Saturday, February 28, 2015

Fertility Rates and Women's Political Rights

By Natasha Wheatley

The impact of fertility rates on nations has been evaluated in comparison to many women’s issues.  The likely logical narrative is that as women gain more power, access to family planning services and contraceptive devices increases, and therefore the fertility rate across a nation decreases. I choose to instead evaluate the argument that fertility rates may be affecting women’s political rights within a country. In the first scenario, fertility rates are the dependent variable, they are only considered as the factor that has been acted upon. The key assumption here is that fertility rates are a byproduct of lack of power or self-determination for women and not an impediment to those things in the first place. Rights are too rarely given freely to the people and they must be fought for but how many women lack the ability to demonstrate or work for those rights because their roles as mothers prevent them from ever finding the time to get engaged.
The potential role of fertility rates has at least been noted by the academic and practical communities of International Relations, Political Science, Anthropology and many others, specifically in the subset of women’s issues. The academic world has attempted to address the link of fertility rates to a variety of issues: obvious ones, such as infant and maternal mortality, education for girls, the availability or use of family planning services and perhaps less obvious links such as economic stability, economic growth, state security, and democratization.2 But there has been a critical lack of research relating to specific issues of women’s political rights and the fertility rates. Therefore, there is a glaring need to evaluate the relationship from the perspective I propose. We need to look at the lack of a potential barrier to engagement, fertility rates, and the potential impact it has on women’s political rights
I evaluate this concept through a simple standard regression process of international data from the Quality of Governance dataset to compare fertility rates to scores of women’s political rights inside countries. Figure 1 shows the data for each country and includes the lowess line which shows the pattern of where an estimate of political rights would fall across fertility that would minimize its collective ‘wrongness’ from the real data. This visualization reflects the chance that a potential correlation could change directions at certain level of fertility.


Figure 1. Fertility Rates and Women’s Political Rights

 Recognizing the work that other factors can play in changing both women’s political rights score and fertility rates, I included the additional variables of income2, an index of women’s social rights3, and female education levels to the quantitative analysis. In the appendix is a brief table summarizing each of the variables that were included in the analysis. Table 1 shows the complete result of the analysis and displays four different models that reflect correlations between these variables. A most basic analysis labeled “Model 1” shows a small correlation that is rendered unreliable by further examination. We cannot be confident enough4 that this data is not just a random occurrence as opposed to an actual correlation when no other variables are included in the analysis. Model 3 of the table reports an even greater lack of confidence that there is a correlation between these two variables because it accounts for the other potential influences and results in an even lower level of confidence in the reported correlation strengths.


Table 1. Affecting Women’s Political Rights

DV: Women’s Political Rights
 Model 1
Model 2
Model 3
Model 4
Fertility Rate
-0.04
-0.32*
0.05
0.17

(-1.70)
(-2.37)
(1.20)
(1.16)
Fertility Rate Squared

0.04*

-0.02


(2.08)

(-0.85)
GNI Per Capita


-0.02
-0.01



(-0.38)
(-0.26)
Avg. Schooling(F)


0.01
0.01



(0.64)
(0.59)
Women’s Social Rights


0.23***
0.24***



(5.30)
(5.32)
R^2
0.02
0.04
0.23
0.23
Note: OLS estimates with t-stats in parentheses.* p < 0.05, ** p < 0.01, *** p < 0.001 Source: Quality of Governance

For Models 2 & 4, a slightly different approach to the raw data was taken. Figure 1 suggested that the impact of fertility rates may not be constant across all values so to reflect that possibility an additional variable, the square of the fertility rate, was added to the analysis. A change in sign of the observed effect (positive or negative) between the fertility rate and the fertility rate squared would provide evidence of this type of correlation.
 And in Model 2, we see just that: that the fertility rate has a negative correlation with women’s political rights scores while the fertility rate squared has a positive correlation with women’s political rights scores. In other words, at first, as the fertility rate increases, women’s political rights for that country decrease but at certain point (highlighted in Figure 1) as the fertility rate continues to increase, the political rights' scores begin to increase also. We can also be confident that, if these were the only variables in a country, this correlation would not have occurred by random chance. Alas, other factors must be taken into account and Model 4 shows us that when we do that, the confidence we once had in Model 2 is wiped away and there is not strong enough evidence of a real correlation between Fertility Rates and Women’s Political Rights.
There is some other interesting information to take away from this model. For one, the obvious critique of my argument that in fact it is not fertility rates affecting female political rights but rather the other way around is struck down because we can observe no statistically significant correlation between the two suggesting that it is more plausible that neither affects the other. It is also interesting to point out that two of other included variables are shown to have no evidence of impact on Women’s Political Rights: both GNI per capita and the average years of schooling for women in a country do not seem to have an effect on women’s political rights in a country. Results that seem to counter popular thought on the issue.






Notes:
1.  John Ross & John Stover. "How Increased Contraceptive use has decreased Maternal Mortality". Maternal & Child Health Journal. Sep 2010, Vol. 14, Issue 5.
 Marayumo, Akiko & Yamamoto Kazihiro "Variety Expansion & Fertility Rates". Journal of Population Economics. Jan 2010 Vol. 23, Issue 1.
Colclough, Christopher. "What are you doing to provide us with an education?" UN Chronicle Jun/Aug 2004, Vol. 41, Issue 2.
2. To better account for the change in what constitutes meaningful differences in income between developed and developing nations, the GNI per capita of a country was logged and then included in the analysis.
3. There are no factors that are included in both the Women’s Political Rights and the Women’s Social Rights Index. Concerns of multicollinearity between Women’s Social Rights and Women’s Political Rights as well as between Women’s Social Rights and other control factors can be alleviated by a similar collinearity test of the variable, the table of results is included in the Appendix but does show a score for women’s social rights that is concerned within the norm and should alleviate any concerns of multicollinearity in the regression.
4. Confidence or lack of confidence in our results is determined by the test statistic and its corresponding p-value which is a calculation of the likelihood that the resulting correlations would appear by random chance and not because there is an actual causal linkage.

Appendix

Table of Variable Summary Statistics
Variable
Mean
Std. Dev.
Min.
Max.
Summary
Women's Political Rights
2.06
0.47
1.00
3.00
Index Score of Rights such as: voting, running for office, participating in political parties, etc. High Score=High Rights
Fertility Rate
2.81
1.41
1.15
7.11
Average Births per Woman in a Country
Fertility Rate Squared
9.86
10.27
1.32
50.62
Average Births per Woman in a County
Squared
Income (GNI per Capita Logged)
8.44
1.61
4.50
11.36
Gross National Income per Capita taken as Log to account for impact in changes to the figure at different levels
Average Schooling Years, Female (25+)
7.39
3.25
0.75
13.29
How many years of Schooling on average a woman 25 or over has in a given country
Women's Social Rights
1.24
1.04
0.00
3.00
Index Score of Rights such as: property rights, marriage & divorce rights, transit freedoms, etc. High Score= High Rights


Table of Multicollinearity Examination

Variable
VIF
1/VIF
Fertility Rate
33.45
0.029896

Fertility Rate Squared
28.74

0.034792
Income (Represented as Log of GNI per capita)
3.50
0.285581
Years of Education for Women Age 25+
3.46
0.289003

Women’s Social Rights
1.71
0.583522
Mean VIF
14.17




1 comment:

  1. Your introduction does a good job of presenting your topic and the importance of the analysis. However, I had to search for your research question. Even if you just start a new paragraph with "I choose instead to evaluate...", I think that would make the question more visible. Just make sure your question isn't lost in your set up of the topic and background information.

    Logistically, I see notes 1 and 3 in the text, but there is only one end note (and note 2 is missing from the text). And I wanted to know more about the index of women's social rights variable (specifically, what the index compiles/measures), but the end note wasn't there.

    Variables: Somewhere, perhaps in an end note, I think a more in-depth description of the outcome variable would be beneficial. You tell us that the data come from the QoG dataset, but I'd like to know what exactly goes into the women's political rights variable (e.g. right to vote, hold office, protest, etc.). Additionally, are you concerned about multicollinearity or skew? Depending on the measures of the women's social rights variable, it seems as though it could potentially overlap with education level. Your post is quite easy for the layman to read, which is great, but you could maybe just include a note that you tested for these things (and perhaps why or why not) to demonstrate to readers in the field that you have been rigorous in your analysis.

    You do an excellent job of explaining your models, particularly the use of the squared variable. However, you note that "we cannot be confident enough..." that the relationship is not coincidental. You might explain where that confidence comes from, without getting too technical. You could maybe just mention (possibly in an end note) that confidence is derived from the p-value, which refers to the probability of chance.

    With regard to the visuals, Table 1 is clear and visually appealing, with a nice use of color. This is just a personal preference, but you might consider formatting Figure 1 to match aesthetically (e.g. either with consistent colors or a standard black/grey). The figure is clear and readable as is, but it might draw the eye more if they are consistent.

    It might be helpful to find someone to proofread the post, as well. You have a very interesting post; a few grammatical tweaks would make it even more readable. Nice job!

    ReplyDelete