This brief tutorial is intended to assure the casual and/or first-time user of Dexter that the application does not have to be as complicated as the input form makes it look. To be sure, some extracts can get pretty complicated, depending on the dataset and what you need to do with the data. But in many cases the program can create useful extracts with little or (almost) no input specifications from the user. The general rule of thumb is that Dexter makes it very easy to extract a lot of data; it only gets more challenging when you try to get exactly what you want.
Note that the section headers on the Dexter query form are also hyperlinks to detailed online help.
The Dexter query form is divided into 5 sections:
...you can limit the # of observations/rows on each output by entering a number here
. By entering a small number (50, for example) you can quickly run a test of the extract and examine the results before going back and doing the final complete extract. Most datasets have variables that are keys to identifying the meaning of the data contained in a row; such identifiers are often geographic codes such as State, County, Place, etc. If you know the codes used for these variables it can be pretty simple to create a filter. For example, if the dataset you are accessing indicates it has data for every county in the U.S. and you are only interested in data for the state of California then you can specify this by making choices/entries in the first of the 5 rows of entry boxes in Section II. (This is by far the hardest part of what we'll be dealing with here -- but remember, many extracts can be done without having to do any such "filtering"). The form uses drop-down menus in the first (leftmost) two columns of Section II, while the 3rd column is a text entry box where you need to type in a value.
Note: there are also checkboxes down the left and right sides of the page in Sec. II that are used to insert parentheses into the logical expression. These are rarely used (only for complex logical expresssions involving 3 or more conditions) and should almost always be ignored, especially by new users.
Assuming that the dataset you are accessing has a variable (field) named State and that State contains a 2-digit FIPS state code (which it always will in our data collection) then you would select State from the pull-down in row 1, column 1; then select Equal To (=)
as the value for Operator from column 2. The 3rd column is the hardest because you have to type in something with no menu from which to choose. Enter the value 06
in the text box. To do this you have to know your FIPS state codes. You also have to understand the importance of typing the leading zero because this field is not stored as a numeric value, but as a character string. As a character string "6" is not equal to "06", so the leading 0 is required to get the proper filter. Typing a value in the 3rd column that does not match any of the values for the specified variable would result in filtering (i.e. eliminating from the resulting output) all the data and after you hit an Extract Data button would result in an error message telling you this and asking you to "Please check your filter ... and try again". At which point you simply click on your browser's Back button and try again.
There are lots of examples of how to code filters in the Dexter online tutorial, and a discussion of how to use the metadata to access web pages showing you the values and meaning of those values for many key variables on the datasets. Two commonly used sources for viewing the values for coded variables are:
See detailed metadata for this dataset.
, where the words detailed metadata are a hyperlink. Click on that link to go to the detailed page. As part of the description of the dataset on this metadata page there will be a section near the top of the page labeled
Key variables for identifying ...:
. The following row is a list of the key variables which are hyperlinks; clicking on one of these links brings up a page that displays the values that the variable takes on in that dataset along with their meaning and a frequency count indicating how many times that value occurs on the set.
For some datasets (those having more than 100 numeric variables) you will see extra options below the select list labeled Filter by regular expression
. We suggest first time casual users just ignore this feature until later. What it allows you to do is to enter something in the text box and then click on the Filter
button to have the list regenerated so that only entries that match the text box entry appear on the menu. This can be very handy when the list has hundreds of entries. But for the large majority of datasets and queries you will never have to use this feature.
To invoke the Dexter program to have it execute your query you need to click one of the Extract Data
buttons found at the end of sections III through V. (You can ignore the extra row of options that follow the Extract Data button at the bottom of the page; these options are intended for internal use by MCDC staff primarily - at least for now.)
The results of the query will be displayed back to your browser. Wait for it to display a message saying Processing completed at <time-and-date>.
. Note that at the the top of the screen - on the 2nd line - you should see a job identifier code. This code will look something like
17OCT05_00003
- a date followed by an underscore and then a 5-digit number. If you encounter an error while using Dexter that you do not understand, you need to report the error to the
Questions and comments regarding Dexter e-mail link at the bottom of the Dexter query form and you need to include this job ID as part of the message you send.
Hopefully, such problems will be rare. If the query works you will see a bolded, underlined link to a Summary Log page and then to the requested output file(s): delimited, report and/or database. The Summary Log page is your hard-copy record of what you specified and provides a concise record of the query. If you need to document your source we strongly recommend that you save this file somewhere.
The links to the output(s) will be to temporary files created in a quasi-temporary directory created when you invoke Dexter. This directory is automatically deleted after 48 hours but can be accessed until then. You do so by simply clicking on the link; what happens at that point will depend on how your browser had been configured to handled files with the specific file extension. These extensions are:
csv
for comma-delimited files
txt
for tab-delimited files
prn
for plain text report files
html
for HTML report files
pdf
for PDF report files
dbf
for DBF database files
sas7bdat
for SAS database files (Windows format for V9)
Note that you can share your results with colleagues or clients by right-clicking on the file links and choosing the Copy link location option to capture (copy to the Windows clipboard) the URL of the file. You can then send e-mail to the person you would like to share your results with, pasting this URL into the body of the e-mail message. This technique is used frequently by MCDC personnel who run queries for users and then simply e-mail a link to the results to the user. (Just remember the 48-hour time limit. This is not a good idea if you are running something late on a Friday afternoon, since the file may be gone by the time the user gets the message first thing Monday morning.)
You can always ask for help with your query. The ideal situation is to have a query system where users can go and access a wide variety of data strictly on their own without any assistance from the people who built the system. While we continue to work toward that ideal, the current reality is that in very many cases users -- especially new ones not familiar with the software and the data archive -- will simply not be able to easily find and/or extract the data they need. These users should keep in mind that the Missouri Census Data Center is housed within the Missouri State Library (within the Secretrary of State's office) and is funded by the state of Missouri specifically to provide assistance to the public in accessing the public data found in this archive. Not asking for help with your data query would be like going to the public library and being afraid to ask the librarian to help you find a book. (Helping you find the book, of course, is not the same as reading the book to you, or telling you what it "means".) As a participant in the Census Bureau's State Data Center program, the MCDC has an affiliate network of agencies throughout the state of Missouri who are available to assist you with accessing these data (see the About MCDC page for details). You can also use the e-mail link at the bottom of the Dexter query page (and this page as well) to ask for assistance. We cannot guarantee, of course, that we have the data you are looking for, or that Dexter will be the appropriate tool for accessing the data and getting it into the format you require; but we are always available to assist you as best we can with the data and tools we have available. In many cases, MCDC personnel can run the query and point you to the results much faster than they can explain to you how to do it. That is often the best solution if you think you are in that category of first- and last-time user.
The MCDC has a number of online powerpoint tutorials related to Dexter and Uexplore. To see what's available visit the Uexplore/Dexter tutorials web page. You can also see a collection of detailed, annotated examples of Dexter queries at the X Samples page.