How to Add Error Bars in SciDAVis

Error bars turn a plain line or scatter plot into an honest picture of your measurement: they show how far each point could plausibly move. In SciDAVis 2.9.x you add them with the Graph → Add Error Bars… command, which opens the Error Bars dialog. This guide walks through the three supported error sources — an existing column, a percentage of the data, or the standard deviation — and explains what SciDAVis cannot do (asymmetric errors). Menu paths and dialog labels were checked against the official SciDAVis Handbook and the current SciDAVis source code.

Where do the error values come from?

The Error Bars dialog offers three ways to define the bar half-width E around each data value V (bars always run from V−E to V+E, i.e. they are symmetric):

Step-by-step: adding error bars to a plot

1. Create a 2D plot of your data first (e.g. select a Y column and choose Plot → Line or Plot → Scatter). Error bars are added to an existing curve — they cannot be created on an empty plot.
2. Click the plot window to make it active, then choose Graph → Add Error Bars....
3. In the Error Bars dialog, the “Add Error Bars to” drop-down lists the curves on the plot — pick the one you want.
4. Tick Y Error Bars (vertical bars), X Error Bars (horizontal bars), or both.
5. Under “Source of errors”, choose Percent of data (%), Standard Deviation of Data, or Existing column (and pick the column if you chose the latter).
6. Click Add — the bars appear on the plot. Then click Close to dismiss the dialog.

Using your own error column (SD, SE or confidence intervals)

This is the most flexible route, and the one researchers use most:

1. In your table, add a column next to the data with one error value per row — for example the standard deviation of three repeated readings, or a 95 % confidence-interval half-width. Tip: mark it clearly, e.g. name it y_sd.
2. Optionally, classify it via Table → Set Column As → Y error (or → X error for horizontal bars). This designation marks the column’s role but is not strictly required by the dialog.
3. Open Graph → Add Error Bars..., select Existing column, and choose your error column from the list.

The error column must have the same number of rows as the data column; otherwise SciDAVis shows an error (“The selected columns have different numbers of rows!”). Also note that error bars are drawn for every plotted point, so trimming your data range in the table changes which bars appear.

Styling: width, caps and color

Error bars added through the dialog are drawn thin (1 px), black, with end caps on both the plus and minus sides, passing through each data point — these are the built-in defaults and the dialog offers no width, color or cap-length controls. Keep that in mind when designing a figure: the bars are deliberately minimal, which suits most journal styles. If your bars are hard to see, enlarge the plot or thicken the data curve itself in the curve’s properties dialog so the black caps stand out against it.

Worked example: a physics lab measurement

Say you measured the period of a pendulum 5 times at each of 6 string lengths, and the table holds the mean period T in one column and the standard deviation of the 5 repeats in a second column:

1. Plot string length (X) vs. mean period (Y) as a scatter plot.
2. Graph → Add Error Bars... → tick Y Error Bars → Existing column → choose your SD column → Add → Close.
3. The plot now shows each mean with ±SD bars — exactly what a lab report expects. If your supervisor wants ±SE instead, add a column computing SD / √5 and use that column as the source.

What SciDAVis cannot do: asymmetric error bars

To be explicit: SciDAVis error bars are always symmetric (V−E to V+E). There is no built-in support for asymmetric errors (different plus and minus widths, e.g. from Poisson counting statistics or log-normal fits). The dialog accepts a single error column per direction. If your analysis genuinely needs asymmetric bars, compute them outside SciDAVis (e.g. in R or Python) and import the finished figure, or draw the interval as a separate shaded curve — do not fake it by averaging the two sides, since that misrepresents the uncertainty.

FAQ

Can I add error bars to a function curve?

No. SciDAVis refuses with the message “This feature is not available for user defined function curves!” — error bars require a table column behind the curve. Plot your computed values as a data column first, then add the bars.

Why are my error bars invisible?

Three usual causes: the error column contains zeros or missing values; the bars are tiny relative to the axis scale (zoom in to check); or you clicked Add and then OK, which is harmless but adds the same bars twice — they overlap and look identical, not invisible. Also confirm the error column has the same row count as the data column.

Can I show X and Y error bars at the same time?

Yes — tick both X Error Bars and Y Error Bars in the dialog before clicking Add. You can also run the dialog twice with different sources (e.g. percent for X, existing SD column for Y).

Do the “Percent” and “Standard Deviation” options change my data?

They don’t touch your data columns, but they do append a new column to the active table with the computed error values. You can rename, edit or delete that column like any other — the bars already on the plot keep a reference to it, though.

Next steps

• Nonlinear Curve Fitting — fit a model through data that carries uncertainties
• How to Import Data into SciDAVis — bring in repeated measurements cleanly
• Export Publication-Quality Figures — prepare the final figure for your report
• Multi-Layer Graphs — compare runs with and without error bars side by side
• All SciDAVis tutorials