Method of Calculating the Syndrome Scale Scores
We created the Patrilineal/Fraternal Syndrome scale (table AI.1) using the following variables (note that the scale date is not the date of the data; all of these scales should be assumed to have examined data for the five-year period from 2010 to 2015):
• Prevalence and Legality of Polygyny (2016), The WomanStats Project, ordinal (0–4), higher scores are worse
• Inequitable Family Law and Practice Favoring Males (2016), The WomanStats Project, ordinal (0–4), higher scores are worse
• Bride Price/Dowry/Wedding Costs (2017), The WomanStats Project, ordinal (0–10); higher scores are worse
• Women’s Property Rights in Law and Practice (2016), The WomanStats Project, ordinal (0–4), higher scores are worse
• Prevalence and Legality of Cousin Marriage (2016), The WomanStats Project, ordinal (0–3), higher scores are worse
• Age of Marriage for Girls in Law and Practice (2016), The WomanStats Project, ordinal (0–4), higher scores are worse
• Legal Exoneration for Rapists Offering to Marry Victims (2016), The WomanStats Project, ordinal (0–1), higher scores are worse
• Son Preference and Sex Ratios (2015), The WomanStats Project, ordinal (0–4), higher scores are worse
• Prevalence of Patrilocal Marriage (2016), The WomanStats Project, ordinal (0–2), higher scores are worse
• Overall Level of Violence Against Women (2014), The WomanStats Project, ordinal (0–4), higher scores are worse
• Societal Sanction for Femicide (2016), The WomanStats Project, ordinal (0–2), higher scores are worse
TABLE AI.1 Syndrome Scale Scores Scaled in 2017 for the Time Period from 2010 to 2015
| Country |
Syndrome Scale Scores 2017 |
| Afghanistan |
15 |
| Albania |
9 |
| Algeria |
12 |
| Angola |
11 |
| Argentina |
4 |
| Armenia |
8 |
| Australia |
0 |
| Austria |
1 |
| Azerbaijan |
8 |
| Bahamas |
2 |
| Bahrain |
11 |
| Bangladesh |
14 |
| Barbados |
2 |
| Belarus |
4 |
| Belgium |
1 |
| Belize |
6 |
| Benin |
13 |
| Bhutan |
8 |
| Bolivia |
6 |
| Bosnia-Herzegovina |
6 |
| Botswana |
11 |
| Brazil |
6 |
| Brunei |
9 |
| Bulgaria |
4 |
| Burkina Faso |
12 |
| Burma/Myanmar |
7 |
| Burundi |
10 |
| Cambodia |
7 |
| Cameroon |
13 |
| Canada |
1 |
| Cape Verde |
3 |
| Central African Republic* |
14 |
| Chad |
14 |
| Chile |
4 |
| China |
9 |
| Colombia |
5 |
| Comoros |
7 |
| Congo |
11 |
| Costa Rica |
4 |
| Cote D’Ivoire |
13 |
| Croatia |
5 |
| Cuba |
4 |
| Cyprus |
2 |
| Czech Republic |
1 |
| Democratic Republic of the Congo |
11 |
| Denmark |
1 |
| Djibouti |
11 |
| Dominican Republic |
5 |
| East Timor |
12 |
| Ecuador |
7 |
| Egypt |
12 |
| El Salvador |
6 |
| Equatorial Guinea |
12 |
| Eritrea |
12 |
| Estonia |
2 |
| Ethiopia |
11 |
| Fiji |
7 |
| Finland |
1 |
| France |
1 |
| Gabon |
13 |
| Gambia |
14 |
| Georgia |
8 |
| Germany |
1 |
| Ghana |
10 |
| Greece |
4 |
| Guatemala |
9 |
| Guinea |
13 |
| Guinea-Bissau |
13 |
| Guyana |
7 |
| Haiti |
7 |
| Honduras |
7 |
| Hungary |
2 |
| Iceland |
1 |
| India |
14 |
| Indonesia |
13 |
| Iran |
14 |
| Iraq |
15 |
| Ireland |
2 |
| Israel |
6 |
| Italy |
1 |
| Jamaica |
3 |
| Japan |
5 |
| Jordan |
14 |
| Kazakhstan |
8 |
| Kenya |
13 |
| Kosovo |
6 |
| Kuwait |
12 |
| Kyrgyzstan |
9 |
| Laos |
8 |
| Latvia |
3 |
| Lebanon |
14 |
| Lesotho |
13 |
| Liberia |
13 |
| Libya* |
13 |
| Lithuania |
2 |
| Luxembourg |
2 |
| Macedonia |
6 |
| Madagascar |
8 |
| Malawi |
13 |
| Malaysia |
10 |
| Maldives |
9 |
| Mali |
14 |
| Malta |
3 |
| Mauritania |
14 |
| Mauritius |
5 |
| Mexico |
6 |
| Moldova |
6 |
| Mongolia |
3 |
| Montenegro |
6 |
| Morocco |
12 |
| Mozambique |
11 |
| Namibia |
11 |
| Nepal |
12 |
| Netherlands |
0 |
| New Zealand |
1 |
| Nicaragua |
7 |
| Niger |
13 |
| Nigeria |
15 |
| North Korea |
7 |
| Norway |
0 |
| Oman |
12 |
| Pakistan |
15 |
| Palestine |
14 |
| Panama |
6 |
| Papua New Guinea |
14 |
| Paraguay |
5 |
| Peru |
7 |
| Philippines |
7 |
| Poland |
2 |
| Portugal |
2 |
| Qatar |
12 |
| Romania |
4 |
| Russia |
5 |
| Rwanda |
10 |
| Saudi Arabia |
14 |
| Senegal |
14 |
| Serbia |
5 |
| Sierra Leone |
13 |
| Singapore |
5 |
| Slovakia |
2 |
| Slovenia |
3 |
| Solomon Islands |
13 |
| Somalia |
14 |
| South Africa |
8 |
| South Korea |
6 |
| South Sudan |
16 |
| Spain |
2 |
| Sri Lanka |
9 |
| Sudan |
15 |
| Suriname |
6 |
| Swaziland |
10 |
| Sweden |
0 |
| Switzerland |
0 |
| Syria* |
13 |
| Taiwan |
7 |
| Tajikistan |
11 |
| Tanzania |
13 |
| Thailand |
6 |
| Togo |
14 |
| Trinidad and Tobago |
5 |
| Tunisia |
9 |
| Turkey |
9 |
| Turkmenistan |
8 |
| Uganda |
12 |
| Ukraine |
4 |
| United Arab Emirates |
13 |
| United Kingdom |
1 |
| United States |
1 |
| Uruguay |
5 |
| Uzbekistan |
9 |
| Vanuatu* |
11 |
| Venezuela |
6 |
| Vietnam |
9 |
| Yemen |
15 |
| Zambia |
11 |
| Zimbabwe |
12 |
*Countries with imputations. Four countries among the 176 analyzed were missing one subscale score in the 2017 scaling used in our present statistical analysis, and their values were imputed to keep the national score comparable to the rest of the nations. These nations are Central African Republic, Libya, Syria, and Vanuatu. The final result of Syria’s imputation was actually 12.5, which was rounded up to 13 for input into the database to allow mapping in chapter 3 to be possible.
The initial exploratory factor analysis (EFA) results of these eleven indicator variables extracted two factors using principal axis factoring and promax oblique rotation methods in the SPSS Statistics 24 software package. These two factors had a Kaiser-Meyer-Olkin sampling adequacy measure of .897, which is in the good range for a sample size of 176 countries.1 We used multiple imputation in R on two subcomponent scores for two countries. The pattern matrix showed some substantial cross-loadings between these two factors. The pattern matrix contains the factor loadings, and if the loadings for two factors are similar for a given variable, this indicates that the variable is measuring the same construct or factor. The correlation between these two factors was also highly significant with r = .666. These results corroborate the theoretical framework’s assertion that these mechanisms of control of females at the household level interlock as the Syndrome. These substantial cross-loadings and correlation between the two extracted factors justify our decision to combine the eleven components into one index (table AI.2).
TABLE AI.2 Exploratory Factor Analysis Pattern Matrix for the Eleven Syndrome Component Variables
| Pattern Matrix |
Factor |
| 1 |
2 |
| Polygyny 2016 |
.993 |
|
| Inequitable Family Law 2016 |
.919 |
|
| BridePrice Dowry 2017 |
.814 |
|
| Property Rights Combined 2016 |
.807 |
|
| Cousin Marriage 2016 |
.668 |
|
| Age Of Marriage Combined 2015 |
.596 |
|
| Rape Exemption 2016 |
.314 |
|
| Son Preference 2015 |
|
.890 |
| Patrilocality 2016 |
.365 |
.480 |
| Physical Security Of Women 2014 |
.458 |
.346 |
| Honor Killing 2016 |
.427 |
.287 |
We also implemented a confirmatory factor analysis (CFA) as a more robust evaluation into whether a one-dimensional scale could adequately represent the eleven subcomponents of the Syndrome. Because each of the eleven variables included in the CFA are ordinal, we use an unweighted least squares estimation technique for the analysis. In the CFA, we test whether the one-factor model fits our data sufficiently well using four model diagnostics: adjusted goodness of fit (AGFI), the normed-fit index (NFI), the standardized root mean square residual (SRMR), and the average value explained (AVE). The AGFI should be greater than or equal to 0.90, the NFI greater than or equal to 0.95, the SRMR less than or equal to 0.08, and the AVE greater than or equal to 0.5. We find that all of our measures of model fit for the one factor model meet these criteria (AGFI = 0.99, NFI = 0.99, SRMR = 0.05, AVE = 0.50). We also calculate the root mean square error of approximation (RMSEA) for this one factor model and find that it again indicates that a single scale is a good fit for the data (RMSEA = 0.00). All p-values for the coefficient estimates corresponding with the eleven subcomponents are significant in the CFA (at alpha ≤ 0.001), indicating that all of the variables are significantly related to the single factor. Overall, we find sufficient evidence to conclude that a single scale to represent these eleven subcomponents is a sufficient fit for the data and significantly describes the variation in the individual variables.
We call this overall index the Patrilineal/Fraternal Syndrome scale. On the basis of this theoretical framework, we used the following algorithm to create the Syndrome scale from the eleven subcomponent scales. Values of this index range from 0 to 16, with higher scores indicating that more of the subcomponents of the Syndrome are present. Based on the EFA, CFA, and the Cronbach alpha’s reliability estimate of .898 for these eleven indicator variables (which is in the very good range for internal consistency), we conclude this index is a valid and reliable indicator of the patrilineal/fraternal security provision mechanism as defined in this paper.
• Use Inequitable Family Law/Practice as the base for the index.2
• Add Patrilocality score to the index.
• If Brideprice/Dowry is present (score of 6 or above), add 1 to the index.
• If Prevalence and Legality of Polygyny is 3 or 4, add 1 to the index.
• If Age of Marriage Combined Law and Practice scale is 3 or 4, add 1 to the index.
• If Cousin Marriage is 3, add 1 to the index.
• If Women’s Property Rights Combined Law and Practice scale is 3 or 4, add 1 to the index.
• If Son Preference and Sex Ratio Scale is 2, add 1 to the index; if it is 3 or 4, add 2 to the index.
• If Violence Against Women (Physical Security of Women scale) is 3 or 4, add 1 to the index.
• If Societal Sanction for Women’s Murder/Femicide (Murder-Scale-1) is 2, add 1 to the index.
• If Exemption to Rape Law is present (1), add 1 to the index.