In NicolasJBM/writer: Functions and interfaces to support academic writing with R

library(writer)
library(tibble)
library(tidyr)
library(dplyr)
library(htmlTable)
library(stargazer)
library(ggplot2)
# Set global options.
options(xtable.comment = FALSE, xtable.floating = FALSE, xtable.timestamp = "", stargazer.header = FALSE)

You can start here is a small abstract of the analysis, so that later you can quickly check what is waas about.

The main purpose of a note book is to keep a commented record of all the preliminary analyses you have conducted before settling on the specific set. Indeed, in the process of researching, you will explore many inconclusive paths. Yet, it is important to keep track of such inconclusive paths to avoid following them again in the future. In addition, the results of the analyses can be store in a specific RData file for use when you write your report or paper.

Each notebook can refer to a specific set or address a specific issue. Then, it is helpful to follow the same sequence of analyses as in a conventional paper:

• Purpose;
• Variables;
• Descriptive statistics;
• Main analysis;
• Robustness tests;
• Additional analyses;
• Conclusion.

Purpose

Although it might appear evident at the time of the analysis, it is important to clearly specify the purpose of the study. It is also helpful to give some background from other sources. First, it forces you to formalize what has to be done. Second, it will help the future you to understand what you tried to achieve here.

Variables

In later stages of the research, you will also benefit from having writen something about the definitions and operationalizaition. Although it might seem redundant to do it several times for the same concepts, it is actually good practice as you might progressively improve your descriptions and justifications, while keeping track of their evolution.

How where the data collected?

Feedback

What items where used? how did you aggregate them? What test did you conduct to assess their validity and reliability?

Satisfaction

What items where used? how did you aggregate them? What test did you conduct to assess their validity and reliability?

Descriptive statistics

Descriptive statistics help noticing some issues which might bias your estimates. It also helps having a better understanding of the main sample characteristics which may drive some of your results.

Univariate

Some basic statistics:

library(modestim)
edwards2007 %>%
  select(comc, satc, fbkc, cen, gen) %>%
  univariate() %>%
  column_to_rownames("Variable") %>%
  htmlTable(
    header = names(.),
    rnames = row.names(.),
    rgroup = c("Dependent","Independent","Moderators"),
    n.rgroup = c(1,2,2),
    cgroup = c("Spread","Distribution"),
    n.cgroup = c(4,4),
    align = "l|rrrr|rrrr|"
  )

ggplot(edwards2007, aes(x = comc)) + geom_density()

Bivariate

edwards2007 %>%
  select(comc, satc, fbkc, cen, gen) %>%
  bivariate() %>%
  htmlTable(
    header = names(.),
    rnames = row.names(.),
    align = "ll|rrrrr|"
  )

Main analysis

Robustness tests

Alternative samples

Alternative operationalization

Alternative specifications

Additional analyses

NicolasJBM/writer documentation built on Aug. 12, 2019, 2:36 p.m.