example.  Only  12  percent  of  the  nation’s  dailies endorsed him. Clearly, what is on the editorial page does not represent public opinion. However, content analysis often shows trends in the media’s  agenda  setting  function.  Most  press  researchers agree  that  mass  media  is  not  effective  at  changing opinions  on  any  given  topic,  but  deciding  what  will  be discussed  on  a  national  scale  (agenda  setting)  is  most often  accomplished  by  media  coverage  of  a  topic. Knowing what topic is coming to the forefront of media attention would be valuable information to public affairs practitioners,  especially  if  such  a  topic  were  to  be military drug use or shipboard safety. A good public affairs office can then be proactive and check into local procedures  or  problems  and  bring  together  a contingency plan or perhaps even point out trouble areas to  those  concerned. Mail  Analysis In relation to mass media, content analysis is mail analysis. This form of research is economical and often overlooked. It is not to be used as hard scientific data, such as the formal surveys you will soon read how to do, but it may serve as early warnings on sources of ill will  or  problem  relationships.  For  most  military  print, broadcast and public affairs activities, reading every letter received from the public is usually not a tough assignment, and it is strongly suggested that each letter be answered. Few stations or offices have ever been inundated  with  bags  of  mail  that  would  require  random selections  of  letters  to  be  read.  This  type  of  data collection is mentioned here to make you aware that letters can be more than just communication between a listener and a disc jockey. They may be the tip of a controversial  iceberg. Call-in Telephone Lines During  Operation  DESERT  STORM,  CHINFO operated  a  toll-free  telephone  line  to  answer  questions from military family members. The line was a huge success and soon became the number to call even for the family  members  of  other  services.  The  reason  for  such a huge success was due, in part, to the analysis of the trends found in the calls. One such trend that the Navy exploited  was  the  huge  amount  of  calls  from Spanish-speaking  families.  By  chance,  during  the  first few  days  of  operation,  a  Spanish  speaking  person  was available to answer questions. Since no other branch of service was able to handle non-English calls, the Navy was soon taking them all. Also, at the time, CHINFO saw  the  trend  in  non-English  calls  and  institutionalized having a Spanish-speaking staffer on hand to take such calls. This  is  a  good  example  of  moving  with  a  fluid situation and of the use call-in telephone lines may be put to in identifying areas of concern. Was the military family  audience  in  this  case  known?  Not  well  enough. Was everything that was being released about the Gulf War going to all of the internal audiences we thought they were? Evidently not. A good lesson was reiterated about knowing the intended audience. It was also a good example of using data that has already been collected, and in particular, using data collected from a call-in telephone  line. FORMAL  SURVEYS For those of you who have been wondering when this chapter on surveys was going to begin looking like a chapter on surveys, your wait is over. This next section will begin dealing with more traditional forms of survey research  techniques. Telephone Surveys The telephone is arguably the cheapest way to do a survey. Most students of journalism know about the early  telephone  surveys  that  did  not  take  into  account “who”  the  owners  of  telephones  were.  In  1936  the Literary   Digest   mailed  out  10  million  surveys  to discover who was going to win the presidential race between  Franklin  Delano  Roosevelt  and  Alfred  Landon. This  was  the  largest  sampling  of  a  population  ever attempted up to that time. The bias of the poll was later discovered  to  be  how  the  “Digest”  found  the  addresses of  their  would-be  respondents.  They  used  the  telephone book. At that time in America, rural areas, as well as many inner  city  areas,  had  substantial  populations  without telephone access. This affected the results by surveying more well-to-do people who could afford telephones. Franklin Delano Roosevelt was planning for a large vote from the urban and rural poor. You can see how the data gathered over the telephone would have favored the candidate  trying  to  represent  the  wealthier  telephone owners.  The  poll  predicted  Landon  by  a  large  margin. Just the opposite occurred with Roosevelt winning in a landslide,  62  percent  to  38  percent.  Telephone ownership   is   not   much   of   a   problem   for   most presidential  pollsters  these  days.  However,  it  is  still  a problem  when  you  conduct  a  telephone  survey  and many   members   of   your   audience   live   in   the 9-3