Public Health Discussion Boards

PUH 6301, Public Health Research 1

Course Learning Outcomes for Unit VI Upon completion of this unit, students should be able to:

4. Evaluate strategies for data analysis to determine the best statistical tests needed for research methods. 4.1 Determine the four levels of measurement as valid research statistical techniques in the public

health research process. 4.2 Explain why proper data and statistical analysis is important. 4.3 Describe the basic types of statistic tests.

Course/Unit Learning Outcomes

Learning Activity

4.1

Unit Lesson Chapter 28 Chapter 29 Chapter 30 Chapter 31 Chapter 33 Blog: “Descriptive vs. Inferential Statistics: What’s the Difference? Unit VI Essay

4.2 Unit Lesson Chapter 28 Unit VI Essay

4.3 Unit Lesson Chapter 29 Unit VI Essay

Required Unit Resources Chapter 28: Data Management Chapter 29: Descriptive Statistics Chapter 30: Comparative Statistics Chapter 31: Regression Analysis Chapter 33: Additional Analysis Tools In order to access the following resource, click the link below: The website below provides a good summary of how the public health researcher can use descriptive and inferential statistics methods to conduct public health research. Market Research Guy. (2011, December 1). Descriptive vs. inferential statistics: What’s the difference? [Blog

post]. http://www.mymarketresearchmethods.com/descriptive-inferential-statistics-difference/

UNIT VI STUDY GUIDE

Data Analysis Plan

PUH 6301, Public Health Research 2

UNIT x STUDY GUIDE

Title

Unit Lesson

Introduction This unit covers the statistical procedures used to analyze the data collected from research tools. During this stage of research, you may begin to draw conclusions and be able to answer the research question(s) and sub-question(s) you developed in Unit I. Use statistics in this stage of research to manipulate the data and make it understandable for others to read. Shi (2008) encourages researchers to know and understand basic statistics and statistical procedures. The data analysis phase of research is important because it makes sense of the data that can be used for future research studies (Jacobsen, 2021). Data Management Data management is the entire process of keeping a record of all the results of clinical assessments conducted during a research study (Jacobsen, 2021). Record keeping includes listing details on potential articles, pulling information from patient charts, tracking responses from surveys, or recording assessment results from cohorts or studies. It is vital that those responsible for collecting and keeping data maintain confidentiality and the integrity of data sets from all outside sources. Once researchers enter the data into the spreadsheet or database, the data should be recoded and double-checked prior to beginning statistical analysis. One way to keep research data organized, analyzed, and entered correctly and easily is by coding it. Creating a codebook to describe each variable and explain how the information that was collected will be entered into the computer database is valuable to a research study (Jacobsen, 2021). When you use quantitative surveys to collect data, it is easier to use numeric or alphabetic codes to organize answers from the questionnaire or survey. For example, when coding yes and no answers, you can use the number 1 to represent the answer for yes and the number two for no answers. In like manner, if you use a Likert scale, you can use numbers 1- 5 to represent worded answers. Strongly agree may be represented by number 1, agree by number 2, neutral by the number 3, and so on. When coding answers from qualitative surveys, using a codebook is even more important because the information from open-ended questions can yield unnecessary data; therefore, the researcher must provide clear and detailed instructions for how to code these responses and comments. The codebook should also specify the following things:

• the name of each variable,

• the type of variable,

• the way the question was asked,

• the options listed on the survey instrument as possible answers to the question,

• the way answers should be entered into the computer database, and

• the way to handle missing responses (Jacobsen, 2021). Researchers can also use the codebook to anticipate data problems and how those problems will be handled. It will also specifically describe how to code answers that are missing or blank. For example, if a participant leaves a question blank on the survey or questionnaire, the codebook will specify that the data should be entered as “missing” or use a number such as 0 or 1 (Jacobsen, 2021). Steps in Preparing Data for Analysis Preparing the data collected by the data collection tool can be cumbersome; however, DePoy and Gitlin (2005) provide seven steps to assist researchers organize and prepare the data for analysis:

1. Check data collection tool for accuracy. Make sure information is completed correctly. It is important to check for missing information and make sure all missing data is corrected before data entry and data analysis begins.

2. Label all variables on the tool. Variables are anything that has quality or quantity. A variable can be time, distance, food intake, or even sea level. If you are using statistical software for analysis, then it is important to make sure your variables are labeled logically.

3. Assign variable labels to locations on computer. In this step, the researcher determines the order in which the variables are entered into the computer program. Generally, researchers assign a coded identification number to every participant. These numbers are considered the raw data.

PUH 6301, Public Health Research 3

UNIT x STUDY GUIDE

Title

4. Develop a codebook and master file. This is considered the dictionary of data. This written record provides copies of the variable labels and the values that represent all of the labels.

5. Entering data into the computer-based system is the fifth step. Double verify or use another quality control procedure. Sometimes, Microsoft Excel is used for smaller data sets; however, it is preferable to use a statistical package such as the Statistical Analysis System (SAS) or the Statistical Package for the Social Sciences (SPSS Statistics) for entering large data sets. Many statistical packages allow researchers to check ranges. Double verification is an important quality control measure. Double verification involves entering the same set of data on two separate occasions. The program will check all entries against itself and alert the researchers if there is a discrepancy.

6. Clean raw data files. Cleaning data occurs when the researchers check the data that were entered and make sure they were accurately transcribed. This is another step to help determine missing information and to make sure information was coded and entered correctly.

7. Develop scores. This step involves summarizing information. The main form of summarizing information is to use descriptive statistics, which is discussed later in this lesson.

Computer software or database programs are commonly used to analyze data received in a research study. The more commonly used software in public health research are SAS, Stata, and SPSS Statistics. These computer programs are “designed to be visually appealing, to facilitate consistent entry of the acceptable responses for each question, and to perform automatic skips between questions” (Jacobsen, 2021, p. 218), which allows for uniformity and consistency of data entry. Alternatively, you can use a spreadsheet to enter data, which does not require a data entry form or code or testing of a system, but it is easy to mix up data and makes data cleaning more complex. Descriptive Statistics It is important for a public health professional to know about variables when conducting research. Variables are important in statistics and data analysis. Variables are any things or any characteristics that can have values assigned to them. Variables are generally broken down into four levels of measurement: ratio, interval, ordinal, and nominal (Jacobsen, 2021).

• Ratio: This level of measurement has numeric responses on a scale where numbers indicate value. For example, if weight is measured in pounds, a person who is 30 pounds is twice a heavy as a person who weighs 15 pounds. Therefore, the ratio is 2 to 1.

• Interval: This level also uses numeric values; however, the spacing between all variable categories is equal. For example, the Likert scale uses the interval variable scale labeled strongly disagree, disagree, neutral, agree, and strongly agree. It is implied that the distance between each variable is equal.

• Ordinal or ranked: This level of measurement assigns a rank order to each variable. For instance, responses can be ordered from least=1 to greatest=4, from best=1 to worst= 3, or from high income=4 to low income=1. No matter the scale, the rank or order of the responses is indicated by numeric value so it is easily coded.

• Nominal or categorical: This level has responses where the variables represent groups with no real rank or order (Jacobsen, 2021). For example, race, ethnicity, age, or blood types are considered nominal variables. A subtype of nominal variables is binomial variables. This type of variable has only two types of possible answers, which are yes and no and are sometimes called dichotomous variables (Jacobsen, 2021).

Ratio variables and interval variables can be further categorized as continuous variables or discrete variables. Continuous variables “can take on a value with a range” (Jacobsen, 2021, p. 236). Discrete variables come from counting something; therefore, there are gaps between values (Jacobsen, 2021). Measure of Central Tendency Descriptive statistics are probably some of the most commonly used statistics in research. Researchers use descriptive statistics when the research is describing averages, and there are three measures of central tendency or average when referring to numerical values, which are mean, median, and mode.

PUH 6301, Public Health Research 4

UNIT x STUDY GUIDE

Title

• The mean of a sample is most frequently used and is calculated by totaling the number of responses to a question, then dividing that sum by the total number of individuals who answered the question (Jacobsen, 2021). This is a good measure of summary but can be affected by extreme values or outliers.

• The median is calculated by putting all the answers given in order from least to greatest, then identifying the number that is in the middle (Jacobsen, 2021). That middle number is called the median. If the responses come to an even number, then the researcher should add the middle two numbers and divide by two, which will yield the median (Jacobsen, 2021).

• The mode is the measure of central tendency that is the number or answer given by participants that appears the most (Jacobsen, 2021).

Advanced Statistics One consideration you must make as the researcher is that the results or analysis can be inaccurate or skewed. Numeric variables have a normal distribution, but they may be skewed. Skewness may be shown with “responses that extend farther from the peak on either the left…or the right…side of the histogram” (Jacobsen, 2021). Understanding whether the error happened as a data collection problem or an analysis problem is important. Ensuring that the data analysis is not falsified, fabricated, or plagiarized is important because that data can unintentionally corrupt a research project if left unchecked. As the researcher, you should be sure that statistical honesty is adhered to by following the established rules of scientific research practices. If at any time you are not sure of your findings or do not understand the statistics, you are always encouraged to check with a professional (Jacobsen, 2021). Professional statisticians are available to make sure that the sampling methods and sample size fit your research, that the questionnaire or survey used will provide usable data, and that the analysis plan is reasonable and reliable.

References DePoy, E., & Gitlin, L. N. (2005). Introduction to research: Understanding and applying multiple strategies (3rd

ed.). Elsevier Mosby. Jacobsen, K. H. (2021). Introduction to health research methods: A practical guide (3rd ed.). Burlington, MA:

Jones & Bartlett Learning. Shi, L. (2008). Health services research method (2nd ed.). Delmar.

Suggested Unit Resources In order to access the following resources, click the links below: You are encouraged to explore the article below. It offers a good review of how public health utilizes descriptive research in epidemiology research. Naito, M. (2014). Editorial: Utilization and application of public health data in descriptive epidemiology. Journal

of Epidemiology, 24(6), 435-436. http://search.proquest.com.libraryresources.columbiasouthern.edu/docview/1622120440?accountid= 33337

You are encouraged to explore the website below. Health data tools and statistics. (n.d.). https://phpartners.org/health_stats.html

Learning Activities (Nongraded) Nongraded Learning Activities are provided to aid students in their course of study. You do not have to submit them. If you have questions, contact your instructor for further guidance and information. Click to access the Unit VI terminology flash cards PowerPoint. Click to access the PDF version.