Walden University SPSS Testing for Bivariate Categorical Analysis Paper

SOLUTION AT Academic Writers Bay

Unformatted Preview
Bivariate  Categorical  Tests   Bivariate  Categorical  Tests   Program  Transcript     [MUSIC  PLAYING]     MATT  JONES:  Up  to  this  point,  we’ve  been  focusing  on  statistical  tests  that   require  metric  or  variables-­-­  that  is,  variables  measured  at  the  interval  or  ratio   level.  But  there  are  a  lot  of  categorical  variables  that  are  of  use  to  the  social   scientist.  We’re  going  to  cover  the  chi-­square  test  for  independence  and   associated  measures  of  effect  of  Cramer’s  V in  SPSS.     Let’s go  to  SPSS.  We  can  test  for  the  relationship  between  two  variables  by   using  the  chi-­square  test  for  independence.  Let’s go  ahead  and  test  the relationship  between  gender  and  views  on  marijuana  legalization.     To  do  this,  we  go  to  Analyze,  Descriptive  Statistics,  and  Crosstabs.  Here,  you  will   see  a  place  to  put  a  variable  in  a  row  and  a  column.  I’m going  to  scroll  down  and   find  my  SHOULD  MARIJUANA  BE MADE  LEGAL  variable  and  enter  that  into  my   row,  and  I  will  scroll  down  to  find  the  respondent’s gender  and  place  that  into  my   column.     I’m going  to  go  ahead  and  hit  OK  to  show  you  the  output  that  we  receive.  And   here,  you  will  see  some  output  that  are  basic  Descriptive  Statistics.  These  are   counts  of  the  number  of  males  and  the  number  of  females  who  felt  that  marijuana   should  be  either  LEGAL  or  NOT  LEGAL.     However,  this  does  not  statistically  test  for  a  relationship  between  these   variables.  We  can  request  the  chi-­square  statistic  by,  again,  going  back  into  our   Crosstabs  box.  So  I  perform  the  same  procedure  of  going  to  Descriptive   Statistics,  Crosstabs,  and  all  of  my  information  is  still  there,  so  I  can  select   Statistics.  You’ll  see  that  the  Chi-­square  statistic  comes  first,  but  I  have  to  go   ahead  and  activate  that.     I’m also  going  to  go  into  the  section  Nominal  to  ask  for  Phi  and  Cramer’s  V.  This   will  tell  me  something  about  the  strength  of  the  relationship  between  the  two   variables.  As  you  know  from  your  reading,  the  chi-­square  tells  us  whether  there   is  a  relationship,  but  it  doesn’t  tell  us  anything  about  the  strength  of  that   relationship.  Find  Cramer’s  V help  us  with  that  follow-­up  should  we  have  a   significant  relationship  with  a  chi-­square.  Continue.   OK.  So  I’m going  to  hit  Cells.  Just  for  ease  of  interpretation,  I’m going  to  request   Percentages  for  Columns.     I’ll  hit  Continue  and  OK.  Here,  you  see,  I  receive  some  Case  Processing   Summary.  This  tells  me  that  there  are  920  valid  cases  in  this  analysis.  580  cases   are  missing.  So  out  of  the  1,500  cases  or  respondents  of  the  survey,  we  have   ©2016  Laureate  Education,  Inc.   1 Bivariate  Categorical  Tests   quite  a  few  of  them  that  either  didn’t  answer,  refused  to  answer,  or  just  left  that   blank.     The  next  piece  of  output  is  the  Crosstabulation  table.  You  can  see  this  looks   similar  to  the  Crosstabs  I  asked  for  in  Descriptive  Statistics  with  just  the  raw   counts,  but  now,  I  also  requested  for  the  percent  within  respondent’s sex.  This   tells  me  55%  of  the  males  believe  that  marijuana  should  be  made  LEGAL  and   44.6%  of  the  males believe  that  marijuana  should  be  NOT  LEGAL  for  a   cumulative  percentage  of  100%.  I  can  interpret  the  female  column  as  the  same   way.  41%  of  females  believe  that  marijuana  should  be  LEGAL  while  59%  believe   that  it  should  be  NOT  LEGAL.  If  there  was  no  relationship  between  these  two   variables,  we  would  see  approximately  equal  percentages.     To  statistically  test  for  this,  we  can  look  to  our  chi-­square  statistic.  Here,  we  see  a   critical  value  of  18.993  with  an  associated  p-­value  of  0.001.  This  test  is  significant at  the  0.01  level  and  certainly  well  below  the  common  0.05  threshold.  Therefore,   we  can  reject  the  null  hypothesis  that  there  is  no  relationship  between  the  two   variables  assuming  that  there  is  some  sort  of  relationship  between  gender  and   position  on  marijuana  legalization.     But  once  again,  we  don’t  know  the  strength  of  that  relationship.  We  can  scroll   down  to  our  Cramer’s  V correlation,  which  tells  us  about  the  strength  of  this   relationship.  A  value  of  0  indicates  no  relationship  whatsoever,  and  a  value  of 1.0 indicates  a  very  strong,  perfect  relationship.     We  can  see  here,  we  have  a  value  of  0.144.  So  while  there  is  a  relationship,  it’s important  to  do  the  follow-­up  test  to  determine  the  strength  of  that  relationship.  In   this  case,  the  relationship  between  these  two  variables,  which  is  statistically   significant  at  the  0.01  level,  is  rather  weak.     [MUSIC  FADING]       ©2016  Laureate  Education,  Inc.   2

CLICK HERE TO GET A PROFESSIONAL WRITER TO WORK ON THIS PAPER AND OTHER SIMILAR PAPERS

CLICK THE BUTTON TO MAKE YOUR ORDER

error: Content is protected !!