Chapter 3: Methodology
3.5 Data Analysis
3.5.3 Stata Software
order to determine which independent variables have no effect as a predictor on the dependent variable. The confidence interval is the value within which the coefficients are located. They provide a range due to possible variation (Lind et al., 2012).
3.5.3 Stata Software
A range of statistical software has appeared over the last decade. As technology has advanced, so too has this variety of computer software. In this way, the gap has greatly reduced between the number of mathematical calculations and their original time-consuming burden.
With the ease of use for statistical software, analysis--particularly in the case of linear regression--has become available for multiple applications and with a high tendency to be the primary option for use in research. More modern methods of quantitative analysis and social research supported by increasingly advanced software applications have greatly reduced the applicability or usefulness of older analytical techniques taught in doctoral training (Jenson, 2008).
The perception of an increasing propensity to select computer software for analysis in recent years is real. The market now offers different software options for selection based on the needs, investigation goals, or preferences of research. Commercially available software suites have played a fundamental role supporting quantitative analysis in the academic field for many years now, including SPSS (SPSS Inc., 2011), SAS (SAS Institute, 2012), and Stata (2013) (Harwell, 2014).
In the case of the regression line, personally I consider Stata to be an adequate software tool with simplicity and a user-friendly interface. Despite some cons, Stata has proven that it works quite adequately on linear regression analysis. While it could be said that Stata is relatively weak on ANOVA with regards to statistical analysis and mediocre at best for factor analysis, it has shown itself to be very strong when utilized in regression analysis, complex survey designs, and limited dependent variables (Acock, 2005).
Stata has been in the market for quite several decades now, during which time it achieved continuous expansion and improvement for statistical analysis. Stata provides a comprehensive, integrated software suite for statistical analysis, offering fast, accurate, and easy use through its point-and-click interface and strongly intuitive command interface that allows its application in economics, education, finance, business, and political science (STATA, 2018b).
Moreover, once commands are input and operations selected, the results are immediately displayed for the user to review (see figure 6).
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Figure 6. STATA interface. Adopted from “STATA,” 2018b, Retrieved from https://www.stata.com/why-use-stata/
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
For Bollen Stata's performance comes from assumptions based on theoretical concepts and structural models, wherein the associated between the parameters given are not descripted in nature but are causal (as cited in STATA manual, n.d). Following this definition, we can state that the measurement models given above are structural models, as too is the linear regression seen in figure 7 below. Based on this notion and in order to come up with the computer results for linear regression, basic models are necessary to obtain the interaction and outcomes from STATA as the example in figure 8 shows.
Figure 7. Variables relationship, Linear structure model. Adopted from “STATA Structural models 1 Linear regression. STATA manual,” by STATA, n.d., Retrieved from https://www.stata.com/manuals/semintro5.pdf
Figure 8: STATA software outcome. Adopted from “STATA features linear regression,” by STATA, 2018, Retrieved from https://www.stata.com/features/overview/linear-regression-and-influence/
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
After running STATA with the corresponding selected data relating to migration and international assistance for Nicaragua, results will be presented in a format similar to the one in the previous image. The outcome of the figure 8 is a summary of the described concepts on the subsections of this chapter.
‧
migration in Nicaragua'. The data collection presented restrictions because the consulted sources did not provide all the completed information necessary to study the interaction of these three variables. Nevertheless, all data is up to date and the sources presented similarities to proceed to the analysis of the figures (see Appendix 1).The information related to total migration was collected from the World Bank (2018c) website. The first quantification of migration that could be identified was for year 1962.
Starting this year, assessed periods of five years were identified. The last evaluation for net migration corresponds to the year 2017 (Appendix 2, Figure 17). During this period, 12 observations with negative values can be counted. Positive numbers in the observations would indicate that the number of migrants who entered and settled in Nicaragua is greater than the number of nationals who left the country. By the same logic, negative values for each of the observations indicate that the number of Nicaraguans who left the country is higher than the number of foreigners who settled in the national territory.
Due to net migration being the dependent variable that presents the lowest number of observations, the variable becomes the bottleneck of this quantitative analysis research with both simple and multilinear regression systems.
As for Net ODA per capita, the data was extracted from the World Bank (2018d) website.
The data correspond to the years between 1960 and 2016 (Appendix 2, Figure 18). The quantification of the official assistance received in Nicaragua corresponding to the year 2017 is not currently available. The lack of information for the year 2017 forced us to eliminate from the study the observations of all other variables for year 2017.
The Gross Domestic Product per capita data are partially available on the official website of the World Bank (2018b). We discovered that the database is limited to the period from 1989 to 2017 (Appendix 2, Figure 19). An alternative complementation was indispensable to successfully supplement the data period from 1962 to 1988. In order to achieve the objectives of this research, and since the observations in the population are a small amount, the basic complementation was a correct step in the data collection to match the 1962-2017 period, which includes the transcendental number of observations of the bottleneck variable.