Understanding Samples and Populations
Mashburn and Myers (2010) developed an article to report their findings in an early childhood longitudinal study examining speech-language impairment. The purpose of the study was to add to the current research on the topic, analyze data regarding “speech-language-hearing research,” and to provide the field with additional evidence (Mashburn & Myers, 2010, p. 61). One of the most important results from the study is the recognition that there are many different subsamples of learning disabilities, and these disabilities can influence how a child learns and applies speech-language during their first years of schooling. In order to understand how the study is applicable to different situations or different groups of children, it is important to understand the population that was studied, the sampling procedure, the size of the sample – and versus the population, the response rate, and if the sampling procedures were effective and appropriate for the study.
Many studies are meant to represent a very large population that cannot be studied in whole. For example, the Mashburn and Myers (2010) study is meant to be a study of the population of kindergarten students in the US, with speech-language impairment, whose information was available through the ECLS database. Specifically, the sample of the population studied in Mashburn and Myers (2010) was derived from a database that is designed to provide researchers with a sample group that is believed to be realistically impartial, as a fair sample for testing, referred to as ECLS-K. One of the difficulties of using this database is that there is no way to ensure that the results are accurate for that specific sample, or student, where their specific results may be the response of bias from the person that diagnosed the child’s condition, or misunderstand the requirements of the information required for the database.
However, this database is nationally accepted as a reasonable resource for researchers, and Mashburn and Myers (2010) used a sample size of “21, 260 children enrolled in kindergarten during 1998-1999 who had information resulting from either a child assessment or parent interview in the fall and/or spring of kindergarten” (p. 65). The size of the sample was 21, 260 children, from a population size in the millions, with a response rate that was qualified prior to use of the information in the study, in that the information was available in the database. Study usage of the information as dependent on the factors the researchers needed to include the students in the study; however, a subsample was developed to draw further conclusions.
The sampling procedure used in this study is acceptable and reasonable because it uses nationally accepted and validated information, works from a population that was a fair representation of the current diversity of the US, though not an exact match. One of the largest difficulties in studies is creating a sample that is representative of the population, while still obtaining the response rate needed to develop conclusions to the study. Many respondents may be accidentally removed from the process through an elimination process meant to be random, lost in the process of developing the study, or simply unavailable during the time of the study. This can cause a sample size to be unbalanced in demographic factors that can be essential to the true application of any results achieved through the study.
These sampling procedures were justified for use in the study, because of the complications of developing a successful study group to represent such a large population. Additionally, the study involved questionnaires for special education teachers, and used descriptive statistics to work with the documented information. One of the complications of the study results included situations where students were retained in the same grade (the study-included kindergarten through 5th grade for review). Developing a solid sample population for students with disabilities must include students held back as part of the calculations. Additionally, the study examined student schedules and delved more deeply into the sample than only the information from the database. Using the database to indentify the sample size from a large population group permitted Mashburn and Myers (2010) to successfully identify a group that could be studied and conclusions were developed.
Populations are identified as the target group that is being examined – whether as part of the objectives to increase the evidence-based research for education or to market a new product. However, populations are often too large to study in completeness, requiring researchers to select a sample of the population that most represents this group. Some samples are hard to identify, specifically in populations that have been limited information. In many cases, students and children are populations where limited evidence or information is available to assist researchers in understanding how their sample sizes will be able to represent the population fully. Just as Mashburn and Myers (2010) did with their study, most researchers will need to understand as much as possible of the big picture of their population before developing their sample.
Mashburn, A., & Myers, S. (2010). Advancing Research on Children With Speech-Language Impairment: An Introduction to the Early Childhood Longitudinal Study — Kindergarten Cohort. Language, Speech, & Hearing Services in Schools, 41(1), 61-69. doi:10.1044/0161-1461(2009/08-0037).