smile 8-bit

where i learned what the data is saying

[smile8bit]

Final Project

Governmental Regime Type and Armed Forces Rate Association

Elton Kwan

[smile8bit]

Introduction

There are many types of governments among the nations of the world. There are democracies, anocracies, and autocracies. Each differing principally in their level of citizen freedom. Predominant among government functions is defense. It is well-known that the United States has the largest military expenditures in the world[1]. This analysis will attempt to examine what the regime type means when it comes to the level of armed forces in a nation. 

[smile8bit]

Methods

Sample

Nations and other geographic areas as chosen by the GapMinder organization. There were a total of 215 governments in the data.

Measures

Income per person or GDP per capita was collected from the World Bank and is for 2010 in constant 2000 US$.

The armed forces rate is measured as percentage of the labor force. 

The polity score is a summary measure of a country's democratic and free nature[2].

Secondary variable income level was created from income per person as a categorical variable with levels, lower, lower-middle, upper-middle, and higher. Countries were classifed under this system based on income per person using the group definitions as specified by the World Bank for country income groups[3].

Regime type is a secondary variable that uses the Polity IV Projects definitions.[2] 

[smile8bit]

Research Questions

1. Does "speak softly, and carry a big stick" hold? Is a large army necessary to protect freedom and democracy?  

2. Higher income countries have the choice to spend more on their military while still feeding their citizens but what does that mean when it comes to the armed forces rate? 

[smile8bit]

Results

Univariate

The mean armed forces rates for all nations is 1.44% with standard deviation 1.709.

Over 70 percent of the 129 nations with polity scores were democracies or open democracies

Bivariate

An analysis of variance was performed between armed forces rate and regime type.  The null hypothesis that armed forces rate is equal across all regime types was rejected as the F-statistic of 7.5 had a p-value less than 0.0001. So the mean armed forces rate is not equal across all regime types. 

A Duncan post-hoc test determined that, in particular, the mean armed forces rate for autocracies, 3.1074, is significantly higher than all other regime types.

[smile8bit]

Regime Type Based on Polity Score from Polity IV Project.

Autocracy (-10 to -6)

Closed Anocracy (-5 to 0)

Open Anocracy (1 to 5)

Democracy (6 to 9)

Open Democracy (10)

[smile8bit]

Regime Type

Autocracy

Closed Anocracy

Open Anocracy

Democracy

Open Democracy

[smile8bit]

The results from the analysis show that the F-statistic is 7.50 and the probability of getting such a value is less than 1 in 10,000, so the result is significant enough that the null hypothesis that the mean armed forces rate of all regime types are equal can be rejected.

Further, a Duncan post-hoc test was performed to determine the difference between each pairs of means from regime types.And as the table suggests the armed forces rate for regime types 2, 3, 4 and 5 are not significantly different. However, each of the armed forces rate for regime types 2, 3, 4 and 5 are significantly different from regime type 1.

For reference, regime types:

1 - autocracy

2 - closed anocracy

3 - open anocracy

4 - democracy

5 - full democracy

[smile8bit]

This graph is a way to visualize the regime types that were and were not significantly different with respect to the income group of the country using the Bonferroni method. The regime types with an ‘A’ were not significantly different from one another but were significantly different from the regime type under ‘B’, 5 or open democracy.

Regime Type

Autocracy

Closed Anocracy

Open Anocracy

Democracy

Open Democracy

[smile8bit]

Moderation

Secondary variable, income level, derived from income per person did not moderate the association between armed rorces rate and regime type much. In cases where the results were statistically significant, the results reflected the original association that armed forces rate and regime type had. In other words, autocracies had a significantly higher armed forces rate regardless of income level.

There was one income level for which the relationship differed from the original relationship. For upper middle income nations, the open democracies and autocracies had significantly different armed forces rate means, 0.6508 and 2.978 respectively.

[smile8bit]

Discussion

What do the results mean?

The analysis ultimately showed that there is a “negative” association between armed forces rate and regime type. The term negative is being loosely used because regime type is a categorical variable. However, the armed forces rate decreased as the regime type level of democracy increased. This association is reasonable as a less democratic and totalitarian regime would need more individuals in society to hold onto power. Also, in a less free regime there may be fewer opportunities for employment relatively which would increase the armed forces rate as a percentage of the labor force.

Strengths

The data are collected from comprehensive and respected sources.

Limitations

There are potential confounders or moderator variablesnot available in the data that may affect either armed forces rate or regime type. There are geographic, geopolitical, cultural and historical variables that potentially affect any given regime’s armed forces rate or regime type. Also, the number of observations was relatively small although the data included the population of all available data.

Recommended further research

The results of the analysis can be further explored with new variables that could be potential moderators, like a categorical variable indicating region of the world. Also, this is just a snapshot of the relationship and could be looked at historically. 

[smile8bit]

[smile8bit]

Assignment 9

In this assignment, an ANOVA was performed on variables Armed Forces Rate and Regime Type while testing for potential moderator variable, Income Level. Performing the ANOVA alone with explanatory variable Regime Type and response variable, Armed Forces Rate, there are significant differences in the means of the Armed Forces Rate per Regime Type. In particular, Regime Type 1 or autocracy has an mean Armed Forces Rate of 3.1074. Regime Types 2, 3, 4, and 5 all have Armed Forces Rate means significantly different from Regime Type 1 ranging from 1.04 to 1.4. 

When testing for potential moderator Income Level where Income Level is a categorical variable describing the relative average gross national income per capita of a country. An Income Level of 1 would indicate a country in the lower income level where GNI per capita is less than $1,025 annually. And the upper income level of 4 indicates a GNI per capita of $12,475 annually. The results of this analysis were that Income Level did not moderate the association between Armed Forces Rate and Regime Type much. In cases where the results were statistically significant, the results reflected the original association that Armed Forces Rate and Regime Type had. In other words, autocracies had a significantly higher Armed Forces Rate regardless of Income Level. 

There was one Income Level for which the relationship differed from the original relationship. For upper middle income nations, the open democracies and autocracies had significantly different Armed Forces Rate means, 0.6508 and 2.978 respectively.

Because the association between Regime Type and Armed Forces Rate and Regime Type and Income Level are similar, that is for Regime Types 2, 3, and 4 the means for Armed Forces Rate and Income Level are not statistically significantly different. If Income Level moderates the relationship between Armed Forces Rate and Regime Type it is observed only in one Regime Level Type. 

[smile8bit]

Armed Forces Rate By Regime Type ANOVA Results.

An ANOVA was performed where the categorical explanatory variable was the regime type. The quantitative response variable used was the armed forces rate.

Null Hypothesis - The armed forces rate is equal across all regime types.

Alternate Hypothesis - The armed forces rate is not equal across all regime types. 

The results from the analysis show that the F-statistic is 7.50 and the probability of getting such a value is less than 1 in 10,000, so the result is significant enough that the null hypothesis can be rejected.

Further, a Duncan post-hoc test was performed to determine the difference between each pairs of means from regime types.And as the table suggests the armed forces rate for regime types 2, 3, 4 and 5 are not significantly different. However, each of the armed forces rate for regime types 2, 3, 4 and 5 are significantly different from regime type 1.

For reference, regime types:

1 - autocracy

2 - closed anocracy

3 - open anocracy

4 - democracy

5 - full democracy

[smile8bit]

Analysis of Life Expectancy and Armed Forces Rate Using Pearson Correlation. 

The graph above is a scatter plot between Armed Forces Rate(as a percentage of the labor force) on the x-axis and Life Expectanc at birth(measured in years)y on the y-axis. The table are the results produced by a SAS using Pearson correlation. The correlation coefficient was calculated as 0.1862 and the associated p-value is 0.0166 making the results statistically significant. The correlation coefficient is relatively low and positive, meaning the the association between life expectancy and armed forces rate is positive but not strong.  In other words, given a country, a higher armed forces rate is weakly associated with a higher life expectancy. 

[smile8bit]

This graph is a way to visualize the regime types that were and were not significantly different. The bars with an 'A' were not significantly different from one another but were significantly different from the bar under 'B'. 

[smile8bit]

Table of Chi-Square Test Comparisons Between Regime Type. 

Using the Bonferroni method, a p-value of 0.005 was necessary for a chi-square test statistic to be significant. Performing comparison between all possible pairs of regime types, the results were that regime types 1, 2, 3, and 4 did not associate with different income groups. Regime type 5, however, was significantly different from each other regime type.