and test in R/JASP/Excel — and describe that step honestly in your methods.

Next steps

• Automate SciDAVis with Python Scripting — run SciPy tests on your tables
• How to Import Data into SciDAVis — clean numeric columns summarize correctly
• FFT Frequency Analysis — the other heavyweight of the Analysis menu
• Nonlinear Curve Fitting — when a model describes your data better than a mean
• All SciDAVis tutorials