courseradataanalysistestblog

[courseradataanalysistestblog]

Week 8 - Moderators

This week I will be testing a potential moderator between Female Employment Rate and Income per person and I chose the variable urban rate as potential moderator as many countries that have high female employment rates are not very industrialized and women are not necessarily educated or even receiving any income from their labor. Since these are two categorical variables I’ll be using a chi-square analysis.

I added this line of code to my SAS program:

PROC SORT; BY UrbanRateGroup; PROC FREQ; TABLES IncomeperpersonGroup*FemaleEmploymentRateGroup/CHISQ; BY UrbanRateGroup; RUN;

For urban rate group 1 we can’t reject the null hypothesis for association of Higher Income with Higher female employment rate.

For urban rate group 2 we see a higher value but we still can’t reject the null hypothesis for association.

For urban rate group 3 we see a low p value and we scan’t reject the null hypothesis for association.

We can conclude that Urban rate does not moderate the relationship between Income per Person and Female Employment rate .

[courseradataanalysistestblog]

Week 7 - Pearson Correlation

This week’s assignment is to conduct a Pearson Correlation test and find the correlation coefficient between two quantitative variables. I’ve chosen the variables Female Employment Rate as my explanatory variable and Income Per Person and Internet User Rate as my response variables.

As we can observe on the table the associations between Female Employment Rate – Income per person and with internet user rate are not statistically significant. Female Employment Rate and Internet User rate have a negative linear relationship.

The scatter-plots presented on previous weeks both suggest a relationship, however, it seems they fail to adhere to a linear form which would explain the weaker linear relationship coefficient.

[courseradataanalysistestblog]

Week 6 - Chi Square Test for Independence

This week’s assignment is to conduct a Chi-Square test for independence. I’ve turned two quantitative variables into categorical groups: Female Employment Rate which is my explanatory variable (categorized in 3 groups: 1=Low, 2=Neutral, 3= High) and Income per Person – my response variable (also 3 groups: 1=Low, 2=Neutral, 3= High).

Chi-Square Test for Independence

H0: There is no association between Female Employment Rate and Income per person.

H1: There is an association between Female Employment Rate and Income per person.

As you can see, after conducting the test I can reject the Null hypothesis that there is no association between these two variables (p=0.01). However, it is still not clear as to why these differences in means exist and Post-Hocs are necessary.

Post-Hoc Tests

Since I am making 3 comparisons the necessary Bonferroni adjustment for the p value is 0.17.

Test for Group 1 and 2

As you can see by the result we can’t reject the null hypothesis that there is no association between group 1 and 2.

Test for Group 1 and 3

In this case we can reject the null hypothesis since p=0.07 < 0.17 (p with Bonferroni adjustment.)

 Test for Group 2 and 3

Again we can reject the null hypothesis for Group 2 and 3 since p=0.005 < 0.17

This graph illustrates the differences in Groups. Post hoc comparisons indicate that income per person is statistically similar among Low to Neutral Female Employment Rate and significantly different in High Female Employment.

[courseradataanalysistestblog]

Week 5 - Data Tools

ANOVA Testing

To examine the association between female employment Rate (quantitative response) and the democratic score ( categorical explanatory), an ANOVA was conducted.

The results revealed that the null hypothesis must be accepted because the p value is 0.068 ( >0.05) so there is no association between Female Employment Rate and Democratic score. The lowest mean was found when Democratic score is -10 and the highest when -1.

Post Hoc ANOVA Analysis – Duncan Test

The Duncan test revealed that Groups -1, 1, 4, -5, 2,-9,6 and -2    are significantly different from all the others and exhibit higher rates of female employment Rate. All other comparisons were statistically similar.

[courseradataanalysistestblog]

Week 4 - Graphics

In this week’s assignment the goal was to provide visual representation of the data. Here is the portion of the code I used for creating the frequency charts, scatter plots and Bar Chart:

FREQUENCY CHARTS

Female employment Rate

The graph is unimodal, with its highest peak at the median category of 50% and the lowest in the category of 10%. It seems to be slightly skewed to the right as there are higher frequencies in lower categories than the higher categories.

Income per person

The graph is unimodal, with its highest peak at 0-500 $. It is skewed right which shows how the majority of people earn less income.

Urban RateGroup

This graph is unimodal, with its highest peak at the category of 6 (60-70% Urban Rate) It seems to be skewed to the left however the skewness is not pronounced.

Internet User Group

This graph is trimodal, with its highest peaks at the category of 5%, 20% and 50%. It is skewed to the left which means overall there’s a lower % of internet users.

Relationship Trends

Using scatter plots and in a few cases Bar Charts I tested the relationship between Female Employment Rate and Income per person, internet user rate, urban rate and the democracy score.

Female Employment rate and Income per Person

As Female Employment Rate and Income per person are both quantitative I used a scatter plot graph. The variables are very clustered and the relationship is not very clear, however, if we draw a regression line there seems to be a positive relationship.

Female Employment and Internet User Rate

Again when looking at Female Employment and Internet Users there seems to be a positive relationship, but not a strong one as the deviation is very large.

Female Employment and Urban Rate

We can see that the scatter graph does not show a clear relationship between the two variables.

Female Employment and Democracy Score

Since the Polity score is a categorical variable in this case I used a C-Q Bar Chart.

The relationship is not clear but there seems to be a positive relationship as Democracy Score seems to be higher when Femlae Employment rate is also higher.

[courseradataanalysistestblog]

Week 3- Data Management

This week I tinkered my Frequency Output by setting aside missing values and grouping variables within individual variable.

Missing Values

I’ve added the statement IF incomeperperson = . THEN IncomeperpersonGroup = .; and repeated for the other variables. Here is an example below:

Grouping Variables

I collapsed the responses for Incomeperperson, femaleemployrate, urbanrate, internetuser and polityscore.  

For income per person, approximately 60% of countries live below an Income of 1500 per person.

for female Employment approximately 55% of countries have a 50-60 Female Employment Rate.

For Urban rate group 6 was the most popular which means between 60-70%.

For internet users approximately 50% of countries have a rate below 30.

Finally for the democratic score the most frequent value is 10 with 20%  but 41% of countries have a percentage of 8 and higher and 29% have a negative score.

[courseradataanalysistestblog]

Week 2 - Running First Program

The primary objective for this analysis is trying to find the relationship between income per person and the female employment rate per country.

As a secondary objective I seek to find out if there is any correlation between female employment rate and Urban Rate, Internet User rate and political score.

Below is the code that I wrote in SAS: notice I categorized Income per Group (the same was done with Female Employment Rate, Urban Rate and Internet User Rate).

Of the sample, approximately 55% of countries have a 50-60 Female Employment Rate, 30% fall below 40 and about 15% go above 60. Missing Frequency is 39.

As for the Income per person, approximately 60% of countries live below an Income of 1500 per person, 34% between 1500-6500 and  only 6% above 8500. Missing Frequency is 23.

As for the Urban Rate, approximately 30% are bellow a 40 urban rate (Group 1,2 and 3), 40% between a 50 and 70 rate (Group 4,5 and 6) and aprox. 31% are above 80 (Group 7,8 and 9). Missing Frequency is 10.

For Internet Users, approximately 50% of countries have a rate below 30. Only 14% have a rate higher than 80. Missing Frequency is 21.

Finaly for the Democratic score. Approximately 29% of countries have a negative score and 41% have a percentage of 8 and higher with 10 being the most frequent. (20%). Missing Frequency is 52.

[courseradataanalysistestblog]

Week 1 - Exploring Research Questions on GapMinder

With the purpose of improving my skills in data science, I registered in the Data Management and Visualization Course by the Wesleyan University through Coursera. Throughout this blog series I will share my experience and provide updates on my progress.

In this first post, I am going to develop a research question to be addressed in the incoming course sections. As I’m about to start working on the development sector I am mostly interest in factors that affect global welfare and development, as such, I chose to work on the GapMinder dataset.

After exploring the codebook I have decided that I’m particularly curious about the female employment rate variable, more specifically how does female employment affect overall economic growth and development? I wonder how much countries excluding such a key segment of the population (as it happens on the ASEAN countries for example) will be negatively affected.

Upon an existing literature review on articles and books using the terms “Female employment and economic growth” and the “role of gender inequalities in labor” I found several relevant variables within the Gapminder codebook.

As part of the 2030’s Agenda for Sustainable Development, The United Nations 8th Goal is to Promote inclusive and sustainable economic growth, employment and decent work for all. Global unemployment increased from 170 million in 2007 to nearly 202 million in 2012, of which about 75 million are young women and men. (UNDP 2018)

A research conducted by the World Bank on the World Development indicators presented a Gender Mainstreaming Strategy launched in 2001 (Dollar and Gatti 1999; Klasen 1999) where it was highlighted that societies that discriminate by gender tend to experience less rapid economic growth and poverty reduction then societies that treat males and females more equally.

Gender inequalities often lower the productivity of labour, in both the short and the long term, and create inefficiencies in labour allocation in households and the general economy. They also contribute to poverty and reduce human well-being. (World Bank Group Gender Action Plan, 2008)

 There is strong empirical evidence for this effect caused by gender-based division which I will present on the following paragraphs.

Current gender disparities in labour force participation suggest a misallocation of talent that impedes the achievement of maximum productivity and curbs economic growth. In the United States, the past 50 years of alleviating the misallocation of talent due to discrimination (gender and race) has been responsible for 15%–20% of growth in aggregate output per worker (Hsieh et al. 2013).

Although the nature and importance of gender issues for poverty reduction and growth vary from country to country, significant gender disparities are found in all regions of the world (including in the member countries of the Organisation for Economic Co-operation and Development—OECD). These disparities tend to be greater in low-income than in higher-income countries, and, within countries, tend to be greater among the poor than the more affluent.

Recent research by Atlas and Cameron made a comparison study focusing on Asian Economy using findings from four country studies and a macroeconomic study calibrated using an average Asian economy. Some of the most relevant patterns and findings include:

i)    A large proportion of women currently out of the workforce state a desire to work but are constrained by various social and cultural norms.

ii)    Proportionately less women enter the labor force than men and retire earlier.

iii)    Higher or intermediate education levels are associated with lower female labor force participation on average, except in the Republic of Korea, as a woman’s choice to work may be related to income level, other household income, and social stigma and norms.

Another relevant pattern is the existence of a U-Shaped relationship between the women’s labour force participation and economic development. This has been discussed and documented in a wide set of papers, including Sinha (1967), Durand (1975, p. 131), Psacharopoulos and Tzannatos (1989), Goldin (1995) and Horton (1996).

For very poor countries, female labor force participation is high, and women work mainly in farm or non-farm family enterprises. Development initially moves women out of the labor force, partly because of the rise in men’s market opportunities and partly because of social barriers against women entering the paid labor force.

However, as countries continue to develop, women’s education levels rise, and women move back into the labor force as paid employees holding mainly white-collar jobs. (Mammen and Paxson 200, Journal of Economic Perspectives)

The linkages of gender to growth through human capital are pervasive and powerful. They involve both males and females, but women are typically at a disadvantage compared to men;

Most authors emphasised the importance of improving women’s rights, resources, and voice. Some examples include:

1-      Improved employment opportunities and higher incomes for women and their families. Educated, healthy women are more able to engage in productive activities, find formal sector employment, and earn higher incomes and greater returns to schooling than their counterparts who are uneducated or suffer from poor nutrition and health. (World Bank Report 2012)

2-      The ability to adopt new technology and respond to economic change. Better-educated women are more able to profit from new forms of technology and the opportunities presented by economic change than are less educated women. ( Klasen, S. and Lamanna, F., 2009)

3-      Time poverty created by poor infrastructure. In many settings— especially in low-income countries and among the poor in all countries—women work many more hours per day or week than men. This limits their ability to engage in income generating activities and to participate in community or national decision making. Because the gender-based division of labour extends to children, girls are often kept out of school to help with household work. (International Labour Organization, 2014)

4-      The quality of governance. Good governance is critical for sustainable development. A growing body of evidence suggests that gender equality in rights and resources is associated with less corruption and better governance. Although the correlation between gender and corruption may reflect the exclusion of women from positions of power, and thus from the opportunity to engage in corrupt practices, (Stojanović, Ateljević, Stević 2016)

Based on the above findings and its relation to the Gapminder variables, I hypothesize that female employment will have an association with income per person.

Moreover, based on the author´s recommendations I also hypothesize that female employment rate will have an association with internet user rate, urban rate and the political score.

References

“Gender Equality as Smart Economics: A World Bank Group Gender Action Plan.” Small Change or Real Change?, 2008, pp. 93–105)

“Sustainable Development Goals .:. Sustainable Development Knowledge Platform.” United Nations, United Nations (2018) , www.sustainabledevelopment.un.org/

“Engendering Development—Through Gender Equality in Rights, Resources, and Voice”. World Bank Policy Research Report (2000), New York: Oxford University Press.

Goldin, Claudia. (1995), “The U-Shaped Female Labor Force Function in Economic Development and Economic History,” in T. Paul Schultz, ed. Investment in Women’s Human Capital. Chicago: University of Chicago Press.

The World Bank (2012), World Development Report: Gender Equality and Development, p. 5.

Klasen, S. and Lamanna, F. (2009), “The impact of gender inequality in education and employment on economic growth: New evidence for a panel of countries,” Feminist Economics, 15: 3, pp. 91-132 (as retrieved from UN Women, Progress of the World’s Women 2015-2016: Transforming economies, realizing rights Chapter 4, p. 199).

International Labour Organization (2014). Global Employment Trends 2014: Risk of a jobless recovery? p. 19, http://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/-publ/documents/publication/wcms_233953.pdf

Mammen, Kristin, and Christina Paxson. (2000) “Women's Work and Economic Development.” Journal of Economic Perspectives, vol. 14, no. 4, 2000, pp. 141–164.