1 CHAPTER 12 VARIABLES LIST Variables list list of variables being measure keyed to the question number in the questionnaire which are designed to measure each variable Ensures proper coverage balance and non duplication of items DATA MANAGEMENT Process by which raw data gathered by some instrument or measurements are converted into numbers for analysis purposes Steps 1 Data Gathering Instrument EX Questionnaire 2 Codebook Codes Instructions for Coding guidebook for the numerical classification of questions to be coded 3 Code sheet Transfer Sheet a blank score sheet on which questions can be coded 4 Data Entry Keyboard Entry Keypunch 5 Data Verification Cleaning edited or verified 6 Data Files Data Disks Tape Computer Memory or Data Decks Editing Prior to the actual coding the questionnaires should be edited Members of the research should check questionnaires to assure that the field interviewer or respondent completed each item accurately Although questions were probably precoded codes may have to be developed for the open ended items on the basis of initial returns A useful procedure in developing codes for open ended items is to read through a sufficient of responses create a preliminary code on the basis of these responses Rule of thumb A code may require revision when 10 or more of the responses are classified as other or as not fitting in the categories the code needs to be revised in order to conclude many of the responses The bulk of survey research work is conducted by those who have the least training often the least commitment to the goal of the research hired hand research Sussman and Haug say that mechanical errors account for a significant level of errors in survey research Coding assignment of to responses gathered by a research instrument Codes go into a codebook The purpose of the conversion of the questionnaire into numerical data is so that the researcher can store the questionnaires work from the summarized information the numbers Coder monitoring Involves checking the work of coders for accuracy coder monitoring is necessary to ensure quality control in research One procedure involves coder verification and reconciliation procedures as essential quality control checks on mechanical error Each questionnaire is double coded meaning that two different coders independently code the same questionnaire 2 In some instances precision and care in early stages of a project are disregarded once the boring routine of number crunching is begun Keyboard Entry The most widely used means to enter input The terminal displays case 1 variable name and prompts asks for info for case then data is entered and the screen prompts for the info for the next case etc Data verification cleaning involves double checking the data file in search errors many of which inevitable despite the conscientiousness of workers They can be checked by computers programs themselves SIMPLE DATA PRESENTATION Summarizing and using univariate statistics Running the marginal single variable tabulations of the type of data which appears in the margins of tables Through the use of rates statistics tables graphic displays frequency distributions and other summarizing procedures it should be possible to communicate the significant findings in an understandable painless fashion Rates proportions or percentages and ratios are meaningful ways of standardizing data so that useful comparisons can be made between unequal populations rate expresses the of cases of the criterion variable per unit of population Example Crime rate of crimes divided by population per 100 00 population Proportions expresses the number of cases of the criterion variable as part of the total population They are less useful where the criterion is relatively a rare event Rate is more useful in summary statistic than the proportion Proportion frequency of criterion variable N Percentages calculated by dividing a frequency by the toal N and multiplying the result by 100 Ratios frequency 1 frequency 2 it simply compares the of cases in one category with the number of cases in another GRAPHIC PRESENTATIONS Graphs or pictorial presentations of data are an attractive means of capturing the reader s attention as well as of summarizing data particularly from frequency distributions EX SPSS Excel Harvard Graphics Pie Charts circles pieces represent some proportion of some phenomenon and total 100 percent Bar Graphs consist of rectangles width represents the class intervals and the height represents quantity or amount score values are usually arranged along the horizontal dimension across percentages are plotted vertically upward or downward Frequency Polygons Line charts the frequency or percentages of the midpoint of each score are plotted and connected by a straight line that begins ends at the baseline Using SPSS for graphs it is considerably in the process of selecting and constructing highly useful graphic displays The graphs gallery gives a main chart with icons The latter makes it possible to envision the type of graphic display one may desire Crime Clock are highly inaccurate means of depicting crime change that fails to control for 3 it fails to control for population growth and uses a constant fixed unit of comparison population growth times it is designed to convey the annual reported crime experience by showing the relative frequency of occurrence of the Index Offenses it is the most aggregate representation of UCR data ELABORATION Process of introducing or controlling for third variables control or test factors by sub classifying original tables STEPS 1 demonstrate a relationship between the variables 2 3 specify the time order which the independent and dependent variable control for or exclude rival causal factors replication repetition of experiments utilizing the same methodology explanation the relationship observed in the original bivariate table weakens or disappears in the Interpretation the relationship observed in the original bivariate table weakens or disappears and the control variable is an intervening one occurs between x and y Suppression when despite no relationship in the original bivariate table a relationship occurs in partial tables the partial tables P133 139 WAYS OF ELIMINATING DISADVANTAGES IN MAIL SURVEYS 2 groups a Those who have yet to respond b Those who refuse to cooperate As long as this rate is small less than 1 percent for example it is an expected loss in surveys Follow up continued efforts to solicit
View Full Document