Mortality Attributable to Low Levels of Education in the United States Paper Example

Section 1: Types of Sampling Procedures

Sampling is one of the critical aspects of empirical research. It is essential since in most cases, it is too expensive or impossible to study the entire population. Sampling procedures can be broadly categorized into probability and non-probability methods.

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Save 25%

Probability Sampling Methods

Probability sampling starts with a sampling frame consisting of all individuals from which the sample is selected (Fowler, 2014). Thus, each has an equal probability of being included in the sample. Probability sampling methods include simple random, stratified, systematic and clustered sampling.

Simple Random Sampling

Each element is selected purely by chance, and all members of the population have the same chance of being selected. It can be implemented by assigning random numbers to each member of the population and using a table of the random numbers to select the sample (Levy & Lemeshow, 2013). It is suitable when the population of interest is large and is impossible to identify each element (Fowler, 2014). This method is beneficial since it reduces sampling bias as all individuals have an equal probability. It is also simple and straightforward. However, it is not suitable when the characteristic of interest is uncommon in the population.

Systematic Sampling

It selects individuals at regular intervals from the sampling frame. The elements in the population are arranged in an order and selected at regular intervals based on the sample size (Warne, 2017). The appropriate interval is determined by dividing the total population by the desired sample size. It is suitable when the population is logically homogenous. It is easier to administer than simple random sampling. However, it can lead to sampling bias is there is periodicity in the data.

Stratified Sampling

Stratified sampling sub-divides a population into subgroups with each subgroup containing a common characteristic (Warne, 2017). Samples are then selected randomly from each subgroup. For instance, in a study of performance among college students, the population can be subdivided into male and females and samples selected from each category. Stratified sampling is suitable when it is possible to subdivide the population into characteristics of interest (Levy & Lemeshow, 2013). It is beneficial since it ensures that every group is included in the sample. Its limitation is that it is complex and difficult to administer.

Clustered Sampling

Clustered sampling involves subdividing the population into groups (clusters) and using the subgroups as the sampling unit (Thompson, 2013). The clusters are randomly selected. Clustered sampling can be single-stage or two-stage. In a single-stage sampling, all elements in each cluster selected are included in the sample. For instance, in a study of tuition fees in public universities in the US, the 50 states can be used as clusters. A sample of 30 states can be chosen, and all the public universities in the states are included in the survey. In a two-stage cluster sampling, public universities are randomly selected from each cluster. A two-stage cluster is suitable when it is impossible or too expensive to include all individuals in the clusters selected.

Clustered sampling is more convenient and efficient than simple random sampling especially when the population covers a diverse geographical region (Thompson, 2013). However, it increases the risk of bias since the clusters may not be a good representative of the population.

Non-Probability Sampling Procedures

Non-probability sampling does not start with a sampling frame hence elements do not have an equal chance of selection (Thompson, 2013). It increases the risk of selecting a non-representative sample. Non-probability sampling methods include convenience, snowball, quota, and judgment sampling.

Convenience Sampling

It picks samples based on availability and willingness to participate in the study. It is suitable during preliminary research (Warne, 2017). It is cheaper and less time-consuming than random sampling. Its limitation is that it does not give a representative sample.

Snowball Sampling

Existing subjects are asked to refer the researcher to the next respondents. Thus, as the sampling continues, the sample size increases. It is suitable when sampling hard-to-reach groups where it is difficult to identify the sampling frame.

Judgment Sampling

In this approach, the researcher relies on judgment to implicitly choose a sample that suits his or her needs (Thompson, 2013). It is cost-effective and straightforward to perform. However, a judgment sample is not representative of the population due to volunteer bias and possible errors in judgment.

Section 2: Article Review

Mortality Attributable to Low Levels of Education in the United States (Krueger, Tran, Hummer & Chang, 2015)

The components of the article include abstract, introduction, methods, results and discussion. The abstract summarizes the main section of the article.

Introduction

The introduction highlights the background of the study. Krueger, Tran, Hummer & Chang (2015) argue that although low education is a strong predictor of adult mortality, little research has been done to determine the number of adults attributable to low education. The article analyses the motility of rates among adults aged between 25 and 85 years with different levels of education.

Methods

The study used the 1986-2004 data from the National Health Interview Survey linked to the Linked Mortality File, which includes data on deaths. Thus, the study relied on secondary data. Krueger, Tran, Hummer & Chang (2015) used judgment sampling to exclude individuals below the age of 25. This was based on the premise that many people under the age of 25 were still enrolled in school hence they were not appropriate for this study. The sample included includes 468,725 males and 540,224 females (Krueger, Tran, Hummer & Chang, 2015). The variables include the cause of mortality, age, educational attainment, race, birth cohort, and gender. Race, gender, and cause of death were measured under the nominal scale. Cause of mortality was categorized into three; all-cause mortality, cancer mortality, and cardiovascular disease mortality. Educational attainment was measured under the ordinal scale and coded as less than high school degree, high school degree, college but no bachelor's degree, bachelors' degree, and post-graduate education.

It used e a complementary log-log discrete time survival model to calculate mortality rates among the different groups. It then compared the estimated mortality rates of adults in a given educational level with the mortality rates of a higher education level to determine the differences. Comparative analysis wee also conducted for other groups based on sex, age, the cause of mortality, and birth cohort.

Results and Conclusion

The results of the analysis showed that there is an inverse association between educational attainment and mortality rates among adults in the USA (Krueger, Tran, Hummer & Chang, 2015). Mortality rates were higher in adults with less than high school degree and lowest in adults with a bachelor degree or higher. It further revealed that the mortality caused by having less education was similar across all gender and racial groups. Krueger, Tran, Hummer, and Chang (2015) concluded that educational attainment has a significant causal effect on adult mortality.

Section 3: Scales of Measurement

Scales of measurement are ways in which variables are defined and categorized (Illowsky & Dean, 2017). Properties of measurement include identity, magnitude, intervals, and a minimum value of zero. They include the nominal, ordinal, interval and ratio scales (Illowsky & Dean, 2017).

Nominal Scale of Measurement

In this level, numbers and data are used only as identifiers and do not have a numerical value (Zeisset, 2009). Thus, it satisfies only the identity property of measurement (Zeisset, 2009). For instance, male and female are categorical data. Male can be coded as 1 and female as 2. 1 and 2 do not have any numerical value hence they cannot be ranked. Variables measured under this scale include gender, race, religious and political affiliation, country of origin, among other variables.

Ordinal Level of Measurement

This scale meets both the identity and magnitude properties of measurement (Freund, Wilson & Mohr, 2010). Responses can be ranked since the number have a numerical value (Freund, Wilson & Mohr, 2010). For instance, when polling the views of students on the performance of the leadership of their educational institution, responses can be classified as strongly satisfied, satisfied, dissatisfied, and strongly dissatisfied. These responses can be ranked based on the level of satisfaction.

Interval Scale of Measurement

It possesses the measurement properties of identity, magnitude, and equal intervals. Under this scale, zero does not imply the absence of the attribute (Burr, 2014). Thus, zero is not the absolute lowest value (Dowdy, Wearden & Chilko, 2011). For instance, temperature is measured on the interval scale. The difference between 70F and 80F is the same as the difference between 210F and 220F. Temperature can be negative hence 00F is not the absolute minimum value.

Ratio Scale of Measurement

It meets all the four properties of measurement. This implies that it is similar to the interval scale but has an absolute zero (Ha & Ha, 2012). The values of a variable measured under the ratio scale cannot fall below zero (Reid, 2013). For instance, the height of students, tuition fees, number of students, and length of a degree program, among other variables are measured using the ratio scale.

References

Burr, I. (2014). Applied statistical methods. Elsevier.

Dowdy, S., Wearden, S., & Chilko, D. (2011). Statistics for Research. Hoboken: John Wiley & Sons.

Fowler, F. (2014). Survey research methods. London: Sage Publication.

Freund, R., Wilson, W., & Mohr, D. (2010). Statistical methods. Amsterdam: Elsevier.

Ha, R., & Ha, J. (2012). Integrative statistics for the social & behavioral sciences. Thousand Oaks, Calif: SAGE.

Illowsky, B., & Dean, S. (2017). Introductory statistics. 12th Media Services.

Krueger, P., Tran, M., Hummer, R., & Chang, V. (2015). Mortality Attributable to Low Levels of Education in the United States. PLOS ONE, 10(7), e0131809. doi: 10.1371/journal.pone.0131809

Levy, P., & Lemeshow, S. (2013). Sampling of populations. Hoboken, N.J.: Wiley.

Reid, H. (2013). Amazon.com Find in a library All sellers >> Books on Google Play Introduction to Statistics: Fundamental Concepts and Procedures of Data Analysis. SAGE Publications.

Thompson, S. (2013). Sampling. Hoboken, N.J.: Wiley.

Warne, R. (2017). Statistics for the social sciences. Cambridge University Press.

Zeisset, R. (2009). Statistics & measurement. Gainesville, FL: Center for Applications of Psychological Type, Inc.