A data file is divided into records (lines). A record is a collection of values. Each value is an occurrence of a variable. For example, a line in the "vehicle" file has 16 values. The first value corresponds to the household number, the second value to household vehicle number, and so on. There are different types of variables, discrete, continuous, ...
To access data and to make statistics or counts of the variables, it is necessary to have a login and a password. This menu points to the registration for this information.
To obtain access, you must complete the registration form. All fields must be filled before the submission, otherwise an error message will ask to complete the form.
The requested fields are :
After having filled all the fields, you must click on the 'submit' button to send your request. An e-mail is then automatically sent to the webmaster and you receive a confirmation page.
When the webmaster receives the mail, she creates the access and replies to let you know that your access is activated with the chosen login and password.
The first time you choose one of the "frequency", "statistics", "crosstables" or "creation of classes" menus in a given session, you must enter your login and password to access the referring page.
The login and the password have to be filled in only once a session.
Otherwise, a table is presented with all the variables of the selected file. The user chooses one of them or more by clicking on the button associated with its/their name(s). He can also click the ``bar chart'' selection to produce such a graph. When this process is completed, he clicks on the ``continue with selected variable(s)'' button to go further (or on the ``reset'' button to revise his choice(s)).
The results page is then displayed. For each value of the selected variable, a table presents the number of times it occurs in the survey and the percentage this represents. The total is also displayed. For privacy reasons, if this number is under a given threshold (determined by the site maintainer), no value is given and an ``N S'' (for non significant) sign is displayed. If a bar chart was previously chosen, it is now displayed. Otherwise, it is still possible to click on the appropriate button and re-display the results page with this graph when clicking on the ``continue with this choice'' button.
The user may also want to produce the table with the results in the Excel format. Under this format, his browser will propose to open this spreadsheet software to display the results or to save the results file on his disk. This option is selected by clicking on the ``Excel format'' label. Finally, the user can also decide to undertake some merging (see associated help section) on the results.
There are two aspects in this item. The first (mean values) gives the statistical values computed on a continuous variable. The values which are computed are the sample size, the total of all the values for this variable, the mean, the median, the minimum and the maximum.
The second (weighted analysis) computes the mean values of such a continuous variable with respect to the values of a second variable (e.g. the mean age by sex).
On the first page, as usual, the user selects the data file where he wants to pick up a continuous variable. He can also select a filter. When clicking on the button associated with this file, a new page is displayed with all the continuous variables contained in this file available for statistics (see the $stat attribute in the definition of the ``dataXXYY.pl'' file). The user may then click on the button(s) associated with the variable(s) for which he wants statistics and then pushes the ``continue with selected variable'' button. The page thereafter produced redisplays the selected variable(s) and presents a ``select another variable'' button to reset this choice and come back to previous stage.
The frame also allows the selection of either ``compute statistical values'' or ``make weighted analysis''. In the first case, a results page is merely displayed with the computed statistical values.
In the other case, the user is asked to select a data file where he wants to pick up a second variable which will be used to discriminate and sort the values of the first (continuous) variable. The usual process occurs: click on the button associated with the chosen file and then on the page thereafter displayed select a variable by clicking on the radio button corresponding to it and go further by clicking on the ``continue with selected second variable'' button. The results page is then displayed: it presents a table with the means and sample sizes of the first variable computed for each value of the second variable.
If the user has chosen a weighted variable, the first option displays unweighted and weighted sample size, weighted total and weighted mean. The second option will produce a table with weighted and unweighted sample sizes and weighted means.
With this item, the user can cross tabulate one variable against another, using filters if required (e.g. method of travel by the purpose of travel for people between 20 and 30). This table can also display percentages by column, by row or even of each cell relative to the total.
After the usual opportunity to select a filter (see filter section ), the first page displays the current choices for row and column variables. There are the ones selected by the user the last time he visited this page. If it is the first time he selects this item, by default, the row variable will be ``usual means of travel to work'' and the column variable, ``sex'', both from the people data file.
Two buttons offer the user the possibility to change the row variable (``change row variable'' button) or the column variable (``change column variable'') button. Both lead to a process where a page is displayed to select a data file to pick the variable type (household, vehicle, etc.). The next page allows theselection of the variable by clicking on the radio button corresponding to it (this process updates the table variables displayed on top of this page) and then pushing the ``continue with selected variable'' button.
When the choice of row and column variables is completed, the user is taken back to the first page where he can now select the type of crosstable he wants. Four kinds of tables are possible:
When the resulting crosstable is displayed, the user has the opportunity to merge values on the rows and/or on the columns (see merging section ).
Don't forget that some variables cannot be crossed. For example, a variable from the vehicle data file cannot be crossed with a variable contained in the people file. This is due to both the structures of the files and the way data are linked in the survey and, according to that, in the database. If the user makes such a selection, a page with a warning will be displayed and he will have to go back to previous page to revise his choice.
This item is an extension of the previous one: a crosstable is created for each value of a previously first selected variable. All the process for producing these 3 ways tables is similar to the one described in the previous subsection except that, over the choice of the row and column variables, there is also a selection of a ``first'' variable (i.e. for each value of this variable, a crosstable will be produced with the row and column variables). For these variables, the default values (i.e. the values used the first time the user selects this item) are ``sex'' for the first variable, ``age (banded)'' for the column variable and ``usual means of travel to work'' for the row one (all contained in the ``people'' data file).
Note that no merging is allowed for the first variable. Moreover, the mergings achieved for the row and/or the column variable applies on each of the crosstables (i.e. for each value of the first variable).
This last item allows the user to cluster values of a variable into classes giving a new name to this newly created variable and also names to each of the formed classes. Doing so the user is thus creating a new variable with new values derived from a variable from the survey. For example, he can define a ``stage of life'' variable from the ``age'' variable with the ``child'', ``teenager'', ``adult'' and ``old'' values clustering respectively the ranges of values: 0 to 12, 13 to 18, 19 to 59 and 60 to 99.
A first page is dedicated to the choice of the data file containing the variable from which the classes will be created. Clicking on the button associated with this file leads to a second page where all the available variables are displayed. The user selects one variable by clicking on the corresponding radio button and then pushing the ``continue with selected variable'' button.
Two cases can then occur:
The ``continue with creation of this class'' button completes the creation of the new ``classes'' variable. Its definition will be stocked in the area owned by the user on the Web site until he removes it through the ``user's page'' item. This variable will appear in all the subsequent variable choices the user will encounter in all computation items (as a new variable in the data file containing the original variable from which this new class variable was derived).
The results of previous queries are retained into a particular area owned by each user of the site. So, during later visits to the site, he can come back and view again previous results tables. He can also choose to delete some of them. This ``user's page'' presents a frame with all these stocked results and their creation date. By clicking on their names, he can display them. He can also remove them by clicking on the radio button associated with the results he wants to delete and then clicking on the ``delete'' button.
As well, the variable classes he has created before are retained in the same area. Their characteristics are displayed when the ``user's page'' item is selected and a ``delete'' button associated with each of them allows the user to remove it
The ``Feedback'' item presents a sheet that the user can fill in to send a e-mail to the site maintainer; he has to give his name and his e-mail address, to select a subjet by clicking on one of the radio buttons associated with each given proposal, to type his message in the appropriate frame and to send (or reset) his mail by clicking on the corresponding button.
All items allowing the user to formulate some statistical queries on the survey data offers him the opportunity to first select a filter to be applied on the data before the computations. A filter is a condition selecting certain data items only (e.g. the user could ask to compute the frequency of each value of ``usual means of transport to work'' variable filtered for people over 40 years only).
It must be mentioned that some crossings are not allowed (for example, the user can not compute statistics on data from the vehicle file using a filter on a variable from the people file). This is due to the links present or not between the data files (and marked by the indices) (e.g., from the survey, it is known that a vehicle belongs to a given household but nothing indicates if this vehicle is linked to one or another member of the family). Nevertheless, if such a situation occurs, a warning message will be displayed and the user will have the opportunity of trying another computation (e.g. with another filter).
For the filter selection, the first page associated with each of these items presents two buttons: a ``filter'' button dedicated to this purpose and a ``delete'' button to remove the current filter. Note that once a filter is selected by the user, it remains active (and it is displayed on the first page associated with each computation item) until he decides to delete it by clicking on the ``delete'' button.
When the user clicks on the ``filter'' button, a new page is displayed. On this page, the user must choose the data file where he wants to select variable(s) to use as filter(s). He also has the opportunity to cancel his choice of selecting a filter thus coming back to previous page. By clicking on the button corresponding to the chosen file, a new page appears. On this page, the user can revise his file selection and come back to the previous page or he may even to decide to cancel his filter selection and then come back to the page where he made this choice (i.e. the first page associated with the current computation item).
But, normally, he selects the table displaying all variables available for filter in the chosen filter and selects the variable(s) he wants to include in the filter. Then, he clicks the ``continue with selected variable'' button to proceed further. The next page offers the possibility to go to another file to select additional variables for the filter. The user can thus repeat this process until he has picked all the variables he wants for his filters (possibly in different files). Once all the filter variables are selected, he then clicks on the ``done'' button. Then, a new page is displayed.
The user must select the values he wants to include in/exclude from the filter for the selected variables. A toggle allows him to select the inclusion or the exclusion of the selected values (for each variable). For discrete variables, he has to click on the values he wants to consider; for continuous variables, he must give the bounds of the range of values to take into account. When the ``select filter'' button is clicked, a new page offers to combine this new filter with the current one using ``and'' or ``or'' connector (provided a filter is already present before this selection). Then a new ``select filter'' button makes the selection definitive. When this process is completed, the site must refresh and displays again the first page associated with the computation item from which the user started his filter selection.
When results are produced after processing in one of the computation items, the user may decide to merge some values. To do this, he clicks on the buttons associated with each value he wants to merge and gives a name for this merger in the appropriate window. Then, he clicks on the ``continue'' button and the results page is re-displayed but, now, with the results collapsed for all the merged values. The name of the merging as well as the values merged are also displayed. The user can repeat this process anytime he wants (e.g. he can merge mergings).