Data collection is planning to hold useful information on a key characteristic (treatment outcome, the prevalence of a disorder…etc) that is important to the targeted audience. This is determined by the research question (s) or hypothesis. Data are collected to make decisions based on facts, these data are either feature data (a character is either present or absent). Alternatively, data can be variables data (measurement of a characteristic) (The Balanced Scorecard Institute Pp. 1-46). The aim of this essay is to outline the basic principles needed for a successful data collection endeavor.
Crewson and Applegate (Pp. 755-761) discussed these basic principles and inferred the first step for data collection is to define the research question. This should serve to identify the research endpoint, next is to identify data requirements that are when, how, and what data to collect. About identifying data requirements, added concepts to consider are the design of data collection form, creating data files, and deciding the unit of analysis which can be an individual patient or a procedure.
The third is to ensure data precision that is data that is data should be repeatable (the same research team member can reach the same outcome many times from the same data using the same instrument). Data should also be reproducible (all research team members can reach the same outcome many times from the same data using the same instrument). It should also be accurate (minimum difference between an observed average measurement and the accompanying previously known standard value). Data stability is a measure of the variation occurring with the same team member measuring the same item over time (Waddick p. 4). Fourth is to get the approval of the local institutional review board which is often needed especially for clinical studies (Crewson and Applegate, Pp. 755-761). Fifth, a properly designed data collection plan should include three items: 1- The statistical analysis plan is concerned with the precision and scale of the data that is whether data are nominal, ordinal, or distributed in intervals. The statistical analysis plan will decide or control the statistical techniques used in the data analysis part of the research. 2- Avoid systematic bias, which may produce errors that make the research results invalid. These errors may result from nonobjective data collection efforts (no control, or absent double-blind methodology), bias inpatient or research material recruitment, data interpretation bias. Response bias commonly occurs in the follow-up phase of the research because of incomplete patient responses. Finally, a data collection plan should include a plan for data management and handling that is how the documents like questionnaires and checklists will be numbered and kept and saved safely and who is authorized to control storage? Data management and treatment includes also data cleaning, which is the scheme put to ensure data consistency and accuracy (Crewson and Applegate, Pp. 755-761).
The results of research depend mainly on data collection as proper data makes it possible to reach conclusions based on correct facts. Systematic data collection makes it possible for a team of researchers to express and examine working assumptions which will lead to clarification of the key characteristic. It also advances decision-making by converging attention to what is happening in the research process. In simple terms, systematic data collection makes it possible to infer the data collection point to the problem is…, instead of inferring, the research team thinks the problem is… This is the core difference between objective and subjective research results.
Works Cited
Crewson, Philip, E., and Applegate, Kimberly, E. “Fundamentals of Clinical Research for Radiologists: Data Collection in Radiology Research.” AJR vol 177 2001. p. 755-761.
The Balanced Scorecard Institute. “Basic Tools for Process Improvement: Module 7 Data Collection.” The Balanced Scorecard Institute. 1996. Web.
Waddick Patrick. “Building a Data Collection Plan.” SixSigma. 2008. Web.