Real Data Analysis!
Third project for the DAND is done! We finally got into what I would call "real data analysis."
I looked at the relationship between a country's economic strength and female participation in employment and education.
Data came from Gapminder.
The original project can be found here, and the updated project here.
Introduction: Female Labor Force Participation and Economic Strength
I was looking to verify reports I have heard that female labour force participation (FLFP) was positively associated with economic strength. I also wanted to look at the type of labour they participated in and the potential relationship to their participation in education as well.
Indicator Selection and Rationale
Income (GDP) per person (fixed 2000 USD)
- Definition: Gross Domestic Product per capita in constant 2000 USD. The inflation but not the differences in the cost of living between countries has been taken into account.
- Source: World Bank
- Rationale:Identified as a standardized metric of economic strength across countries.
- Definition: Percentage of female population, age above 15, that has been employed during the given year.
- Source: International Labour Organization
- Rationale:Identified as a metric to capture female participation in the workforce.
- Definition: Percentage of all female labor that works in agriculture sector.
- Source: International Labour Organization
- Rationale:Included to allow for examination of the relationship between employment sector and economic strength.
- Definition: Percentage of all female labor that works in industry sector.
- Source: International Labour Organization
- Rationale:Included to allow for examination of the relationship between employment sector and economic strength.
- Definition: Percentage of all female labor that works in service sector.
- Source: International Labour Organization
- Rationale:Included to allow for examination of the relationship between employment sector and economic strength.
- Definition: The average number of years of school attended by all people in the age and gender group specified, including primary, secondary and tertiary education.
- Source: Institute for Health Metrics and Evaluation
- Rationale:Included to allow for examination of the relationship between educational achievement (up to tertiary) and economic strength, and the relationship between educational achievement and workforce participation, including sector.
Background Information
An International Labour Organization (ILO) working paper, Female labour force participation in India and beyond, authored by Ruchika Chaudhary and Sher Verick, provided valuable background information for the project.
Key Findings
- The log of GDP per capita (due to the skew of the data) is typically used as a measure country economic strength.
- The three sectors are the standard sectors used to capture areas of employment.
- The relationship between FLFP and economic strength (from a cross-country analysis) is varied in findings, but has been suggested to be U-shaped, such that greater FLFP is associated with both higher and lower economic strength.
- This U-shaped relationship indicates that increased participation in employment is not always a positive sign. Women may have been forced into the workforce as a result of family finacial pressure, especially in times of economic downturn, and, this participation is often at lower rates and less desirable working conditions than men.
- Greater years of participation in education may decrease employment rates among younger women, but typically education encourages FLFP once education is completed.
- Increased education is also typically associated with more favorable employment outcomes/conditions.
- Global history tells us that typical country economic development is associated with employment transition from agriculture, to industry, and in more recent years, to the service industry.
- Women’s participation in the labour market has remained relatively stable from 1993 to 2013, with women accounting for approximately 40% of the global labour force.
Research Questions
- How does economic strength vary across the globe?
- How does women's participation in the workforce vary across the globe?
- How does women's participation in education vary across the globe?
- Does the data support a theory of a U-shaped relationship between FLFP and economic strength?
- Does the data show a relationship between employment sector participation and economic strength?
- Is there a positive relationship between economic strength and education participation?
- Does the data show relationships between education participation and overall FLFP, as well as in the three employment sectors?
Substantial data wrangling and cleaning was completed to produce the final dataset. Information about this process can be found in the project links above.
How does economic strength vary across the globe?
The globe spread of economic strength (log of income per person) ranged from 4.55 to 11.35 with an average of 7.89 and a relatively normal distribution.
The spread of economic strength was also inspected across the United Nations country groupings.
Observations
Average regional income per person ranged from Africa with the lowest per person income (μ = 6.42) to Western Europe and Other with the highest per person income (μ = 10.15). The eight non-UN countries reported income data with the smallest range and a simiar mean of per person income (μ = 9.95) but a right-skewed dataset.
Variability in income per person tended to decrease as the mean for the region increased. Africa and Asia-Pacific (μ = 7.63) had the lowest means and the greatest variability, Eastern Europe (μ = 7.92) and Latin America and Caribbean (μ = 8.22) had moderate income per person and variability with the region, and Western Europe and Other had the highest mean income per person and the lowest variabilty for UN regions.
Research Question Answer
Countries that are economically strong are more likely to be found within the Western Europe and Other region, and less likely to be found in the Africa and Asia-Pacific UN regions.
Research Question 2: How does women's participation in the workforce vary across the globe?
Employment Participation
Observations
While the average FLFP of the UN regions were relatively close, ranging from 43.9% in Asia-Pacific to 52.9% in Africa, the variability of FLFP ranged greatly between the regions. Eastern Europe had the most consistency of FLFP and Africa and Asia-Pacific had the greatest variability.
In addition, because the UN regions are ordered by increasing income per person in the above chart, it possible to make some comparisons relative to income per person from the FLFP boxplots. It can be seen that as income per person increases that variability shrinks a range of approximately 20% between the countries in Eastern Europe, and then grows again with increasing income per person.
That is, if income per person is kept in mind, when considering the UN regions, this figure gives an indication of the U-shaped relationship between income per person and FLFP. (It is not possible to make a comparison between how the non-UN coutnries' FLFP compares to their income per person because the data for each variable is typically not shared with the other.)
Research Question Answer
The greatest participation of women in the workforce is found in the African and Western Europe and Other regions. For the countries that reported, females are the least likely to participate in the workforce when their country is not part of the UN, but of the UN regions, FLFP is lowest in the Asia-Pacific and Eastern Europe regions. When considering the UN regions labor force participation does appear to align with a U-shaped correlation with economic strength.
Employment Sector
While it was not confirmed whether the same countries were reporting, the number of countries reporting employment sector information was substantially reduced for the Africa, Asia-Pacific and Latin America and Caribbean regions. As such, confidence in region-specific analysi for these regions is limited.
In addition, because Eastern Europe and Western Europe and Other are comparatively over represented, caution should be made when interpreting relationships between employment sectors and other variables.
Observations
Africa had the largest proportion of women participating in the Agricultural sector, with over one-third of females employed in this sector. In all other regions the proportion of females employed in the Agricultural sector was less than 20%. Africa also had the smallest proportion of females in the Service sector, with just over half of employed women working in this sector. The remaining regions typically had more than two-thirds of females employed in the Service sector, with the highest proportion of females employed in the Service sector.
The Latin America and Caribbean, Western Europe and Other regions had the largest proportion of females working in the Service sector.
Eastern Europe differs from all other regions with the largest relative amount of females employed in the industrial sector. It also has the second highest proportion of females employed in the Agricultural sector.
(The data for the Latin America and Caribbean region was slightly adjusted to accommodate for the sum of the sectors greater than 100%.)
Research Question Answer
The type of employment females participate in appears related to economic strength, with greater participation in service industries associated with economic strength.
How does female education participation vary across the globe?
Observations
Apart from Eastern Europe, the regions were arranged in order of increasing female education participation, with Africa with the lowest female education participation (μ = 3.30 years) and Western Europe with the highest female education participation (μ = 11.24), in line with increasing economic strength.
Instead, Eastern Europe was most similar to Western Europe and Other in terms of female education participation ( μ = 11.04) and also had the least variance in female education participation, followed by Western Europe and Other. These were the only two regions that had countries with average years of education completed by females over 12.
The Asia-Pacific region had the greatest variability in female education participation.
Only two countries that are not in the UN have information regarding the particiaption of females in education.
Research Question Answer
Generally the pattern of participation in education for females follows that of relative economic strength. The region that does not follow this trend is Eastern Europe, with a far greater participation in education for females compared to the region's economic strength.
Does the data support a theory of a U-shaped relationship between FLFP and economic strength?
To test the hypothesized U-shaped relationship between a country's FLFP and economic strength the countries were split into third quantiles of economic strength - low, medium, and high.
The labor force participation was then compared between these three categories utilizing an one-way ANOVA to confirm variation across the regions.
A one-way ANOVA was conducted to confirm the presence of a difference between the categories (F = 16.5, p < 0.00001).
Comparisons were made between the three categories. Each category was found to be significantly different from the other (p < 0.01667).
Research Question Answer
The t-tests supported the presence of a U-shaped relationship between country economic strength and FLFP. However, each of the categories were shown to have different FLFP. For countries of low economic strength the FLFP was an average of 54.6%, for those of medium economic strength the FLFP was an average of 40.9%, and for countries of high economic strength the FLFP was an average of 46.3%.
Does the data show a relationship between employment sector and economic strength?
Due to its skewed nature, female participation in the agricultural sector was log transformed to create a more normally distributed statistic. The correlation was tested for significance. A negative correlation (r = -0.738, p < 0.00001) was found between the log of the rate of female participation in agricultural sector employment and a country's economic strength.
That is, countries with higher economic strength have less women working in agricultural sectors.
A scatter plot was also completed to view the log relationship between economic strength and female pariticipation in the agricultural sector.
The correlation appeared to be due to increasing variation in participation in the agricultural sector for countries of lower economic strength, with less variation for countries of higher economic strength.
There was no relationship observed between a country's economic strength and female participation in the industrial sector.
A positive correlation (r = 0.716, p < 0.00001) was found between the rate of female participation in service sector employment and a country's economic strength.
That is, countries with higher economic strength have more women working in service sectors.
Research Question Answer
There are relationships between female participation in the agricultural and service sectors and a country's economic strength. Countries of higher economic strength are more likely to have high employment of females in the service sector and low employment in the agricultural sector. The participation of females in the industrial sector ranges up to approximately 30% across countries.
Is there a positive relationship between economic strength and female education participation?
Research Question Answer
There was a highly significant positive correlation (r = 0.722, p < 0.00001) between a country's economic strength and female participation in education.
Does the data show relationships between education participation and FLFP, including in the three employment sectors?
Overall labor force participation
It was identified that the pattern of the relationship between education participation and FLFP looked similar to the relationship between economic strength and FLFP. A one-way ANOVA was conducted to confirm a difference between low, medium and high education after confirming the suitability of the data.
The test indicated that there were differences in the means for the three categories (F = 4.88, p < 0.01).
Research Question Answer
The ANOVA confirmed a similar U-shaped pattern, with the FLFP for countries of low (μ = 50.8) and high (μ = 48.7) female education higher than countries of medium female education (μ = 42.7).
Employment Sector
The significance of the correlation between female education participation and the log of agriculatural sector employment was conducted. A moderate logarithmic negative correlation (r = -0.386, p < 0.001) was found between the female education participation and female agricultural sector employment.
That is, countries with higher female education have less women working in agricultural sectors.
To view the log relationship between education and female pariticipation in the agricultural sector a scatter plot of arg and education was completed.
The correlation appeared to be due to some increasing variation in participation in the agricultural sector for countries of lower female education, with less variation for countries of higher female education.
There was no relationship observed between a country's female education participation and female participation in the industrial sector.
A significant moderate positive correlation (r = 0.413, p < 0.001) was foundbetween the rate of female participation in service sector employment and a country's female education participation.
That is, countries with higher female education participation have more women working in service sectors.
Research Question Answer
The relationships between female education participation and female participation in the three employment sectors is similar but less strong than the relationships between the employment sectors participation and a country's economic strength.
Countries of higher female education are more likely to have high employment of females in the service sector and low employment in the agricultural sector. The participation of females in the industrial sector ranges up to approximately 30% across countries.
Conclusions
Overview
The following relationships were found:
- A U-shaped relationship between a country's economic strength and female labor force participation, supporting the findings of the ILO report.
- A negative log relationship between a country's economic strength and female agricultural sector employment.
- A positive relationship between a country's economic strength and female service sector employment.
- A positive relationship between a country's economic strength and female education.
- A similar pattern of relationship, with reduced correlations between education and all employment variables
Regional Data
| Region | Income | FLFP | Agr. | Ind. | Ser. | Education |
|---|---|---|---|---|---|---|
| Africa | 6.40 | 52.9 | 34.1 | 12.3 | 50.6 | 3.3 |
| Asia-Pacific | 7.63 | 43.9 | 16.7 | 13.8 | 66.8 | 6.5 |
| Eastern Europe | 7.92 | 44.0 | 18.1 | 19.2 | 62.6 | 11.0 |
| Latin American and Caribbean | 8.22 | 46.1 | 6.5 | 15.0 | 79.4 | 7.8 |
| Western Europe and Other | 10.15 | 48.9 | 4.8 | 12.3 | 82.6 | 11.2 |
| non-UN | 9.95 | 39.8 | 6.1 | 8.7 | 85.0 | 8.8 |
Impact of Education
Based on the regional data, female education participation suggests that some of the reduced correlation strength for education results from Eastern Europe's high education participation that does not match it's lower economic strength and labor force participation. One hypothesis for Eastern Europe's increased participation in education in comparison to its relative economic strength is that female education participation rates are influenced by the countries' close location to Western European countries.
Some of the reduced correlation may also be due to the varying impact of increased education on FLFP based on age (as indicated in the ILO report), but the data was not available to investigate this.
However, the data did support the findings of the ILO report that increased education is typically associated with more favorable employment conditions with as shown by greater female education being associated with less participation in agriculture and more in service employment.
Type of Employment
The data supported the ILO findings of a relationship between economic health and the type of employment participation for females, however, it did not show a sequential progression of agricultural to industrial to service sector employment. Instead, the data showed the expected relationships for agricultural (negative) and service (positive) sector employment in relation to economic health, but no relationship for industrial employment compared to economic health.
Eastern Europe's disproportionality high participation for female in industrial employment suggests that factors beyond country economic strength and education influence the type of employment participation. One hypothesis is that political ideologies may also influence the type of work that is available within a country.
Limitations
More information was available for higher economically developed countries than lower economically developed countries and this may distort the results, especially with comparisons related to employment sectors.
While theories exist regarding the causality of the results, the data was unable to support a determination regarding the direction of the causality between female access to employment and economic strength because longitudinal data was not included. The increased education for Eastern Europe region countries suggested that increased education access may precede increased economic strength but this could not be confirmed definitively.
Another limitation of the current data is the intentional exclusion of data related to male participation in employment and education. As such it is not possible to determine whether the pattern of employment, education and economic strength for females is similar to that for the general population, or males (and therefore not a product of gender), or whether female particiapation in employment and education have an additional contribution to a country's economic strength.
Final Note
If I were looking for goals in creation of data visualizations, this would definitely be right up there: from Gapminder themselves.