Enriqueta

 D.  Reston,  Ph.D.    

University  of  San  Carlos  
Cebu  City  
 
What comes to your
mind when you hear
the word statistics?
How about research?
¡  “The  world  of  statistics  starts  with  a  question,  
not  with  data.  Statistics  need  data.  Data  come  
from  good  measures.  Good  answers  to  
important  questions  require  good  data  from  
relevant  measures.  “  
                                                                               -­‐  Dana  Keller,  The  Tao  of  Statistics  
}  Research  is  Authentic  Inquiry  
◦  At  its  core,  research  is  inquiry,  the  attempt  to  answer  significant  
question  (Kilpatrick,  1992)  
}  Research  Advance  knowledge  in  the  Field    
 The  bottom  line  for  judging  research  is,  does  it  advance  the  
current  knowledge  in  the  field  in  a  significant  way?    
“A research question refers to a question that represents
the interests of the field, not the idiosyncratic setting to
be used for the study.  Observation of any group of
students, any teacher, or any classroom is not
necessarily of interest to the field. The question must be
justified based on what is known in the field and a
reasoned argument as to need for the intended
knowledge.”
¡  At the heart of Facts Worth
the discipline Familiarity

¡  Run across

grade levels
¡  Many times Concepts/
run across Strategies
Important to
curricular Know/Use
areas
¡  Overarching
understandings   “Big Ideas”
for  the  entire  field   Worth
of  Statistics   Understanding
1.    Know  where  the  data  come  
from:    
Ø  from  a  population  or  from  a  
sample?  
Ø  If  from  a  sample,  is  selection  
random  or  non-­‐random?  
 
 Target Population

Sample

s from a
• Sampling is the process of selecting the sample or the study unit
previously defined population.
 The observed sample
statistics are generalized to the
unknown population
Population of
unknown
parameters

Known
A sample is selected Sample
from a population with
unknown parameters.
2.  Know  the  type  of  data:    
Ø qualitative  or  quantitative?  
Ø Discrete  /  frequency  counts  or  
continuous  measures?  
Ø Level  of  measurement:  
nominal,  ordinal,  interval  or  
ratio  
Textual/  Narrative  
Tabular  Presentations  (tables)  
•  Frequency  Distribution  Tables,  Crosstabulations  

Graphical  Displays                                            (Area  graphs,  

bar  graphs,  line  graphs,  dot  plots,  etc.  
Summary  Statistics  /  
Numerical  Summary  Measures  
Types  of   Levels  of   Descriptive  Measures   Graphs  
Data   Measurement  

Qualitative   Nominal     Mode,  frequency  counts     Pie  chart  

&  percentages   Horizontal  bar  charts,  
component  bar  graphs  

  Ordinal   Median,  Measures   Vertical  Bar  charts  

  of  Position  
 
Quantitative   Interval   Means,  range,   Line  graphs,  box  
variances,  standard   plots,  scatter  plots  
deviations  
Ratio  
3.  Be  guided  by  
the  research   Theory  
problems:  
questions  to  
Empirical   Research  
be  answered   Generalizations   Problem  
Hypothesis  

or  objectives  
to  be  
attained.     Observations
/  Data  
 
 
¡ Statistical  –  pertaining  to  data  
¡  Data,  as  used  in  research,  means  research  
results  from  which  inferences  are  drawn:  
usually  numerical  results  but  may  be  also  
qualitative  or  categorical.    
¡  Analysis  -­‐  the  categorizing,  ordering,  
manipulating  and  summarizing  of  data  to  
obtain  answer  to  research  questions  
¡ To  reduce  data  to  
intelligible  and  
interpretable  forms  so  that  
relationships  of  research  
problems  can  be  studied  
and  tested.    
¡ Data  is  broken  down  into  
constituent  parts  to  obtain  
answers  to  research  
questions  and  to  test  
research  hypotheses  
 
4.  Always  consider  issues  of  
data  collection.    
ü  Constructs  to  be  investigated?  
ü  Variables/indicators  of  the  construct  
ü  Measurement  issues:  research  instrument  
validity  and  reliability  
ü  Sampling  issues  
•  Sample  representativeness  
•  Sample  size  adequacy  
•  Sampling  errors  
Sampling errors Chance
Non-sampling errors differences
a. Non-response from sample
to sample.
Follow up on
b. Coverage non-responses.

Excluded from
frame.

c. Measurement Bad Question!

5.  Examine  visual  displays  of  the  data.  Do  an  
exploratory    data  analysis  (EDA).  
 
  20

Minimum Infection Rate (MIR )

18
16
14
12
10
8 y = 0.258x + 3.133
6 R² = 0.03277

4
2
0
0 2 4 6 8 10 12 14 16 18
No. of Dengue Cases
6.  Statistics  lets  you  make  general  
conclusions  about  the  target  
population  from  limited  data.  
 
The  methods  of  inferential  statistics  allow  us  to  extrapolate  
from  limited  (sample)  data  to  make  a    conclusions  or  
generalization  about  the  population.  "Descriptive  
statistics"  simply  describes  data  without  reaching  any  
general  conclusions.  But  the  challenging  aspects  of  
inferential  statistics  are  all  about  reaching  general  
conclusions  from  limited  sample  data.  
7.  All  statistical  tests  are  based  on  
assumptions.  
Every  statistical  inference  tests  is  based  on  a  list  of  
assumptions.  An  assumption  behind  every  
statistical  calculation  is  that  the  data  were  
randomly  sampled,  or  at  least  representative  of,  a  
larger  population  of  values  that  could  have  been  
collected.  If  your  data  are  not  representative  of  a  
larger  set  of  data  you  could  have  collected  (but  
didn't),  then  statistical  inference  makes  no  sense.  
8.  Statistical  conclusions  are  always  
presented  in  terms  of  probability.  
¡  In presenting results of statistical tests in research
reports, support with the associated p-value or
significance probability value (denoted by Sig. in SPSS)
¡  A p-value specify the probability of a chance occurrence
of findings for a sample and provides a measure of
support from the empirical data against the null
hypothesis Ho.
¡  A small p-value indicates a sample result that is unusual
given the null hypothesis is true. Small p-value lead to
rejection of Ho whereas large p-values indicate that Ho
cannot be rejected.
¡  While
practical significance ask the
question “So what’s the use of these
findings?,” statistical significance
addresses the question, “Can we
generalize these findings from the sample
to the population, or are they just simply
due to chance or sampling error?”
The  general  format  for  presenting  an  inferential  statistic  
is:    
Statistic(df)  =  value,  probability  =  value.    
Example:    t  (29)  =  4.253,  p=  0.0025  
•  Note  that  exact  p-­‐values  are  preferred.  Also,  if  the  
computer  output  says  the  probability  is  .0000,  
then  report  it  as  .001.    
•  When  actually  presenting  the  results,  emphasize  the  
meaning  of  the  statistics.  That  is,  clearly  describe  
what  it  is  that  is  being  tested  and  what  significance  
means  for  the  variables  involved.    
9.  Recognize  that  insignificant  results  do  not  
necessarily  mean  that  no  effect  exists.  
•  This  simply  means  that  data  set  obtained  
have  not  provided  sufficient  evidence  to  
reject  the  null  hypothesis.    
If  a  difference  is  not  statistically  significant,  you  can  conclude  
that  the  observed  results  are  not  inconsistent  with  the  null  
hypothesis.  This  is  especially  a  problem  with  small  sample  
sizes.    It  makes  sense  to  define  a  result  as  being  statistically  
significant  or  not  statistically  significant  when  you  need  to  
make  a  decision  based  on  this  one  result.    
10.  Understand  how  statistical  inferences  are  
made  and  recognize  its  limitations  in  research.    
A  confidence  interval  estimate  quantifies  
precision  about  a  population  parameter  
based  on  sample  statistic.    
The  logic  of  hypothesis  testing  is  proof  by  
contradiction.  The  evidence  of  this  proof  
comes  from  sample  data.    
 
The Statistical Process
¡             Frequency  Analysis  
¡   Graphs  and  Graphing  
¡   Measures  of  Central    
           Tendency    and  Variability  
¡   Measures  of  Relations  
¡         Analysis  of  Differences      
¡ Analysis  of  Variance  &  Related  
Methods    
¡   Profile  Analysis  
¡   Multivariate  Analysis  
¡             Indices  
¡             Social  Indicators  
 “
“Statistical  thinking  will  one  
day  be  as  necessary  for  
efficient  citizenship  as  the  
ability  to  read  and  write.”  
-­‐H.G.  Wells