Qualtrics is an online survey platform similar to SurveyMonkey that is used by researchers to collect data. Until recently, one had to manually download the data in either SPSS or .csv format, making ongoing data analysis difficult to check whether the trend of the incoming data supports the hypothesis.

Jason Bryer has recently developed an R package published to Github for downloading data from Qualtrics within R using the Qualtrics API (see his Github repo). Using this package, you can integrate your Qualtrics data with other experimental data collected in the lab and, by running an Rscript as a cronjob, get daily updates for your analyses in R. I’ll demonstrate the use of this package below.

When presenting data, confidence intervals and error bars let the audience know the amount of uncertainty in the data, and see how much of the variance is explained by the reported effect of an experiment. While this is straightforward for between-subject variables, it’s less clear for mixed- and repeated-measures designs.

Consider the following. When running an ANOVA, the test accounts for three sources of variance: 1) the fixed effect of the condition, 2) the ability of the participants, and 3) the random error, as data = model + error. Plotting the repeated-measures without taking the different sources of variance into consideration would result in overlapping error bars that include between-subject variability, confusing the presentation’s audience. While the ANOVA partials out the differences between the participants and allow you to assess the effect of the repeated-measure, computing a regular confidence interval by multiplying the standard error and the F-statistic doesn’t work in this way.

Winston Chang has developed a set of R functions based on Morey (2008) and Cousineau (2005) on his wiki that help deal with this problem, where the sample variance is computed for the normalized data, and then multiplied by the sample variances in each condition by M(M-1), where M is the number of within-subject conditions.

See his wiki here for more info.

By using LaTeX to author APA manuscripts, researchers can address many problems associated with formatting their results into tables and figures. For example, ANOVA tables can be readily generated using the xtable package in R, and graphs from ggplot2 can be rendered within the manuscript using Sweave (see Wikipedia). However, more complicated layouts can be difficult to achieve.

In order to make test items or stimuli easier to understand, researchers occasionally organize examples in a table or figure. Using the standard \table command in LaTeX, it’s possible to include figures in an individual table cell without breaking the APA6.cls package. For example: