RESEARCH METHODOLOGIES
SPRING 2025
UNIVERSITY OF SHARJAH
DATA ASSIGNMENT 2
Please turn in a Word document (NOT PDF) with the answers to the following questions and
copy and paste your R code into the file. DO NOT USE SCREENSHOTS OF R CODE!!!
You can ONLY use packages, functions, and arguments to functions that I use in the class
materials. If you use and packages, functions, or arguments to functions that are not in
the class materials, you will receive 0 points for that question. Also, if I can determine
that you use AI to provide the answers, you will receive 0 points.
There is no need to create a “fancy” document with things like tables of contents, etc. Just
answer the questions that are asked.
The data set cabinet.csv contains data on how long governments survived in office for 15
countries in the post-war period. The variables are as follows:
durat Duration of the government in months.
polar Polarization index: a measure of support for extremist parties.
fract Fractionalization: an index characterizing the number and size of parties in
parliament such that higher values indicate dispersion into a larger number of
relatively smaller blocs.
format Number of formation attempts.
minority Majority/minority government (0=Majority, 1=Minority).
postelec Post-election formation: whether or not the government was formed
immediately after an election or not (0=Immediately, 1=Delayed).
caretakr Caretaker government: a small subset of governments specifically established
only pending new elections (0=Caretaker, 1= Not Caretaker).
The aim of this assignment is to assess the impact of various factors on the longevity of
governments that have not reached their full term. Those governments that effectively
reached full-term and had to call an election within the year may have survived for much
longer had they been allowed to, so our data-set only includes those that ended before their
time was constitutionally up.
Part 1a (30 points)
Regress durat on each of fract, polar, minority, format, postelec and caretakr separately.
There should be 6 models. Present the regression results in a single table using the stargazer
function.
For each model, provide a brief interpretation of the coefficient, including whether or not the
effect is significant and what the effect is. Which of the 6 independent variable is the best
predictor of duration and how do you know this?
1b: (5 points) Create a scatterplot of duration and polarisation and include the linear fit line.
Be sure to include axis labels.
1c: (5 points) Create a scatterplot of duration and fractionalisation and use different coloured
points for minority and majority govts. No need to include the linear fit line. As always, be
sure to provide axis labels.
Part 2a: (15 points)
Rather than regressing durat on each of fract, polar, minority, format, postelec and caretaker
separately, include all these variables in one multiple regression. Produce a table of results
using stargazer How do these results differ from your conclusions from the bivariate
regressions? Give some possible explanations for these differences.
2b: (5 points ) Make a plot of the coefficients from this regression using the function
coefplot in the package ‘arm’.
2c (6 points) Make an effects plot for polarisation and for caretaker (2 separate graphs). Be
sure to include the partial residuals in the plots and to supply a meaningful title to the plots)
2d (4 points) Make an effects plot for the effect of polarisation for each value of minority.
This should be one plot with 2 parallel lines on it.
2e (8 points) Compute the predicted duration for a minority government that was formed
immediately, is not a caretaker, and has the mean values fract, polar, and format. You need
to show your work here, not just provide the answer.
