Coming soon

These How-Tos are on the way. Each will stand on its own, like the ones already listed on the How-Tos page.

Using your own data

The bridge from the practice datasets to a file of your own: opening an Excel, CSV, or commercial-software file, keeping it in a data folder, and taking a first look with jscreen().

Working with data from commercial statistical software

Importing and working with data from SPSS, Stata, and SAS, including value and variable labels and the different missing-value conventions.

Reading jstats output: Case Processing and Missing data

Not how to interpret the statistics, but how to read the accounting printed above them: the Case Processing lines and tables that say how many cases an analysis used and which were set aside, and the Missing data breakdown that shows where each variable’s missing values went.

Recoding and relabeling variables

Reverse-coding an item, collapsing categories, and fixing variable and value labels with jrecode() and jrelabel(), keeping the original column intact.

Text answers into coded variables

Turning a column of words (“Yes”, “No”, “Don’t know”) from a spreadsheet or survey export into a labeled, coded variable with jencode().

Building and checking a scale

Combining Likert items into a scale and checking its reliability with Cronbach’s alpha.

Working on part of your data

Analyzing a subgroup with jsubset(), and using jcomplete() so a set of analyses all run on the same cases.

Comparing groups: t-tests and cross-tabulations

t-tests with jt() and cross-tabulations with jcrosstab(): the same formula style as jaov(), how to read what comes back, and the Welch version of the t-test (welch = TRUE), which does not assume equal variances. One-way ANOVA is already available.

Regression with categorical predictors

Linear and logistic regression with jlm() and jlogistic(). jdummy() is the quick way in for a categorical variable: register its dummy coding and reference category once, and it goes into a model by name — no building dummy variables by hand, and no converting to base R’s factors.

Handling missing values across platforms

What survives when data move between SPSS, Stata, SAS, and R: round trips between formats, converting declarations from one convention to another, missing-value ranges, and the SAS interchange format. Declaring missing values in the first place is covered in Telling jstats about missing values.