To perform logistic regression to analyze the effect of a one unit increase in the Syndrome on a country’s likelihood to be in the “worse outcomes” category for each significant variable, we dichotomized each of the variables in this appendix to define “worse outcomes” (see chapter 7). This appendix specifies the split between better and worse outcomes that we chose for each variable and justifies the selection. For continuous variables, the number at which the variable was split indicates that that number and below would be marked as different from all values above that number. If the variable’s directional meaning is that higher scores are worse, then the lower numbers are assigned 0 and the higher numbers are assigned 1 for the logistic regression. Alternatively, if the variable’s directional meaning is that lower scores are worse, then the lower numbers are assigned 1 and the higher numbers are assigned 0 for purposes of the logistic regression.
Our general rules of thumb for demarcating “better” from “worse” outcomes are as follows: (1) if a continuous variable’s histogram showed two clear separate distributions, then we split between those; (2) if a continuous variable’s histogram was extremely skewed and one or two adjacent bins had the highest peaks, then we created a split to ensure that those bins were grouped together; (3) if a continuous variable did not have a clear split in its histogram, then we split at the mean; (4) ordinal variables with an even number of scale points were split in half; and (5) ordinal variables with an uneven number of scale points were split with one extra scale point in the “0” group so that the “worse” outcome category was smaller than the “better” outcome category. Rare exceptions to these rules are discussed, when applicable.
Dimension 1. Political Stability
1.1 Fragile States Index: Split at 70.45, approximately at the mean.
- Security, Stability, and Legitimacy Factor: Split at 1.67, natural split in the histogram.
1.2 Government System and Effectiveness Factor: Split at 2.171, mean + ½ standard deviation. We added half a standard deviation to the mean because this would make the worse outcome higher than just above average which is a stricter criterion.
- Regime Types: Split 1–3 and 4. This split is different from our rule, but we chose this for theoretical reasons, namely, that values 1–3 are all different levels of autocracy and 4 is the only scale point that indicates that there is some sort of democracy.
1.3 World Bank Corruption: Split at −0.13, approximately at the mean.
1.4 World Bank Rule of Law: Split at −0.13, approximately at the mean.
- Private Property Rights: Split at 41.38, approximately at the mean.
1.5 Lack of Freedom Factor: Split at 0, approximately at the mean.
- Civil Liberties: Split at 6.67, natural split in the histogram.
1.6 Freedom to Establish Religion: Split between 5 and 7.5. This is a factor variable with possible values of 0, 2.5, 5, 7.5, and 10. The general rule is that a factor variable with an uneven number of scale points are split with one extra in the “0” group. But here only one observation has the lowest value (0), so the split was made so that 0, 2.5, 5 = 1, and 7.5, 10 = 0.
1.7 Percent of Seats in Parliament Held by Women: Split at 21.56, approximately at the mean.
- Government Participation of Women: Split 0–2 (“better”) and 3–4 (“worse”).
Dimension 2. Security and Conflict
2.1 Violence and Instability Factor: Split at 3.794, mean + ½ standard deviation. We added half a standard deviation to the mean. This would make the worse outcome higher than just above average which is a stricter criterion.
- Political Stability and Absence of Violence/Terrorism and Freedom of Domestic Movement Factor: Split at −1.5, natural split in the histogram.
- Trafficking of Women: Split 0–2 and 3–4.
- Political Terror Scale: Split 1–3 and 4–5.
2.2 Societal Violence Scale: Split 1–3 and 4–5.
2.3 Military Expenditure and Weapons Importation Factor: Split at 0, natural split in the histogram.
- Access to Weapons: Split 1–3 and 4–5.
2.4 Monopoly on the Use of Force: Split 1–5 and 6–10.
2.5 Global Terrorism Index: Split at 0.67. We note that the bin in the leftmost side of the histogram has the highest frequency and we used the upper limit of the bin as the cutoff.
- Terrorism Injury and Violent Conflict Factor: Split at 0, both natural split in the histogram and also approximately the mean.
- Terrorism Incidents and Internal Conflict Factor: Split at 0, both natural split in the histogram and also approximately the mean.
- Terrorism Impact: Split at 1.25.
We note that the bin in the leftmost side of the histogram has the highest frequency and we used the upper limit of the bin as the cutoff.
- Overall Index of Disappearances: Split 0–5 and 10. The three possible scale points are 0, 5, and 10.
2.6 Perceptions of Criminality: Split 2–3 and 4–5.
Dimension 3. Economic Performance
3.1 Food Security: Split at 57.92, approximately at the mean.
3.2 Reliance on Agriculture and Lack of Prosperity Factor: Split at 0, approximately at the mean.
3.3 GDP per Capital PPP 2017 (log transformed): Split at 9.32, approximately at the mean.
3.4 Poverty and Economic Decline: Split at 5.78, approximately at the mean.
3.5 Wealth Infrastructure and Economic Freedom Factor: Split at 0.88, approximately at the mean.
3.6 Global Competitiveness Index: Split at 70, approximately at the mean.
3.7 Economic Inequality Factor: Split at 0, approximately at the mean.
3.8 Female Labor Force Participation: Split at 36.7, natural split in the histogram.
Dimension 4. Economic Rentierism
4.1 Natural Resources as Percentage of GDP: Split at 3.33. Note that the highest bar in the histogram stood on its own.
Dimension 5. Health and Well-Being
5.1 Healthcare Access Factor: Split at 0, approximately the mean.
- Percent of Pregnant Women Receiving Prenatal Care: Split at 90, approximately the mean.
- Sustainable Society Index Human Well-Being: Split at 7, close to the mean.
5.2 Health Expenditure per Capita: Split at 1,333.
We note that the two bins in the leftmost side of the histogram had the highest frequencies so we grouped these bins together.
- Health Expenditure as Percentage of GDP: Split at 5.83 due to a natural split in the histogram.
5.3 Preventable Death Factor: Split at 0, approximately the mean.
- Illness and Mortality Factor: Split at 0, approximately the mean.
- Life Expectancy at Birth for Females: Split at 73.7, approximately at the mean.
- Maternal Mortality Rate: Split 0–2 and 3–4.
- Deaths Due to Diarrhea of Children Under Five: Split at 2.
We note that the two bins in the leftmost side of the histogram had the highest frequencies so we grouped these bins together.
5.4 Total Alcohol Consumption per Capita: Split at 6, approximately the mean.
5.5 Prevalence of HIV Among Women Ages Fifteen and Over: Split at 45 due to a natural split in the histogram.
5.6 Global Hunger Index: Split at 17.13, approximately at the mean.
5.7 Female Genital Cutting/Mutilation (FGM): Split 0–1 and 2–4. This split is different from our rule, but we chose this for theoretical reasons, namely, that 0–1 scores indicate that FGM rarely or never occurs, which we defined as the only acceptable “good” score for this variable.
Other variables:
- Percent Under Five Who are Stunted: Split at 20, due to a natural split in the bimodal histogram. The mean is 24.40 and the median is 23.90, so the split is lower than these values.
- Prevalence of Wasting Percentage Under Five: Split at 5, due to a natural split in the histogram. The mean is 6.26 and the median is 4.90, so the split is between these two values. There are two “distributions” in the histogram—one more evenly distributed and one right-skewed distribution.
- Percent Under Five Who Are Underweight: Split at 10, due to a natural split in the histogram. The mean is 12.57 and the median is 11.10, so the split is lower than both these two. There were two right-skewed “distributions” in the histogram.
Dimension 6. Demographic Security
6.1 Mother’s Mean Age at First Birth: Split at 23.95, mean.
6.2 Demographic Pressure: Split at 5.82, mean.
6.3 Total Fertility Rate: Split at 3.67, due to a natural split in the histogram. The mean is 2.82 and the median is 2.34, so the split is higher than both of these values. This histogram is also bimodal, but also right-skewed, and the second peak is much lower than the first.
6.4 Contraceptive Prevalence: Split at 49.42, mean.
6.5 Youth Risk Factor: Split at 0.35, due to a natural split in the bimodal histogram. The mean is .412 and the median is .415, so the split is lower than either of these values.
Dimension 7. Education of the Population
7.1 Average Years of Schooling: Split at 8, close to the mean. We did not choose the precise mean because we wanted years of school through the completion of primary school to be grouped together.
7.2 Access to Basic Knowledge: Split at 90. We used the natural split in the histogram.
7.3 Access to Information and Communications: Split at 60, due to a natural split in the histogram.
7.4 Overall Literacy for Fifteen and Over of Population: Split at 95.
We used the natural split in the histogram.
7.5 Female Literacy Rate Ages Fifteen to Twenty-Four: Split at 95.
We used the natural split in the histogram.
7.6 Survival Rate to the Last Year of Primary School for Females: Split at 90.
We used the natural split in the histogram.
Dimension 8. Social Progress
8.1 Human Development Index: Split at 0.7, approximately the mean.
8.2 Gender Inequality Index: Split at 0.4, approximately the mean.
8.3 Happiness Index: Split at 5.67, due to a natural split in the bimodal histogram. The mean is 5.36 and the median is 5.24, and the split is a little above these two values.
8.4 Percentage of Pensionable-Age Persons Receiving Social Security or Pensions: Split at 79.9. This split is different from our rule, but we chose this for theoretical reasons, namely, that we did not consider the value for this variable to be “good” unless 80 percent or more of pensionable-age persons were receiving their benefits.
- Social Safety Nets: Split 0–5 and 6–10.
8.5 Percent of Population with Access to Electricity: Split at 93.3.
We note that the bin on the right side of the histogram has the highest frequency and we used the lower limit of the bin as the cutoff.
8.6 Hofstede Individualism Score: Split at 39, approximately at the mean.
8.7 Gender Gap Index: Split at 0.7, approximately the mean.
8.8 Discrimination and Violence Against Minorities: Split at 5.5, due to a natural split in histogram.
- Religious Tolerance: Split 1–3 and 4. This split is different from our rule, but we chose this because more than 60 countries scaled at a 4, so the typical split, 1-2 versus 3-4, would have produced uneven sample sizes in each of the two groups.
8.9 Legal Declaration of Gender Equality: Split 0–1 and 2.
- Government Framework for Gender Equality: Split 0–2 and 3–7. This was partially chosen because there are many scale points in this distribution. We considered using the continuous variable rule and then found a clear split in distributions in the histogram (this is where we split the variable). The split is close to the mean. The mean is 2.43 and the median is 2.00, so the mean is right in between the split.
Dimension 9. Environmental Protection
9.1 Water and Environmental Well-Being Factor: Split at 2, due to a natural split in the histogram.
9.2 Air Pollution Factor (Without Outlier): Split at −0.10, approximately at the mean.
9.3 Biodiversity and Pest Regulation Factor: Split at 0, approximately at the mean.
9.4 Global Climate Risk Index
- Outdoor Air Pollution Attributable Deaths: Split at 40, approximately at the mean.
- Air Quality: Split at 87, mean + ½ standard deviation. We added half a standard deviation to the mean. This would make the worse outcome higher than just above average which is a stricter criterion.