Data Analysis ProceduresRaed Dakhil KareemPhD Candidate, University of Baghdad, College of Arts, Dept. of English1- Introduction: What Is Data Analysis?This is a diverse lot of questions with a common element: The answers depend, in part, on data.Human beings ask lots of questions and sometimes, particularly in the sciences, facts help. Dataanalysis is a body of methods that help to describe facts, detect patterns, develop explanations,and test hypotheses. It is used in all of the sciences. It is used in business, in administration, andin policy.The numerical results provided by a data analysis are usually simple: It finds the number thatdescribes a typical value and it finds differences among numbers. Data analysis finds averages,like the average income or the average temperature, and it finds differences like the difference inincome from group to group or the differences in average temperature from year to year.Fundamentally, the numerical answers provided by data analysis are that simple.But data analysis is not about numbers — it uses them. Data analysis is about the world, asking,always asking, “How does it work?” And that’s where data analysis gets tricky.The tools of the trade for data analysis begin with just two ideas: Writers begin their trade withtheir A, B, C’s. Musicians begin with their scales. Data analysts begin with lines and tables. Thefirst of these two ideas, the straight line, is the kind of thing you can construct on a graph usinga pencil and a ruler, the same idea I can represent algebraically by the equation “y = mx + b”.This first idea, the straight line, is the best tool that data analysts have for figuring out howthings work. The second idea is the table or, more precisely, the “additive model”. The first idea,the line, is reserved for data we can plot on a graph, while this second idea, the additive model, isused for data we organize in tables. For example, a table may represent daily mean temperaturesfor two cities and two dates: The two rows of the table show mean temperature for the twocities, the two columns show mean temperatures for the two dates.The additive model analyzes each datum, each of the quantities in the table, into fourcomponents — one component applying to the whole table, a second component specific to therow, a third component specific to the column, and a fourth component called a “residual” — aleftover that picks up everything else.There you are, lines and tables: That is data analysis, or at least a good beginning. So what is itthat fills up books and fills up the careers of data analysts and statisticians? Things begin to get“interesting”, that is to say, problematical, because even the best-behaved data show variance:
Measure a twenty gram weight on a scale, measure it 100 times, and you will get a variety ofanswers — same weight, same scale, but different answers. Find out the incomes of people whohave completed college and you will get a variety of answers.Steps / procedures to Data AnalysisThe steps that are needed in analyzing data are described below:1.Start with a Plan 2. Collect and Clean Your Data 3. Determine a Coding System 4.Tabulate Your Data 5. Transfer Your Information 6. Check Your PlanStep 1 Starting with a PlanBefore you begin your data analysis, plan how you will analyze your data. First, consider thegroups about which you want to report data. Refer to your desired result or standard to chooseyour unit of analysis to be a collection of individuals, groups, or activity locations (e.g., parkusers, school district principals, blocks within a neighborhood). This will help you select surveysto include in the analysis and to identify appropriate types of analyses. Then, choose a type ofanalysis for each question: frequency, percent distribution, mean, change in score from pre-testto a post-test, or content analysis. Next to each question on your instrument, list the type ofanalysis you are going to conduct.Step 2 Collecting and Cleaning Your DataKeep all of the forms you received for each survey, test, or interview in one place. At thispoint, you may wish to remove forms that are substantially incomplete or do not make anysense. Record the number of forms you remove and the reasons why they were removed. Youmay want to include this in your report. Assign identification numbers (ID) to each form tohelp keep track of which forms you have processed.With Qualitative Data:The purpose of doing qualitative data analysis is to reduce the amount of text and organizeresponses to identify trends in your data. One method of doing qualitative data analysis iscontent analysis. Content analysis creates a structure to organize open-ended information.
a) Identify the Unit of AnalysisBegin by identifying the unit of analysis. The unit of analysis is the smallest block of textexamined in the content analysis determined by the person conducting the analysis. Potentialunits of analysis include respondent, response, paragraph, sentence, idea, or word.b) Arrange the Raw DataAfter you determine your unit of analysis, arrange the raw data by unit of analysis. (Forexample, if your unit of analysis is “response,” then put all the responses to question onetogether, all the responses to question two together, etc., keeping each person’s responsesseparate.) If possible, type the data into a word processing program.Step 3 Determining a Coding System Quantitative Dataa) Code the QuestionIf you are going to use a spreadsheet or a database, you will need to convert “check box” typeanswers to a numbered code. This will speed up data entry, since you will only type the numbercorresponding to the answer into the computer, rather than the whole word. On a blank copyof your instrument, assign a number, or code, for each potential answer to each question.b) Organize your ResponsesBy Hand: Write the name and number of each question from your instrument on a blankpiece of paper. Create a column for each potential answer. Go through your stack ofinstruments, question by question, and tally how many people gave each type of response.By Computer: After you have assigned numbers to potential responses, input the responsesinto the computer. Set up your spreadsheet allowing each column to represent one survey itemand each row to represent one person’s form. Start each row with the survey’s ID number.You can even set up a data entry screen that looks like your printed form! (Be aware that usinga “0” for a blank answer may cause your calculations to be inaccurate. Some programs, such asExcel, will read the “0” as a number and include it when it calculates an average. However,Excel does ignore non-numeric responses.Step 4 Tabulating Your DataConduct the type of analysis you chose for each item. Four common types of analysis usedwith quantitative data are: frequencies, percent distributions, means, and change from pre-teststo post-tests. Most computer programs for statistical data analysis have functions for tabulatingautomatically. Conduct a content analysis for qualitative data with open-ended items.
With Quantitative Data:a) Calculating a FrequencyFrequencies tell how often that a particular answer was selected. Frequencies can be calculatedfor any level of data. To calculate a frequency, take one item that you have coded for dataanalysis. Tally by hand, or use a computer program to count how many times each answer wasselected.b) Calculating A Percent DistributionPercent distributions, or percentages, tell what proportion of respondents selected a particularanswer. Since percentages reflect the number of times each answer would be selected out of100 responses, they can be used to help put your data in perspective. These are particularlyuseful when you have a large number of responses.c) Calculating a MeanThe mean is the average response given. A mean can be used when you want to describe thegroup as a whole. To calculate a mean, add up all the responses you got for the question.Divide the total by the Number of responses (n).d) Analyzing Pre-Post DataThe way you write the “standard of success” in your objective will determine how you analyzeyour pre-post data.With Qualitative Data:a) Classify the DataPlace each data unit (e.g., responses to a question, paragraph, sentence, or word) in one of thecategories you identified. Each unit should be in at least one category. It may be useful to give afriend or co-worker your list of categories and your organized data and see if they classify theanswers in the same way you did. They probably will not match completely, but if your analysislooks completely different, you may want to describe your categories better, or choose newcategories.b) Reduce the Volume of Text (But Dont Lose the Information)There are several ways to reduce the amount of text within each category. You can use a count,a composite, or a short paragraph:1-Count the ResponsesCount the number of responses placed in each category, then report the number of responsesalong with the category description you have created.2-Create a Composite Response
Create “composite” responses (or use a quote) that reflect the content of all the responses ineach category. Use these composite responses along with the count of all responses in eachcategory to report your data.3-Write a Short ParagraphFor more complex or extensive data, you may want to write a short paragraph describing thecontents of each category. These can be reported with, or without, a count of the individualresponses.Step 5 Transferring Your InformationTransfer the information to a copy of the original instrument. For quantitative data, record thefrequencies or percentages for each response. For qualitative data, present the categories youdeveloped, summary of comments and the number of responses by category for each question.Step 6 Checking Your PlanGo back and check the data analysis plan you created in Step 1. Did you conduct the analysisthe way that you planned? Compare individual answers from pre and post-tests, or compareresults from your clients and with those from a control group.ReferencesIntroduction to Data analysis: The Rules of Evidence, Joel H. Levine Joel H. Levine,1996http://www.dartmouth.edu/~mss/data%20analysis/Volume%20I%20pdf%20/006%20Intro%20(What%20is%20the%20weal.pdf Accessed May, 2013.Data Analysishttp://www.nationalserviceresources.org/filemanager/download/Evaluation/users_guide/dataanal.pdf Accessed May, 2013.