Chapter

 4  
Theore,cal  framework  and  
hypothesis  development    
 

Ali  BELAIDI  
Research  ques3ons  are  crucial  because  they  will:  
 
• guide  your  literature  search;  
• guide  your  decisions  about  the  kind  of  research  design  
to  employ;  
• guide  your  decisions  about  what  data  to  collect  and  
from  whom;      
• Guide  your  analysis  of  your  data;  
• guide  your  wri3ng-­‐up  of  your  data;  
• stop  you  from  going  off  in  unnecessary  direc3ons;  
• provide  your  readers  with  a  clearer  sense  of  what  your  
research  is  about.  
We  need  to  know:    
• what  is  already  known  about  the  topic;  
• what  concepts  and  theories  have  been  applied  to  
the  topic;  
• what  research  methods  have  been  applied  to  the  
topic;  
• what  controversies  about  the  topic  and  how  it  is  
studied  exist;  
• what  clashes  of  evidence  (if  any)  exist;  
• who  the  key  contributors  to  research  on  the  topic  
are.    
• Concepts  are  the  way  that  the  sense  of  the  
social  world  is  made.    
• Concepts  are  a  key  ingredient  of  theories.  
Indeed,  it  is  almost  impossible  to  imagine  a  
theory  that  did  not  have  at  least  one  concept  
embedded  in  it.    
• Literature  review  is  a  cri3cal  examina3on  of  
exis3ng  research  rela3ng  to  the  phenomena  
of  interest  and  of  relevant  theore3cal  ideas.  
 The  need  for  theore3cal  framework  
• A  theore3cal  framework  represents  your  
beliefs  on  how  certain  phenomena  (or  
variables  or  concepts)  are  related  to  each  
other  (a  model)  and  explana3on  of  why  you  
believe  that  these  variables  are  associated  
with  each  other  (a  theory).  
• Both  the  model  and  the  theory  flow  logically  
from  the  documenta3on  of  previous  research  
in  the  problem  area.  Integra3ng  your  logical  
beliefs  with  published  research,  taking  into  
considera3on  the  boundaries  and  the  
constraints  governing  the  situa3on,  is  pivotal  
in  developing  a  scien3fic  basis  for  
inves3ga3ng  the  research  problem.          
 The  process  of  building  theore3cal  
framework  includes:  
• Introducing  defini3ons  of  the  concepts  or  
variables  in  your  model.  
• Developing  a  conceptual  model  that  provides  
a  descrip3ve  representa3on  of  your  theory.  
• Coming  up  with  theory  that  provides  an  
explana3on  for  rela3onship  between  variables  
in  your  model.  
 Theory  and  research  

• Characterizing  the  nature  of  the  link  between  

theory  and  research  is  by  no  means  a  straight  
forward  maQer.  There  are  several  issues  at  
stake  here,  but  two  stand  out  in  par3cular.  
– First,  there  is  the  ques3on  of  what  form  of  theory  
one  is  talking  about.    
– Secondly,  there  is  the  maQer  of  whether  data  are  
collected  to  test  or  to  build  theories.    
• Theory  is  important  to  the  social  researcher  
because  it  provides  a  backcloth  and  ra3onale  
for  the  research  that  is  being  conducted.  It  
also  provides  a  framework  within  which  
organiza3onal  phenomena  can  be  understood  
and  the  research  findings  can  be  interpreted.  
 Sort  of  theories    
• Grand  theory  and  social  research  
• Labour  process  theory:  a  middle-­‐range  theory  
• Background  literature  as  theory:  emo3onal  
labour  
 Influences  on  the  conduct  of  
social  research  
 
Theory      Prac3cal  considera3ons  
Epistemology  
  Social  research  
 
Values                        Ontology  
 Criteria  in  social  research    
 
• Reliability  
• Replica3on  
• Validity  
• Rela3onship  with  research  strategy  
 Reliability  
 
• Reliability  is  concerned  with  the  ques3on  of  whether  
the  results  of  a  study  are  repeatable.  The  term  is  
commonly  used  in  rela3on  to  the  ques3on  of  whether  
the  measures  that  are  devised  for  concepts  in  the  
social  sciences  (such  leadership,  mo3va3on,  decision)  
are  consistent.    
• Reliability  is  par3cularly  at  issue  in  connec3on  with  
quan3ta3ve  research.  The  quan3ta3ve  researcher  is  
likely  to  be  concerned  with  the  ques3on  of  whether  a  
measure  is  stable  or  not.    
•  We  would  consider  it  an  unreliable  measure—we  
could  not  have  faith  in  its  consistency.  
 
 Replica3on  
 
• In  order  for  replica3on  to  take  place,  a  study  must  be  
capable  of  replica3on—it  must  be  replicable.  
• This  is  a  very  obvious  point:  if  a  researcher  does    not  
spell  out  his/er  procedures  in  great  detail,  replica3on  is  
impossible.    
• Similarly,  in  order  for  us  to  assess  the  reliability  of  a  
measure  of  a  concept,  the  procedures  that  cons3tute  
that  measure  must  be  replicable  by  someone  else.    
• An  inves3ga3on’s  capacity  to  be  replicated—
replicability—  is  highly  valued  by  many  social  
researchers  working  within  a  quan3ta3ve  research  
tradi3on.    
 Internal  validity  
• How  confident  can  we  be  that  the  
independent  variable  really  is  at  least  in  part  
responsible  for  the  varia3on  that  has  been  
iden3fied  in  the  depend  ent  variable?  
 External  validity.    
 
The  ques3on  of  whether  the  results  of  a  study  
can  be  generalized  beyond  the  specific  research  
context.    
In  this  context  that  the  issue  of  how  people  are  
selected  to  par3cipate  in  research  becomes  
crucial.  This  is  one  of  the  main  reasons  why  
quan3ta3ve  researchers  are  so  keen  to  generate  
representa3ve  samples.  
 Ecological  validity.  
• Do  our  instruments  capture  the  daily  life  condi3ons,  opinions,  
values,  a\tudes,  and  knowledge  base  of  those  we  study  as  
expressed  in  their  natural  habitat?    
• This  criterion  is  concerned  with  the  ques3on  of  whether  
social  research  some3mes  produces  findings  that  may  be  
technically  valid  but  have  liQle  to  do  with  what  happens  in  
people’s  everyday  lives.  
• To  answer  a  ques3onnaire  may  mean  that  the  findings  have  
limited  ecological  validity.    
Some  qualita3ve  researchers  propose  the  
alterna3ve  terms  and  ways  of  assessing  
qualita3ve  research  are  required  :  
«  trustworthiness  »  as  a  criterion  of  how  good  a  
qualita3ve  study  is.  Each  aspect  of  
trustworthiness  has  a  parallel  with  the  
quan3ta3ve  research  criteria.  
• Credibility,  which  parallels  internal  validity—
that  is,  how  believable  are  the  findings?  
• Transferability,  which  parallels  external  
validity—that  is,  do  the  findings  apply  to  other  
contexts?  Dependability,  which  parallels  
reliability—that  is,  are  the  findings  likely  to  
apply  at  other  3mes?  
• Confirmability,  which  parallels  objec3vity—
that  is,  has  the  inves3gator  allowed  his  or  her  
values  to  intrude  to  a  high  degree?  
 Variables  
• A  variables  is  anything  that  can  take  on  
differing  or  varying  values.  
• The  value  can  offer  at  various  3me  for  the  
same  object  or  person,  or  at  3me  for  different  
objects  or  persons.  
 Four  main  types  of  variables    
• The  dependent  variables  (also  known  as  the  
criterion  variable).  
• The  independent  variable  (also  known  as  the  
predictor  variable).  
• The  modera3ng  variable.  
• The  media3ng  variable.  
 Dependent  variable  
• It  is  the  variable  of  primary  interest  to  the  
researcher.  The  research’s  goal  is  understand  and  
describe  the  dependent  variable  ,  or  to  explain  its  
variability;  or  predict  it.  
•  it  is  the  main  variable  that  lends  itself  for  
inves3ga3on  as  viable  factor.    
• Through  the  analysis  of  the  dependent  variable,  it  
is  possible  to  find  in  quan3fying  and  measuring  
the  dependent  variable,  as  well  as  other  variables  
that  influence  this  variable.  
 Independent  variable  
• It  is  generally  conjectured  that  an  
independent  variable  is  one  that  influences  
the  dependent  variable  in  either  a  posi3ve  or  
nega3ve  way.  In  other  words,  the  variance  in  
the  dependent  variable  is  accounted  for  by  
the  independent  variable.  To  establish  that  a  
change  in  the  independent  variable  causes  a  
change  in  the  dependent  variable.  
 All  four  of  the  following  condi3on  
should  be  met:  
• The  independent  and  dependent  variable  
should  covary:  in  other  words,  a  change  in  the  
dependent  variable  should  be  associated  with  
a  change  in  the  independent  variable.  
• The  independent  variable  should  precede  the  
dependent  variable.  In  other  words,  there  
must  be  a  3me  sequence  in  which  the  two  
occur  the  cause  must  occur  before  the  effect.    
• No  other  factor  should  be  a  possible  cause  of  
the  change  in  the  dependent  variable.  Hence,  
the  research  should  control  for  the  effects  of  
other  variables.  
• A  logical  explana3on  (a  theory)  is  needed  and  
it  must  explain  why  the  independent  variable  
affects  the  dependent  variables.  
 Modera3ng  variable  
• The  modera3ng  variable  is  one  that  has  a  strong  
con3ngent  effect  on  the  independent  variable-­‐
dependent  variable  rela3onship.  That  the  presence  of  
a  third  variable  (the  modera3ng  variable)  modifies  the  
original  rela3onship  between  the  independent  and  the  
dependent  variables.    
• The  third  variable  has  a  modera3ng  effect  on  the  
independent  variable  and  dependent  variable  
rela3onship.  
• The  variable  that  moderates  the  rela3onship  is  known  
as  the  modera3ng  variable      
 Media3ng  variable  
• A  media3ng  variable  (or  intervening  variable)  is  
one  that  surfaces  between  the  3me  the  
independent  variables  start  opera3ng  to  
influence  the  dependent  variable  and  the  3me  
their  impact  is  felt  on  it.  
• There  is  thus  a  temporal  quality  or  3me  
dimension  to  be  media3ng.  In  other  words,  
bringing  a  media3ng  variable(s)  opera3ng  in  any  
situa3on,  and  helps  to  conceptualized  and  
explain  the  influence  of  the  independent  
variable(s)  on  the  dependent  variable.      
 How  can  we  develop  the  theore3cal  
framework  for  our  research?  
• The  TFW  is  the  founda3on  on  which  en3re  
deduc3ve  research  project  is  based.  It  is  logically  
developed,  described,  and  elaborated  network  of  
associa3ons  among  the  variables  deemed  
relevant  to  the  problem  situa3on  and  iden3fied  
through  such  process  as  interviews,  observa3ons,  
and  LR.    
• The  rela3onship  between  the  LR  and  TFW  is  that  
the  former  provides  a  solid  founda3on  for  
developing  the  laQer.        
• The  TWF  represents  and  elaborates  the  
rela3onships  among  variables,  explains  the  
theory  underlying  these  rela3ons,  and  
describes  the  nature  and  the  direc3on  of  the  
rela3onships.  The  TFW  provides  the  logical  
base  for  developing  testable  hypotheses.  
 The  components  of  TFW  
• There  are  three  basic  features  that  should  be  
incorporated  in  any  TFW:  
– The  variables  considered  relevant  to  the  study  
should  be  clearly  defined.  
– A  conceptual  model  that  describes  the  
rela3onship  between  variables  in  model  should  be  
given.  
– There  should  be  a  clear  explana3on  of  why  we  
expect  these  rela3onship  to  exist.  
 A  conceptual  model  
• A  conceptual  model  helps  you  to  structure  your  
discussion  of  the  LR.  A  conceptual  model  describes  you  
ideas  about  how  the  concepts  (variables)  in  your  
model  are  related  each  other.  A  schema3c  diagram  of  
the  conceptual  model  helps  the  reader  to  visualize  the  
theorized  rela3onships  between  variables  in  your  
models  and  than  obtain  a  quick  idea  about  how  you  
think  that  the  management  problem  can  be  solved.  
Both  a  schema3c  diagram  of  the  concept  model    and  a  
descrip3on  of  the  rela3onship  can  between  the  
variables  in  words  should  be  given,  so  that  the  reader    
can  see  and  easily  comprehend  the  theorized  
rela3onship.  
 Hypothesis  development    
• A  hypothesis  can  be  define  as  a  tenta3ve,  yet  testable,  
statement,  which  predicts  what  you  expect  to  find  in  
your  empirical  data.  
• Hypotheses  are  derived  from  theory  on  which  your  
conceptual  model  is  based  and  are  oben  rela3onal  in  
nature.  
• Hypotheses  can  be  defined  as  logically  conjectured  
rela3onship  between  two  or  more  variables  expressed  
in  the  form  of  testable  statement.  
• By  tes3ng  the  hypotheses  and  confirming  the  
conjectured  rela3onships,  it  is  expected  that  the  
solu3on  can  be  found  to  correct  the  problem  
encountered.  
 Statement  of  hypotheses  formats  
• If-­‐then  statement  
• Defined  as  a  testable  statement  of  rela3onship  
among  variables,  hypothesis  can  also  test  
whether  there  are  differences  between  two  
groups  with  respect  to  any  variables.  
• To  examine  whether  or  not  the  conjectured  
rela3onships  or  differences  exist,  these  
hypotheses  can  be  set  either  as  a  proposi3ons  or  
in  form  of  if-­‐statements.      
 Two  formats  can  be  seen  
• Employees  who  are  more  healthy  will  take  sick  
leave  less  frequently.  
 
• If  employees  are  more  healthy,  then  they  will  
take  sick  leave  less  frequently.    
Direc3onal  and  non-­‐direc3onal  
hypotheses    
 Null  and  alternate  hypotheses  
• A  null  hypothesis  (H0)  is  hypothesis  set  up  to  be  
rejected  in  order  to  support  an  laternate  
hypothesis,  labeled  HA.  
• When  used,  the  null  hypothesis  is  presumed  true  
un3l  sta3s3cal  evidence,  in  the  form  of  
hypothesis  test,  indicates  otherwise.  
•  typically,  the  null  statement  is  expressed  in  
terms  of  there  being  no  significant  rela3onship  
between  two  variables  or  no  (significant)  
difference  between  two  groups.  
 The  alternate  hypothesis  
• The  alternate  hypothesis,  is  the  opposite  of  
the  null,  is  a  statement  expressing  rela3onship  
between  two  variables  or  indica3ng  
differences    between  groups.  
The  steps  to  be  followed  in  hypothesis  
tes3ng  are:  
• State  the  null  and  the  alternate  hypotheses.  
• Choose  the  appropriate  sta3s3cal  test  
depending  on  whether  the  data  collected  are  
parametric  or  nonparametric.  
• Determine  the  level  of  significance  desired  
(0,05  or  more,  or  less).  
• See  if  the  output  results  from  computer  
analysis  indicate  that  the  significance  level  is  
met.  
 Hypothesis  tes3ng  with  qualita3ve  
research  
• Nega3ve  case  analysis:  
• Hypotheses  can  also  be  tested  with  qualita3ve  
data.  
• To  test  the  hypothesis