« Previous | Contents | Next »
Listen
SHS Lite - User Guide: A guide to using the Scottish Household Survey simplified dataset
4. Variables and Weights
4.1 Variable Naming
Most of the variables in the dataset are derived directly from the answer given to a question in the questionnaire. In many cases, the variable name will be the same as the question number.
Example variables
- HA2 - Highest income householder
- HC4 - Number of bedrooms
- HD8 - Number of motor vehicles
- RA1 - How long lived at current address
Looping Questions
Some variables are asked of all household members, creating up to ten responses for each household from the same question. The variable names for these 'looped' questions have a common root based on the question number plus an additional number to indicate the household member referred to. For example, question HA5 asks the age of up to 10 household members - the resulting variables are therefore named HA5-1 to HA5-10.
Examples of 'looped' variables
- HA7-1 to HA7-10 - Economic status
- HA9-1 to HA9-10 - Ethnic origin
Some questions allow the respondent to select a number of responses that reflect their views. For example, question RB2 asks respondents to say what it is about their area that they like. There are 15 variables recording these options, with Yes/No responses for each case. These types of multiple response variables are named by taking the root from the question name (RB2) and adding letters to indicate each response option (in this case a-o). The variables are therefore named RB2a to RB2o.
4. 2 Viewing Variable Information
A Variables dialog box is available that displays definition information for the currently selected variable (see Figure 5). This includes data format, variable label, user missing values and value labels.
Viewing Variable Information
- Select Utilities, Variables... from the menu bar or click
- Select the desired variable from the variable list on the left
Figure 5 - The Variables dialog box
! Note | While in Data View, you can use this dialog box to quickly navigate to one of the variables by selecting it in the variable list and clicking the button. |
4.3 Display Variable Names in Dialog Boxes
By default, SPSS displays Variable Labels in dialog boxes instead of Variable Names (see Figure 7 and Figure 8). You may want to change this setting because it can be much easier to select the variable by viewing the name rather than its lengthy label. To change this, you need to set the options in SPSS as follows:
- Select Edit, Options... from the menu bar (see Figure 6)
- Choose the General tab
- Within Variable Lists, select the Display Names option
- Click
- Click again to accept these changes
Figure 6 - The Options dialog box
Figure 7 - Frequencies dialog box showing Labels 
| Figure 8 - Frequencies dialog box showing Names 
|
4.4 Variable Sets
4.4.1 Using Sets
Using Sets restricts the variables displayed in dialog boxes to the selected sets that you have chosen. Small variable sets make it easier to find and select the variables for your analysis and can also enhance performance. A full list of variables grouped together by analysis set can be found within the file named SHS Lite Variable Listing.pdf on the accompanying CD.
Using sets
- Select Utilities, Use Sets... from the menu bar or click
on the toolbar - Select the current sets in use and remove them by clicking
- Select the sets that you would like to use and add them by clicking
- Click
Figure 9 - Selected sets to be removed 
| Figure 10 - Selected sets to be added 
|
| ! Note | You will always require the ADMIN set because it contains the weighting variables. |
| ! Note | Each time you close the SHS Lite dataset, the variable sets will return to ALLVARIABLES and NEWVARIABLES . |
! Note | To select consecutive sets, click the first set, press and hold down the [Shift] key on the keyboard, and then click the last item. To select non-consecutive sets, click the first set, press and hold down the [Ctrl] key on the keyboard, and then click each additional set. |
4.4.2 Defining Sets
It is possible to create further subsets of variables. This can be very useful when analysing because you will only see the variables contained in currently selected sets within dialog boxes such as frequencies. Set names can be up to 12 characters long and can include spaces. Any combination of numeric and string variables can be included in a set and any variable can belong to multiple sets.
Defining variable sets
- Select Utilities, Define Sets... from the menu bar
- Select the variables you would like to add and click
- Type a name for the new set and click
- Click
Figure 11 - Selected variables to be added to the new set
! Note | The  button will only become available once you have typed a name for the new set. |
4.5 Variable Recoding
You can recode variables into new variables. This will make it possible for you to reassign the values of an existing variable to a new variable. For example, you could group household ages into a new variable containing age range values. An illustrative example of how to do this is given in Section 8.7 on page . For easy reference however, the basic steps that you would need to take are listed below.
Recode values of a variable into a new variable
- Select Transform, Recode, Into Different Variables... from the menu bar
- Select the variable you want to recode (e.g. randage) and click
! Note | You can select a variable by typing the start of its name instead of scrolling through the list of variables. For example, type r twice to select the randage variable. |
- Enter a name for the output variable (e.g. randage2)
- Enter an optional label for the output variable (e.g. Age ranges of random adult)
- Click
Figure 12 - Recode into different variables after performing the steps listed above
Click 
to recode the values
- Specify the Old Value or Range of values
- Specify the New Value and click
- Repeat these steps for all values or value ranges to be recoded
- Click
- Click
Figure 13- Old and New Values showing 2 ranges added as new values 1 and 2
! Note | The new variable will be displayed at the end of the dataset. |
4.6 Defining Value Labels
Value labels provide us with a useful description for each of the variable's values. For example, if you have recoded a variable, this will allow you to give meaningful labels to the new values. Figure 14 (below) shows some sample Value Labels. These labels can be used to view the information in Data View ( see Section 3.2.2).
Define Value Labels
Figure 14 - Value Labels showing labels added for values 1 and 2
4.7 Weights
Although the SHS covers all local authorities, the sample does not represent each authority in proportion to the population distribution between authorities: small authorities are over-sampled to allow analysis of individual local authorities after two years. This means they have more interviews than a proportionate allocation would give them. This is compensated by the fact that some local authorities - the larger ones - are under-sampled. When the data for the whole of Scotland is analysed or comparisons are made between two or more authorities, the data need to be weighted to ensure that each local authority represents the correct proportion of the population.
Similarly, the random adult data need to be weighted both to correctly represent local authorities and to account for the fact that adults in large households have less chance of being sampled than adults in smaller households. Random adult data always need to be weighted.
The correct weight to be used with each of the variables can be found within the file named SHS Lite Variable Listing.pdf on the accompanying CD. Where there is no random adult or random child data, the value of the weight will be zero.
LA-WT | This is the weight that adjusts for differences in sampling fractions and response rates between local authorities. This should be used when analysing household, household member or vehicle variables. This includes all variables beginning with H (except those from HE6 to HE17) and derived household variables about the household, the highest income householder (HIH) or the spouse of the HIH. |
IND-WT | This contains the individual weight to be used when analysing the Random Adult data. This includes all variables beginning with R and the derived random adult variables. |
KID-WT | This contains the individual weight to be used when analysing the Random Schoolchild data (variables from questions HE6 to HE17 and the derived random schoolchild variables). |
The weighting is straightforward when the variables being analysed all need the same weight. In cases where you want to mix household and random adult data, the weight needs to be IND-WT. For example tenure is a household variable and travel to work is a random adult variable. A table of tenure by travel to work would be weighted by IND-WT because tenure is being used as a characteristic of the random adult. The rules to adopt are:
- Household variables - use LA-WT
- Random adult variables - use IND-WT
- Combination of household and random adult variables - use IND-WT
- Combination of household and random schoolchild variables - use KID-WT
! Note | You can only choose 1 weight. IND-WT and KID-WT both incorporate LA-WT. |
! Note | For more information on weighting see the SHS Technical Report on the CD. |
Weighting Data
Figure 15- Weight cases showing LA-WT as the Frequency Variable
! Note | The Status bar at the bottom of the screen indicates that a weight has been applied. |
! Note | Once you apply a weight variable, it remains in effect until you select another weight variable or turn off weighting. If you save a weighted data file, weighting information is saved with the data file. You can turn off weighting at any time, even after the file has been saved in weighted form. |
« Previous | Contents | Next »