iNZight provides users with an easy interface for adding inference information to plots, using "normal theory" or bootstrap methods. The inference options available again depend on the type of plot drawn.
If a scatter plot is drawn, you must first add trend lines before you can obtain inference information. This is done using the Add to Plot button, selecting Add trend line and selecting one or more trends (this includes the smoother).
Having added a trend line or lines, the inference information will generate 30 bootstrap trend lines and display these on the graph, giving users an idea of the precision of the estimated curve.
Dot plots, by default, provide a box plot which provides an estimate of the Median, as well as the 25% and 75% quantiles. Users are then presented with three options:
Parameter: the parameter for which inference is obtained
Mean: this will generate inference of the mean of the data
Type of Inference: Depending on the parameter, users can select the type of inference used to obtain the intervals displayed on the graph
Normal: this uses standard normal theory to approximate a confidence interval
Year 12: for medians, the year 12 (NCEA L3) method of computing an interval is used
Type of Interval: the type of interval displayed
Confidence Interval: this displays a 95% confidence interval
The inference information for bar plots is computed for the estimated proportions, using normal theory or bootstrap techniques.