Davids Key Points About Quantitative Research

1
Davids Key Points About Quantitative Research
2
Generalizability

• The goal of quantitative research to be able to
  make strongly supported claims that can be
  generalized to a large population
• One corollary of this is that the claims tend to
  be quite narrow
• All the procedures of quantitative research are
  aimed at this goal generalizability and strong
  support

3
Sampling

• A census is a study of a whole population (not
  necessarily the population of a country), and is
  an ideal (no generalization required!)
• A sample is intended to represent the whole
  population when a census is impossible
• It needs to be (a) large enough and (b)
  representative enough
• Simple random sampling is the ideal, but is
  modified for various reasons

4
Reliability

• Do all the bullets from the rifle hit the target
  in the same place?
• If I measure this again, or someone else
  measures it with this instrument, will the
  results be the same?
• Test/retest and alternate forms, as well as
  some statistical measures

5
Validity

• Do all the bullets from the rifle hit near the
  bulls-eye on the target?
• Does the instrument measure what I think it
  measures?
• Content, criterion and construct validity (p. 165
  of Creswell)

6
Experimental Studies

• Chapter 11 in Creswell
• What effect does this treatment (independent
  variable) have on that results (dependent
  variable)?
• Since its difficult to control for all other
  variables in education, try to submerge them in
  large sample sizes

7
Correlational Studies

• Chapter 12 in Creswell
• Do subjects with this characteristic also have
  this other characteristic?
• If I know this characteristic, can I predict
  this other characteristic?
• Key point correlation does not establish
  causality

8
Survey Studies

• Chapter 13 in Creswell
• Dr Maryanne Doherty discussed large-scale survey
  research using one of her studies as an example
• Political and other polling is another example
• A census is like a survey study but takes data
  from the whole population, rather than a sample

9
Analysis Software

• Dont underestimate MS Excel for data entry,
  organization and simple analyses
• SPSS, SAS/STAT and a variety of others
• Dont have to know how to calculate the stats,
  but should know what they mean!

10
History of Stats Tools

• p. 40 in Creswell, left hand side
• correlation
• regression
• factor analysis
• inferential statistics
• meta-analysis
• effect sizes

11
An example of a correlational study

• Perth high school physics, 1987-1997
• Just under 26,000 students
• Analysed value added difference between
  actual score and predicted score based on test of
  academic aptitude vs socioeconomic status
• Results are by school, by sector
• HFNC high fee non-Catholic
• LFNC low fee non-Catholic

12
(No Transcript)
13
(No Transcript)
14
(No Transcript)
15
(No Transcript)
16
Where to from here?

• If you think you might want to do a quantitative
  study
• Read Chapters 11-13 of Creswell
• Find a good stats-based research methods course,
  probably in the Ed Psych department, like EDPY
  500, EDPY 505 or EDPY 605