• No results found

RECODING THE VALUES OF VARIABLES

In document Quantitative Analysis With SPSS (Page 62-67)

Further steps with SPSS 8 for Windows

RECODING THE VALUES OF VARIABLES

Sometimes it is necessary to change or to recode the values of some variables. For example, it is recommended that the wording of questions which go to make up a scale or index should be varied in such a way that people who say yes to everything (yeasayers) or no (naysayers) do not end up with an extreme score. To illustrate this, we have worded two of the four questions assessing job satisfaction in the Job Survey on pp. 40–41 (‘6c Most of the time I have to force myself to go to work’ and ‘6e My job is pretty uninteresting’) in the opposite direction from the other two (‘6b My job is like a hobby to me’ and ‘6d Most days I am enthusiastic about my work’). These questions are answered in terms of a 5-point scale ranging from 1 (‘strongly disagree’) to 5 (‘strongly agree’). While we could reverse the numbers for the two negatively worded items (6c and 6e) on the questionnaire, this would draw the attention of our respondents to what we were trying to accomplish. It is simpler to reverse the coding when we come to analyze the data. Since we want to indicate greater job satisfaction

46 Further steps with SPSS 8 for Windows

with a higher number, we will recode the answers to the two negatively worded questions, so that 1 becomes 5, 2 becomes 4, 4 becomes 2, and 5 becomes 1.

We can do this with the Recode procedure which is on the Transform menu. We can recode values using either the same (Into Same Variables…) or different (Into Different Variables…) variable names.

If we want to compare the original values with the recoded ones or if we want to retain the original values, we use the Into Different Variables… option which opens the Recode into Different Variables dialog box presented in Box 3.3. For instance, if we wish to recode satis2 and satis4 into the respective two new variables of rsatis2 and rsatis4, we select satis2 which puts it in the box entitled Numeric Variable -> Output Variable:, type the new variable name of rsatis2 in the box entitled Name: select Change and repeat the procedure for satis4, renaming it rsatis4.

Then we select Old and New Values…which opens the Recode into Different Variables: Old and New Values subdialog box displayed in Box 3.4. In the box entitled Value: (in the Old Value section where the V of Value: is underlined to distinguish it from Value: in the New Value section where the 1 of Value: is underscored) we type the first value to be changed (e.g. 1) while in the box called Value: (in the New Value section) we type the new value (e.g. 5) and select Add. We do this consecutively for the three other old values of 2, 4, and 5. For the values which remain the same (e.g. 3) we can type the old value in the box called Value:, and select Copy old value[s] and Add. The values for each case are recoded from left to right and are only changed once so that when 1 is initially recoded as 5 (1->5) in the above example, it is not subsequently reconverted to 1 (5->1). If there are missing values, as in this case, select System- or user-missing (in Old Value section) and System-missing (in New Value section).

After doing this we select Continue to close the Recode into Different Variables: Old and New Values subdialog box and OK to close the Recode into Different Variables dialog box. We can then check the recoded values in the appropriate columns (satis2, rsatis2, satis4, and rsatis4) in the Data Editor.

Alternatively, we can check the recoded values of rsatis2 and rsatis4 by selecting Summarize Cases and listing the values of satis2, rsatis2, satis4, and rsatis4 for, say, the first 10 cases as shown in Table 3.2.

ªStatistics ªSummarize ªCase Summaries…[opens Summarize Cases dialog box shown in Box 3.5]

ªselect variables [e.g. satis2] ª䉴button ªLimit cases to first and type number [e.g. 10]

ªOK

If we do not want to retain the original values, we select the Into Same Variables…option which opens the Recode Into Same Variables dialog box presented in Box 3.6. For example, if we want to recode satis2 and satis4 we select them, which puts them in the box entitled Variables:.

We then proceed as previously by selecting Old and New Values…which opens the Recode into Same Variables: Old and New Values subdialog box displayed in Box 3.7. In the box entitled Value: (in the Old Value section) we type the first value to be changed (e.g. 1) while in the box called Value: (in the New Value section) we type the new value (e.g. 5) and select Add. We do this consecutively for the three other old values of 2, 4 and 5. We can also do this for Box 3.4 Recode into Dif ferent Variables: Old and New Values subdialog box

48 Further steps with SPSS 8 for Windows

Table 3.2 Case Summaries output showing receded values of rsatis2 and rsatis4

Box 3.5 Summarize Cases dialog box

3. If there are missing values, as in this case, select System- or user-missing (in Old Value section) and System-missing (in New Value section).

Alternatively, we can select All other values, type 3 in the box called Value: (in the New Value section) and select Add.

After doing this we select Continue to close the Recode into Same Variables: Old and New Values subdialog box and OK to close the Recode into Same Variables dialog box.

Box 3.6 Recode into Same Variables dialog box

50 Further steps with SPSS 8 for Windows

For each variable there can only be one new value, whereas there can be any number of old values. For example, if we wished to form a 3-point scale with only one agree, one disagree, and one undecided answer, we could do this by receding 1 and 2 into 1, 3 into 2, and 4 and 5 into 3. Since only one value can be entered at a time into the box entitled Value we would have to do this sequentially.

We can specify a range of old values by selecting the Range: option (which is distinguished by having the n of Range: underlined) and typing the lowest value in the first box and the highest value in the second box. For example, we could recode ethnic group into whites and non-whites by putting 3 into the first box and 5 into the second box and receding 3 through 5 as 2.

If we do not wish to find out what the lowest value is, we can specify it by selecting Range:. If we wanted to specify the highest value without determining what it was, we would select Range:. For example, we could use these options to categorize our sample into those 40 years old or above and those below by receding the Lowest through 39 as 1 and the 40 through highest as 2.

If we had ages which were not whole numbers and which fell between 39 and 40, such as 39.9, they would not be recoded. To avoid this problem, we would use overlapping end-points so that Lowest through 40 is recoded as 1 and 40 through highest as 2. In this example all people aged 40 and less would be coded as 1. Since values are recoded consecutively and once only, age 40 will not also be recoded as 2.

In document Quantitative Analysis With SPSS (Page 62-67)