Team: Anna Blythe, Travis Meng, Hannah Smay
Background
Building on our lab from last week, we continued to study the number of threatened fish species among the 25 2014-2015 Lewis & Clark overseas programs. In order to begin to get a better idea of what causes the variance in threatened fish species among these countries, we picked three variables as the drivers of our environmental variable (number of threatened fish species) and used SPSS to understand the relationship (if any) between them. These variables include the proportion of seats held by women in national parliaments (expressed as a percent), infant mortality rate (per 1000 live births), and public spending on education (percent of GDP) 2009-2012. It’s important that we research the relationship between driver variables and environmental variables because it’s a way for us to begin untangling issues such as threatened fish species among countries. Even if our results tell us that there is no significant relationship we can apply to all countries between our chosen driver variables and environmental variable, it’s still important because we are then able to eliminate variable(s) and begin to look at other variables for a significant relationship.
Procedure
- For our first step, we chose three variables that we thought would be somewhat relevant to our environmental variable: endangered fish species. Out of the class list of driver variables we were able to choose from, we decided on using Percent Proportion of Seats Held by Women in Parliament (% Women), Percent of GDP spent on Education (% GDP/Education) and Percent Infant Mortality Rate (Per 1000 births). We then each selected one of the three chosen variables.
- We began analyzing data by first using SPSS to individually derive descriptive statistics such as mean and median for our own driver variable.
- We then proceeded to do a difference of means test where we individually derived inferential statistics such as p value.
- Next, we combined the data from our three separate variables and tried to find a correlation between them and the enviro variable.
- Lastly, we again used our combined data to look at the linear regression relationship.
Refer to https://moodle.lclark.edu/mod/page/view.php?id=204928 for specific directions on using SPSS to produce statistical results.
Results
From our statistical analysis, we found a variety of descriptive and inferential statistics. We analyzed the individual independent variables and as well as the relationship between all independent variables and our environmental variable of endangered fish species.
Figure 1: Descriptive Statistics of Driving Variables
[table]
Stat, % Women, % GDP/Education, Infant Mortality
Mean, 29%, 5%, 17.5%
Median, 24%, 5%, 7%
[/table]
By running basic descriptive statistics, we found several averages for our independent variables in the 25 countries we examined. The mean percentage of women in parliaments was 29%, the mean percentage of GDP dedicated to education was 5%, and the mean infant mortality rate was 17.5%.
Figure 2: R and p Values of Driving Variables
[table]
Stat, % Women, % GDP/Education, Infant Mortality
Sig. 2-tailed (p), .473, .465, .553
Pearson (R), -0.212, -0.189, 0.059
[/table]
All p-values of our independent variables were well above 0.05, which shows that our data is not statistically significant and our results cannot be extrapolated to the entire world. The R-value of percent of women in parliament was the most significant, at -0.212. Because this value is greater than the absolute value of two, there seems to be a negative correlation between the percent of women in parliament and the number of endangered fish fish species in this sample. However, because the p-value is so high, these results do not describe the world. The R-values of the other two driving variables are lower than the absolute value of two, and so there is no correlation.
Figure 3: Beta of Driving Variables
[table]
, % Women, % GDP/Education, Infant Mortality
Beta, -0.194, -0.162, 0.71
[/table]
When we ran a linear regression, the Beta (equivalent to the R-value) illustrated the correlation between the driving variables and the dependent variable when each driving variable stood alone. We found that the percent of women in parliaments declined below the absolute value of two, and the percent of GDP spent on education also declined. The Beta value for infant mortality increased. However, all Beta values were too small to indicate correlation.
Figure 4: Adjusted R Square for Driving Variables
[table]
Adjusted R Square
-0.073
[/table]
The Adjusted R Square illustrates the how well the driving variables explain the numbers of endangered fish species. According to the data, our chosen driving variables only account for 7% of the variation in endangered fish species. This means that 93% of the variation is left unexplained by our three chosen variables.
Discussion
Through analyzing our data with SPSS, we discovered that there is little to no correlation between our driver variables and our environmental variable (number of threatened fish species). This implies that the percentage of women in parliament, portion of GDP spent on education and infant mortality have no effect on, and are not affected by, the number of threatened fish species in the countries used for our research. We believe that this dearth of correlation is due to the lack of an association between ecological and socio-political factors in the majority of select countries used. It is also possible for us to infer from the results that indicators of human welfare do not necessarily affect or have an impact on a country’s aquatic biodiversity.
Since our research and analyses yielded no significant results, we know there were several errors and therefore improvements that we could make to our experiment. Firstly, our chosen independent variables could have been selected with more relevance to the dependent variables, we believe that these would have yielded more significant results if they were more closely related with ecological indicators (especially aquatic health). Another issue was that the countries used for sampling were not economically or geographically random because they were carefully selected for study abroad purposes by Lewis & Clark College, this led to a skewed range of data for several variables. If we were to create an accurate probability sample, we believe the results would have had more of a statistical outcome that could be applied to countries worldwide.