Beruflich Dokumente
Kultur Dokumente
In order to
analyse and explain the phenomenon of political stability, this paper draws its statistical data from the
Norris cross-national subset. However, it needs to be said that a complete analysis of all the factors
affecting political stability can neither be collected nor analysed. The following hypothesis includes the
variables which as this paper will show have the strongest effect on the dependent variable, ‘political
stability’ (Kaufmann political stability 2002).
Upfront, it needs to be said that there are a wide variety of variables that influence political stability, this
paper, however, will choose the ones which are more common than for example the amount of police per
1000 citizens or similar variables.
The hypothesis is that political stability mainly depends on the rule of law (kaufmann rule of law 2002), a
high degree of autocracy or democracy (Democ²) and the amount of trade (% of GDP; mean 1990-99)
that the country engages in.
This suggests three separate hypotheses which support the main hypothesis stated above.
Firstly, this work suggests that political stability is getting stronger when the rule of law increases. Pinherio
agrees that a lack in of the rule of law leads to a lack of legitimacy, and therefore also consequently
political stability (Pinherio 1996: 18). If the rule of law decreases, the political stability is weakened. The
rule of law has an influence on the political stability because it is an indicator for how well the executive
does its job. Even in an authoritarian regime is it important that the army follows the rules set up by the
dictator. Moreover, the variable rule of law 2002’ is an ordinal variable which goes from -2.05 to 2.03.
Secondly, the degree of political stability high if there is a high degree of democracy or
autocracy. Beetham argues that the biggest changes in the relationship between society and legitimacy
are a shift in the political or social order (Beetham 1991: 75). Strong democracies or strong autocracies
are best equipped to withstand this shift, and hence provide political stability. Therefore, states with a
small degree of democracy or autocracy are less politically stable. In audition this computed ordinal
variable goes from 0 to 100.
Finally, political stability is also dependent on economic aspects. The reason is obvious: people are
encouraged to invest and trade when they are confident in the future, and few things seem more likely to
undermine business and consumer confidence than the prospect of political unrest and sudden changes
in the economic “rules of the game” (Goldsmith 1989: 471). Therefore, trade is also an indicator for
political stability, but also a part of economic stability. A lack of trade does not only mean that there is a
lack of production, but would also suggest that there is a lack of political stability, because political
stability is only possible with economic stability. Furthermore, trade is an indicator for how developed the
society is, if there are certain goods produced which other countries cannot produce because of their lack
of technical advantage. Trade can also be seen as an indicator of how well the regime is embedded with
its neighbours. If that is not the case, than the degrees of political stability and volume of trade are low.
This is an interval variable which goes from 3.8 to 360.
This work will start the analysis by drawing a causal diagram in order to illustrate the dependency of the
dependent variable on the independent variables. Afterwards, there will be an examination of the
frequency table, in order to assure that there are sufficient cases which this paper can draw from. In
addition, there will be a correlation computed in order to illustrate that the independent variables do not
produce multicolinearity. Finally, there will be an ordinary least square regression (OLS) used between all
the variables in order to examine the strength, form and significance of this model. This is necessary to
conclude that the hypothesis stated above is true.
The aim of the causal diagram is to suggest the positive correlation between the independent variable
and the dependent variable. Moreover, this causal diagram serves to illustrate what the paper aims to
explain and which variables will be used to do so.
The second obligation of this research is to assure that there are in fact sufficient cases to work with. A
dataset in which specific variables for a variety of cases are missing would undermine the research. To
assure that this is not the case, a Frequency table will be presented (table 2). This table shows that there
are sufficient valid cases in the dataset. There is just a small amount of cases in the dataset, in which the
dependent or independent variable is missing. There is just a maximum of 40 cases missing. It appears to
be reasonable to reject one or more independent variables when there are more than 60 cases are
missing. The dataset all together has 190 cases. As a result, this work concludes that it is able to use the
chosen independent variables.
Frequency table
Trade (% of GDP)
kaufmann political kaufmann rule of mean(1990-99) -
stability 2002 law 2002 WDI/STF3 Democ²
Number of valid
177 186 156 151
cases
Number of missing
14 5 35 40
cases
Table 2
The variable Polity Combined 20-pt score 2000 was squared for this research in order to compute the
variable democ2. In doing so this paper calculated a variable which indicates the degree of stability the
regime enjoys. If the number is high, then the regime has the ability to sustain itself. This variable does
not account for democracy. It just assures that it is either a very stable democracy or a stable autocracy.
In order to carry out the test of the hypothesis it was necessary to compute the variable because no
variable with these specific features existed before.
The third important operation is a correlation. This correlation (table 3) assures that there is no
multicolinearity between the independent variables. It can be said that two independent variables are
multicolinear when the correlation between them is 0.7 or above.
Correlations
Trade (% of
GDP)
kaufmann mean(1990-
rule of law 99) -
2002 WDI/STF3 democ2
kaufmann rule Pearson Correlation
1 ,274(**) ,669(**)
of law 2002
Sig. (2-tailed) ,001 ,000
N 186 156 151
Trade (% of Pearson Correlation
GDP)
,274(**) 1 ,020
mean(1990-99)
- WDI/STF3
Sig. (2-tailed) ,001 ,814
N 156 156 137
democ2 Pearson Correlation ,669(**) ,020 1
Sig. (2-tailed) ,000 ,814
N 151 137 151
** Correlation is significant at the 0.01 level (2-
tailed). Table 3
As table 3 shows, the biggest correlations between the independent variables are less than 0.7. Therefore
it can be rejected that there is multicolinearity between the variables. The three operations in this part of
the paper reassured that the independent variables in use have explanatory power, sufficient cases inside
the dataset and do not depend on each other
The hypothesis and the statistic
In order to illustrate the dependency between the dependent and the independent variable an OLS
regression was computed in Table 4.
Coefficients(a)
Unstandardized Standardized
Coefficients Coefficients
Model R²= .817 B Std. Error Beta t Sig.
1 (Constant) -.521 .157 -3.324 .001
kaufmann rule
.624 .072 .636 8.706 .000
of law 2002
Trade (% of
GDP)
.003 .001 .142 2.711 .008
mean(1990-99)
- WDI/STF3
democ2 .005 .002 .188 2.635 .009
a Dependent Variable: kaufmann political stability
2002 Table 4
The regression equation of this model is Y= a + b1x1 + b2x2 + b3x3 + e. The Y is the independent
variable, eand in this case political stability. The a-value describes the y intercept and is in our case -.521.
Moreover, b1, b2 and b3 are the ratio of the relative predictive power of the independent variables and
X1, X2 and X3 are the ratio of the relative predictive power of the independent variables. This means that
the regression equation for this model is as follows: political stability = -.521 + .624 rule of law + 0.003
Trade (% of GDP) mean (1990-99) + 0.05 democ2 + e.
This table contains three indicators which will show that the rule of law, trade, and the level of democracy
or autocracy are in a correlation with political stability.
Firstly, the strength of this model shows that the independent variables chosen do explain the dependent
variable. The R² is 0.8e17. This indicates that 81.7% of the variation in political stability in the dataset is
explained by the independent variables. Therefore, the independent variables chosen have quite a
substantial explanatory power. However, the constant just has a small difference within; therefore other
variables would score a high R² as well.
Secondly, the form of the model indicates that the there is a positive correlation. The constant in relation
to the coefficient show that that there is a slight positive relationship between the independent and the
dependent variable. The slope of this model is just very slight because the constant is quite small. Due to
the circumstances that the maximum and minimum of the dependent variable is quite small, the
coefficient is small as well. This, however, does not suggest that the slope is so slight that the fact that
there is a positive correlation might be rejected. This is just a mathematical phenomenon and should not
be regarded as an argument against the hypothesis.
Finally, the models’ significance is strong enough to support the hypothesis. In order to reject the null-
hypothesis two indicators need to be examined: the t-ratio and the p-value. The t-ratio in all cases is
above 1.96 or below -1.96 which indicates that the null hypothesis should be rejected. The t-ratio is
calculated by dividing the B- coefficient through the standard error. The t-ratio for the rule of law is 8.706,
the t-ratio from the trade variable is 2.711 and the t-ratio of the democ2 variable is 2.635. This concludes
that the relationship between political stability and trade, rule of law and a high degree of autocracy or
democracy is significant. However, the p-value might suggest otherwise on the first glance. The
significance level of the b-coefficient (p-value) should be below 0.005 in order to reject the null hypothesis
with 99.9% certainty. This is just the case in the rule of law. The significance of the trade and democ2
variable on the other hand is 0.008 and 0.009. This is above the widely acknowledged limit if 0.005.
However, as mentioned before, the constant has values which are small and lie close together. This is the
explanation why the p-value is above 0.005. I would not be surprising to see higher values. The values of
the trade and democ2 variable however is still quite small and the t-ratio is well above 1.96 therefore the
null-hypothesis might be rejected.
This model states that if the rule of law would rise by 1 then political stability would rise by 0.624. If the
amount of trade would rise by 1% then political stability would rise by 0.003. Moreover, if democ2 would
rise by one political stability would rise by 0.005. This does not sound as it would make a big difference
but as mentioned before the variable of political stability goes from -2.42 to 1.63. Therefore, just slide
increases or decreases would have a big affect on the political stability of a country.
Conclusion
To conclude, this model was computed in order to show that political stability is mainly caused by the rule
of law, the amount of trade and the degree of democracy or autocracy in a country.
In the first part, this paper shows that there are a sufficient number of cases in the dataset. This leads to
the acknowledgement that the independent variables chosen have enough cases to continue in the
argumentation.
The second part of this work showed that there is no multicolinearity between the independent variables.
This states that no independent variable is caused by or is in a close relationship with the other
independent variable. This measure was computed in order to reject any arguments which could criticize
the dependency between the independent variables.
The final part of the paper examined a regression computed in order to prove that there is in fact a
positive correlation between political stability and the independent variables. The strength of the model
showed that the explanatory power of the independent variables is in fact sufficient to explain a
substantial amount of political stability in the dataset. The model form indicated a slope with a slight
positive correlation. The model significance indicates that we can reject the null hypothesis and that the
independent variables are significant in explaining the dependent variable. Some data got corrupted due
to the circumstance that the constant is quite weak.
All these examinations sustain the hypothesis stated in the beginning. It is not just true that political
stability grows when there is a growing rule of law, an increased amount of trade but also when there is a
growing degree of democracy or autocracy in the country. The numbers showed that the independent
variables chosen in this paper do in fact explain what political stability in a country is all about. However,
this is true using the dataset provided. It can be argued that political stability is in fact so difficult to assess
that statistical proof can always be disputed.
Bibliography
Goldsmith, A. A. (1987) ‘Does Political Stability Hinder Economic Development? Mancur Olson's
Theoryand the Third World’, in Comparative Politics, Vol. 19, No. 4: 471-480.