r-data-generator

Two sets of R scripts for creating statistically significant data illustrations for bivariate regressions and difference-of-means tests.

File Description

Regression Package

1. reg/preamble-reg.R defines the parameters for the scripts in this package. Change the variables accordingly to match your own research
2. reg/null-reg.R generates a dataset where there is no significant relationship between the two variables.
3. reg/datagen-reg.R randomly generates a 20-point dataset with high r-squared value and low p-value.
4. reg/interpolate-reg.R generates Sample size - 20 extra data points using the principal dataset in Step 1, while maintaining its high R2 value. (this part is deliberately separated from step 3 for the user to test different interpolation models without resetting the principal dataset)
5. reg/combined-reg.R combines (3) and (4) for those who do not want the hassle of running two separate scripts.

Bar Chart Package

1. bar/preamble-b.R defines the parameters for scripts for this package. Change the variables accordingly to match your own research.
2. bar/null-b.R generates a bar graph where no significant difference is detected across the groups.
3. bar/sig-b.R generates a bar graph with significant differences in the means of the data groups. Means are sorted according to the order that the groups are defined in the xlabels parameter.

Dependencies:

• Regression package dependencies (run reg/dependencies-setup-reg.R to install and setup)
  • maptools
  • jtools
  • stargazer
• Bar graph package Dependencies (run bar/dependencies-setup-b.R to install and setup)
  • ggplot2
  • dplyr
  • pander
  • Pandoc (requires restart after install)

Please cite as:

• Koo, J (2020). R data generator. R package version 0.0.3. Available online at: https://github.com/tanxpyox/r-data-generator.

Q&A

Why create this script? Is this created to do sketchy stuff?

• A: No, this was originally created to complete my politics coursework, which required me to present two different hypothetical datasets, where I (i) fail to reject or (ii) can reject with confidence the null hypothesis. I ask that all users of this script to abide by the academic standards of ethics and indicate that the dataset is only hypothetical and created for illustration.

Will more data patterns be added?

• A: Sure, leave your suggestions in the issues tab.

Will this be available as an R package?

• A: I currently have no plans of making this into a package, partly because this code is still very crude and will need some major rewriting to accomodate for all possible users.--But, with your help, maybe I can!

Licence

• The scripts in this repo are released under the GPL 3.0 licence. See LICENSE for the full terms and conditions.