Gender and Presence of Children: Impact on Perceptions of Food Safety, Assignments of Community Health

Download Gender and Presence of Children: Impact on Perceptions of Food Safety and more Assignments Community Health in PDF only on Docsity! February, 2001 Sociology 475: Sociology of Health Care Systems HANDBOOK State of the State Surveys Harry Perlstadt, PhD, MPH INTRODUCTION TO THE ASSIGNMENT The purpose of this assignment is to increase your knowledge of medical sociology and sociological research. Upon completion, you will have tested hypotheses or replicated findings using the data collected on a questionnaire which gathered information about the backgrounds, activities and attitudes of respondents using data collected from a survey questionnaire. This handbook will help you select hypotheses, obtain computer generated statistics, arrange those statistics into meaningful contingency tables, analyze the tables and organize your term paper. The computer is a convenient tool for compiling data and calculating statistics. A computer lab session is scheduled for the course at which time you can receive assistance with this assignment. You may also consult with classmates, roommates or other students, but be aware that this assignment has certain procedures and conventions that may not be used by others. You will be using the windows version of the Statistical Package for the Social Sciences. This is available on almost all computers in the University computer laboratories. You cannot access SPSS via modem from your room or dorm. You can purchase the SPSS Advanced Package for Graduate Students from the Computer Store 3rd Floor of Computer Center on campus for about $150 disks and $170 CD. Computer programs are constantly updated and each terminal or computer lab may have a slightly different configurations, set ups, and operating systems. This handbook cannot take into account all changes or the differences between various versions of the same program. If you run into a problem or consistent error, do not waste a lot of time and energy or get frustrated. If you cannot figure out how to correct your error or get the result you want, ASK for help. All other work--the selection of hypotheses, the review of frequency distributions, the construction of contingency tables, the analysis of the tables, and the final writing of the term paper--must be your own individual work. The instructor will be available during regular office and computer laboratory hours to discuss any problems you may have concerning this aspect of the assignment. -1- NOTE: While the data base contains a variety of items, you are to do your paper on a health related topic. -2- 5. To access the CATI script Click on CATI Script.soss18 and open 6. To locate a specific question in the script click on edit | find type in the question ID you want such as L8c Hit find and it should go to that question. -5- PART II: SELECTING HYPOTHESES A hypothesis is a sentence which relates several variables to each other, often in a causal manner. Hypotheses are really questions addressed to the data, and as such, a hypothesis may be presented as a question as well as a formal statement. The most important part of asking questions or formulating hypotheses in a study is to have them inter-related through a rationale. Rationales often include the following: (a) a specific problem area discussed in lectures or readings (b) a general belief gathered from popular culture (movies, books, TV) (c) an exploration of personal observations and/or experiences This is a class on sociology of health care delivery. Do not pick variables that are unrelated to health such as right to strike or job performance of President or Governor or right to strike) The example used throughout this manual is taken from items on the SOSS10 rather than SOSS 18. In the past few years, several cases of food poisonings and food processing recalls have made the headlines. A few children have died. This an be explored using the SOSS data. SOSS10 asked "How important is the need to provide food safety information in your community" [X5] and "how well is providing food safety information being addressed in your community" [X5a]. These are the issues we wish to explain and are termed the dependent variables. The answers depend on the characteristics of the respondents or their circumstances. For example, people with children under 18 might have different views on food safety than people without children, and females may have different views on food safety than males. Or mothers (females with children under 18) may have different views than others. The easiest way to begin is to write out questions or statements that relate ideas together. For example: Are people with children under 18 concerned about food safety? Are females concerned about food safety? These can then be expanded to fit the questions or measures asked in the survey: Are people with children under 18 concerned about getting information about food safety? Do people with children under 18 believe getting information about food safety is being addressed? Are females concerned about getting information about food safety? Do females believe getting information about food safety is being addressed in their community? Note that the two questions about food safety have been separated and each now has its own question or hypothesis with regard to people with children or females. It is desirable to see how the independent variables interact together rather than separately. This may enable us to determine the influence of each independent variable has on the dependent variable in the presence of (or controlling for) the other. This is accomplished by including the second independent variable in the hypothesis and looking at combinations of independent variables with the dependent variable. One combination of independent variables involved females with children, that is mothers. This combines two variables, gender- females [CD1] and households with children under 18 [CD12]. The questions would be: Are mothers (females with children) concerned about getting information about food safety? Do mothers (females with children) believe getting information about food safety is being addressed? PART III: OBTAINING FREQUENCIES Your next step is to obtain frequencies for the variables in which you are interested. The frequencies procedure in the SPSS produces frequency tables, measures of central tendency and dispersion, histograms, and bar charts. You can sort frequency tables by value or by count, and you can display frequency tables in condensed format. You will use this information to decide which variables to include and how to combine categories of each variable. Below are the steps necessary to obtain the frequencies. How to Obtain Frequency Tables From Computer Labs (216 Berkey) 1. Log in: type your ID and Password. The Windows 95 setting will appear. 2. Click on Start | Programs | Math Appls | SPSS | SPSS 10.0 3. In the SPSS Window, Click: File | Open OR dot in Open Existing File and Click OK Select File type: SPSS Select Look in: Root on >Mlabsafs=(U:) At the File name, Type: MSU\course\soc\475 or go through the file folder icons: MSU | course | soc | 475 4. Click on SOSS18WT.SAV and then open From the top menu, 2. Click on: Analyze/ Descriptives /Frequencies This opens the Frequencies dialog box, In the Frequencies box, from the left panel box, 3. Select variables that you are interested in They are listed by name but in order of appearance in survey. 4. Move them to the right panel box by click on the arrow( >>=). 5. Click on >statistics= Mark >mean= & >median=, and click on >continue=. This takes you back to the Frequencies dialog box. In the Frequencies box, 7. Click on >OK=. The SPSS will produce the frequencies tables for the variables you have selected in SPSS Output Navigator. PART IV: REVIEWING FREQUENCY DISTRIBUTIONS Obtaining and reviewing frequency distributions for the variables in a study is a necessary first step because it helps identify potential some problems: (a) The variable of interest may not vary. It is pointless to analyze a variable in which almost every case falls in a single category. For example, 91.9% of the respondents say that child safety is very important [X12]. This is not worth further analysis since almost everyone says it is very important and the survey is not large enough to support an analysis of those who say less than very important. (b) Some categories for a variable may result in calculations and inferences based upon a very small number of cases. Since only 3.4% of respondents say they have 4 or more children under 18 currently in their household, then it is unwise to create a category of large families (four or more children). But the survey does contain the total number of children born and adopted (CD14) which indicates that 32.1% of respondents had three or more children. (c) Some groups cannot be combined meaningfully. For example, single, never married cannot be combined with currently married or currently remarried. Single, never married could be combined with widowed, currently divorced, or currently separated if we wish to examine people living together and living alone. But that might be better examined under number of people in the household. Variables are considered to be either independent or dependent. Independent variables are causes, predisposing factors, treatments or exogenous. Age, race, and gender are examples of independent variables because they influence the attitudes and behavior of the respondents. Dependent variables are effects, outcomes or results. Some degree of causality is usually implied in sociological analyses, and a dependent variable generally follows or is the consequence of independent variables. Sometimes, given your problem and theory, a variable (such as importance of addressing food safety) may be considered to be either independent or dependent. We have already taken it as a dependent variable as a concern of mothers. But it could be an independent variable if concern for food safety leads one to prefer one political party over another. The important point is for you to clearly think through and state your hypotheses so you know whether a variable is independent or dependent. The type of analysis you will be doing involves examining one dichotomized dependent variable at a time. A dichotomized variable is one divided into two nearly equal parts. This is done using the RECODE feature of SPSS. The RULE is to dichotomize a variable as close to the median as possible. The median, which is similar to the term 50th percentile, is the value which divides a variable into two halves. The closer to a 50-50 split the better. It is not advisable to use splits lower than 20- 80. Variables can be continuous or discrete. A continuous interval variable is expressed as a set of increasing or decreasing numbers on a scale such as age or year of birth (CD2). A continuous ordinal variable is expressed as a set of rank ordered groups on a scale such as very important, somewhat important, and not very important or excellent, very good, good, fair and poor. A discrete or categorical variable is expressed on a nominal scale that cannot be rank ordered such as gender (male and female). The SPSS output on the next page contains frequencies for the three variables that are in Hypothesis 1. These tables will be used to help determine how and where to divide variable values into two or three groups for purposes of analyses. [NOTE: These instructions are for SPSS 10.0 and may differ for other versions and operating systems] PART V: OBTAINING CROSSTABS Once you have decided which variables you want to use in your analysis, you will do a run on the computer which will produce cross tabulations. This run has two parts. You should first combine, that is, recode categories-values of the variables into a usable set. The reasons for combining categories have been explained in the section on Reviewing Frequencies. You will then need to list the actual tables you want produced. Below are the steps for carrying out this part of the assignment. MAKE A LIST OF THE TABLES AND RECODES YOU WANT BEFORE STARTING COMPUTER WORK 1. RECODING. The effect of the RECODE instruction is to regroup the variables. We need to dichotomize the dependent variable importance of food safety [X5]. As discussed in the previous section, 64.6% of the respondents said it was very important. This means we can easily dichotomize between those who say it is very important and those who say it is somewhat or not very important. This would be recoding 1 B> 1 and 2 thru 3 B> 2. The independent variable children [CD 12] must also be recoded. Given our hypotheses, a comparison between households with and without children will suffice. Therefore CD12 can be recoded 0 B> 0 for households without children and 1 through 6 new value 1, that is, 1 - 6 B> 1 for households with children. We do not have to recode variable gender [CD1] because it has only two categories of Male (1) and Female (2). NOTE: It is recommended that you recode to the highest value of the low group and the lowest value of the high group. That is, 1 thru 3 B> 3 and 4 thru 5 B> 4. This will make the labels that are printed correspond to your cut points. From Computer Labs (216 Berkey) 1. Log in: type your ID and Password. The Windows 95 setting will appear. 2. Click on Start | Programs | Math Appls | SPSS | SPSS 10.0 3. In the SPSS Window, Click: File | Open OR dot in Open Existing File and Click OK Select File type: SPSS Select Look in: Root on >Mlabsafs=(U:) At the File name, Type: MSU\course\soc\475 or go through the file folder icons: MSU | course | soc | 475 4. Click on SOSS18WT.SAV and then open A. Generally it is "best practice" to recode into a different variable and then rename the recoded variable. To recode the values of an existing variable into a Different Variable. A-1. Choose: Transform / Recode / Into Different Variables This opens the Recode into Different Variables dialog box which contains the source variable list.  Make sure it is blank . Best to hit Reset button. If any recodes are in the Old ( New box, REMOVE them . A-2. Select a variable for recoding by highlighting it and then clicking on the arrow (>>=). This moves the variable into the Numeric Variable B> Output Variable box. A-3. In the Output Variable Name box write in the name you want to give the recoded variable. For example, if you were recoding CD12 number of children under 18 living in household, you could rename it CD12R meaning CD12 recoded, or you could call it NEWKIDS meaning the new (recoded) variable for children under 18 at home. A-4. Once you type in the name, the CHANGE button will appear and you hit it to shift the new name into the Numeric Variable B> Output Variable box. To define the values to recode, A-5. Click on Old and New Values. This opens the Old and New Values dialog box. A-6. Specify the old value and the new value, and then, click on Add. For our example of the dependent variable [X5], in the left Old Value panel box, choose value and enter 1. In the New Value box, enter 1, and click on add. The old and new value will appear in the Old B> New box. 1 B> 1. Then return to the Old Value panel box and chose range. Enter 2 in the first range box and 3 in the second range box so it reads 2 through 3. Then move to the New Value box and enter 2. Finally click on add. This will place the recode in the Old B> New Box and it will read 2 through 3 B> 2. Continue (which takes you back to the Recode Variables dialog box) and then click on OK. 2. CROSSTABS: We will be creating three way tables, that is, crosstabs with two independent and one dependent variable. In addition to cell counts, you should obtain row percents. To obtain Crosstabulations of our first hypothesis that mothers (females [CD1] with children [CD12]) are very concerned about food safety [X5], from the menus at the top of the SPSS screen, 2-1. Choose: Analyze/ Descriptives / Crosstabs This opens the Crosstabs dialog box. 2-2. Select an independent variable from the left source list box and move it to the Row(s) by clicking on the arrow (>>=). In our example click on (CD12) and then the > arrow. 2-3. Select a dependent variable from the left source list box and move it to the Column(s) by clicking on the arrow (>>=). In our example click on ((X5) and then the > arrow. 2-4. Select a second independent variable (CD1) from the left source list box and move it to the Layers of control variables by clicking on the arrow (>>=). In our example click on (CD1) and then the > arrow. To get row and total percentage, and the number of case in each cell, 2-5. In the Crosstabs dialog box, click on Cells This opens the Crosstabs Cell Display dialog box. 2-6. Choose: Observed and Row. Then click on Continue. This takes you back to the Crosstabs dialog box. 2-7. Click on OK to obtain the output of the crosstabs. Since you will be exploring three to five hypotheses and each hypothesis requires generating three way tables, you will be following the same procedures as above to obtain the crosstabs for our earlier hypotheses as well. The SPSS will produces the output similar to the following: VI: CONSTRUCTING PERCENTAGE TABLES Each SPSS crosstab table should consist of two independent variables and one dependent variable. The dependent variable should be the columns. You construct these final tables by coping the percents and numbers by hand onto your paper or you can type them into a tables format in Word or Wordperfect. Looking at the computer output on the previous page, you will see that each box contains a number and percent. For example, the upper left hand cell indicates that 148 male respondents with no children reported that food safety was very important. The next cell on the same row indicates that 102 male respondents with no children reported that food safety was somewhat or not very important (remember we recoded and combined somewhat and not very important). The far right cell on the top row indicates that a total of 250 male respondents had no children. The percents in the first row add across so that 59.2% of males with no children said food safety was very important and 40.8% of males with no children said food safety was somewhat or not very important. You will create each three-way percentage table for your paper as follows: The name of the dependent variable is listed on top as % saying, in this case, food safety very important. One of the independent variables is entered on the side and the other along the top. The table is filled in by taking the percent of males with no children who said food safety was very important (59.2%) and the total number of male respondents with no children (250). These numbers are placed in the upper right hand cell. % Saying Food Safety Very Important [X5] Presence of Children [CD12] None Presence of Children [CD12] 1 or more Gender [CD1] Male 59.2% (250) Gender [CD1] Female Similarly, the percent of male respondents with children who said food safety was important was 67.7%, and 217 respondents were males with children. These numbers would go in the upper right hand cell for males with children. Do the same for the female respondents. The completed table looks like this: % Saying Food Safety Very Important [X5] Presence of Children [CD12] None Presence of Children [CD12] 1 or more Gender [CD1] Male 59.2% (250) 67.7% (217) Gender [CD1] 67.3% 65.4% % Saying Food Safety Very Important [X5] Female (257) (234) Table n = 958 Missing = 17 Sample N = 975 PART VIII: INTERPRETATION AND DISCUSSION OF FINDINGS Table 1 shows the relationships of gender and having children under 18 at home on attitude towards importance of food safety. One way of analyzing data is to perform a quasi-experiment by statistically controlling for the effects of first one independent variable and then the other on the dependent variable. This takes advantage of natural or convenience groupings rather than the more rigorous random assignment. But it would be difficult and perhaps unethical to randomly assign people to have children or not. For example we can control for CD1, that is, whether respondent is male or female. We will then discover how having children influences being attitude on importance of food safety. Significance: The rule of thumb in this study is that a difference of 10% is significant, and a difference between 5% to 10% is marginally significant. A difference of less than 5% is not a significant difference. A. CONTROLLING FOR CD1-----CONTROLLING FOR THE SIDE VARIABLE By LOOKING ALONG ROWS we control for the variable with categories on the side (or stub) of the three way table, in this case, gender. By looking along the top row of Table 1 we find that among males who had no children 59.2% said food safety was very important compared with 67.7% of males who had children and said food safety was very important. The difference (67.7% - 59.2% = 8.5%) is marginally significant. We can therefore conclude that having children makes some different for male attitudes on food safety. Continuing our analysis by looking across the second row of Table 1, we find that among females without children, 67.3% said food safety was very important compared with 65.4% of females with children who said food safety was very important. The difference (67.3% - 65.4% = 1.9%) is not significant. We can therefore conclude that having children makes no difference on female attitudes toward food safety. The first set of conclusions can then be summarized as follows: Controlling for gender, we find that children has an affect on male attitudes toward food safety, but that children have no affect on female attitudes toward food safety. B. CONTROLLING FOR CD12-----CONTROLLING FOR THE TOP VARIABLE By LOOKING DOWN COLUMNS, we can control for the variable with categories on the top of the three way table, in this example, CD12 presence of children. By looking down the first column of Table 1, we find that for respondents without children, 59.2% of males said food safety was important compared with 67.3% of females who said food safety was important. The difference between them (67.3% - 59.2% = 8.1%) is marginally significant. We can therefore conclude that being female makes a difference in attitudes toward food safety among those with no children. In the second column, we find that for those with children, 67.7% of males said food safety was important compared with 65.4% of females who said food safety was important. The difference between them (67.7% - 65.4% = 2.3%) is not significant. Therefore gender makes no difference in attitude on food safety for people with children. The second set of conclusions can then be summarized as follows: Controlling for presence of children, we find that gender had an affect on attitude on food safety for those without children, but had no affect for those with children. The first hypothesis was that mothers (females with children) will indicate that information about food safety is more important that fathers or people without children. The data analysis of Table 1 does not strictly uphold this hypothesis. The data suggest that males without children are less likely to say that food safety is important than males with children or females with or without children.