## Description

Group-based assignment in quantitative methods

An independent group-based assignment is part of the examination for this course. As explained in

the syllabus (and in the course plan), the student must receive a pass on this assignment to receive a

final grade for this course.

The group is free to base its assignment on the data from either the World Value Study (WVS) or the

QoG-institute. That is, the assignment can either be based on analyses of data where countries are

the units of analysis (the QoG) or data where the units of analysis are individuals nested in different

countries (WVS). Building on either of these two databases, the group should select one dependent

variable and present hypotheses about independent variables that may explain variations in the

dependent variable.

The assignment should consist of no less than 4000 words (excluding Tables and Figures), but may

not exceed 4500 words. Moreover, statistical results must not be “copied and pasted” from the

output in the statistical software. Instead of using unedited output, students should present tables

and figures that could be published.

The group assignment is due on Wednesday the 8th of January at 14.00.

The assignment should consist of the following steps:

1. Introduction

a. Present a brief research problem where you highlight why it is theoretically

important to review variations in the phenomena (dependent variable) you have

chosen for your assignment. Your arguments should be based on relevant literature,

that you cite properly to support your claims. Moreover, given the global reach of

both datasets, your research problem should have a comparative perspective – that

is, the outcome you study should most likely differ between countries. Here, you

need to motivate why.

b. Present a clear and unambiguous aim of your assignment.

2. Theory and hypotheses

a. Present a brief, but theoretically grounded, discussion about the factors

(independent variables) that may affect variations in the dependent variable. Here it

is required that your arguments are based on relevant literature, that you cite

properly to back up your claims.

b. Based on this, present one or more hypotheses that you will test in your assignment.

Present these hypotheses properly.

c. Motivate a set of control variables that you use in your subsequent regression

analyses. Moreover, given the global nature of both datasets, countries are evident

control variables that should be included and motivated.

3. Descriptive statistics

2

a. Review the central tendency of your variables and discuss their variation in

substantive terms.

b. Explore the distribution of your dependent variable. Is it normally distributed? If not,

how may this affect your results? In case of highly shewed dependent variable, what

have you done to minimize this problem?

4. Correlation and bivariate regression analyses

a. Present correlations between all of the variables in your analyses. Interpret your

results in substantive terms. What do the correlations say about the suitability of

including all your variables in a multivariate regression analysis? Should some of the

variables be removed and/or replaced?

b. Present bivariate regression analyses to test the separate effect of each of your

independent variables. Comment substantially upon the strengths of the regression

coefficients as well as measures of ‘goodness-of-fit’ (R-squared and the standard

error of the estimate). Also, comment upon the statistical significance of the

relationships.

5. Multivariate regression analyses

a. Set up a model where all of your variables are regressed against the dependent

variable. Compare the results with the bivariate regressions and comment in great

detail upon regression coefficients as well as measures of ‘goodness-of-fit.’ Also,

comment substantially upon how the estimates differ between the bivariate and

multivariate analyses.

b. Add new models where you control the results from the previous model (5a) for (at

least 10 well argued) country dummies. Do these controls affect the results? What

do differences between coefficients for the country-dummies mean?

c. Add a new model with interactions between county-dummies and the one or more of

the central variables in the analyses. Comment substantially on the results in relation

to previous models.

6. Conclusions

a. Summarize your main findings. Is your aim fulfilled? Are your hypotheses supported by

the analyses. Why/why not?

## Reviews

There are no reviews yet.