Pure Random Generation

  Previous topic Next topic JavaScript is required for the print function Mail us feedback on this topic! Mail us feedback on this topic!  



Pure Random Number Generation screen


You can generate up to 9,999 pure random phone numbers at a time.


Here's how pure random phone numbers are generated:


An Area Code/Prefix combination is selected at random from your prefix list.  (These can be added/changed/deleted by choosing Data Management / Random Numbers / Change Prefixes from the menu).


The last four digits of the phone number are generated randomly.  For example, if the Area Code/Prefix combination of (619) 268 is selected, the phone number (619) 268-XXXX is generated, where XXXX is any random number between 0000 and 9999.


The number generated is checked against phone numbers presently in your system.  If the number exists, it is not added and the process to add a new number starts all over again, by selecting another random Area Code/Prefix combination.  If the phone number does not already exist, it is added to your respondent's file.


If you want the generated random numbers to skew more heavily toward specific Area Code/Prefix combinations, you can enter them more than one time in Change Prefixes.


HandCalling random numbers will significantly lower your "hit" rate of reaching qualified respondents; perhaps by as much as 70-80%.  


The reason for this is that we now have multiple area codes in most markets.  Many of these exchanges include many non-working or "reserved" numbers; others are for "non-human" purposes; fax machines, pagers, cellular phones (which are mostly turned off), data lines, etc.


Twenty years ago, you could call any number at random, and there was a pretty good chance a human would answer it.  Now, with voice mail, answering machines, and the previously discussed mechanical components, it is difficult to get good results using pure random phone numbers.


Here is one analogy:  If you were trying to locate people who drove VW Beetles, you could:


A) Go house to house and knock on people's doors at random, and ask them if they had a VW Beetle in the garage.  You would run into: people not home, locked gates, and a lot of footwork required to find 100 people who drove bugs.




B)  Buy a list from the DMV.


Both produce the same result; a list of people who drive VW Beetles.  The first, while generated randomly, is very time consuming and not generally budget-worthy of undertaking.


So, it seems if you were trying to survey a specific group, for example, Women 25-44 with a propensity to listen to Country music, you could achieve this in a similar manner.  Call randomly until you find enough respondents, or call from a list that has already been pre-screened based on demo, gender and either cluster group or zip code based on high listening levels to Country music.


As you have probably deduced, we are not big fans of pure random calling.  ComQuest is designed to efficiently administer multiple music tests at the same time.  It's not much more productive at random calling, or finding a needle in a haystack, than just fielding those calls manually.


See also

Changing Prefixes

Random Number Generation