Research Methods in Social Relations

1
Research Methods in Social Relations

• Professor Mike Gallivan
• Georgia State University
• Atlanta, Georgia, USA
• Class 7 June 24, 2009

2
Overview of Class 7 today

• Review of old material
• Multiple regression analysis in SPSS
• Sivo, Saunders, Chang Jiang (2006)
• This article appeared in Journal of the AIS
• New material for today
• Students will present their evaluation of
  articles
• More discussion of Sivo, Saunders, Chang Jiang
  (2006)
• Writing the Results section of papers
• Writing the Discussion section of papers
• Getting papers published in academic journals

3
Here are the articles you can evaluate

• Bhattacharjee, S., Tung, Y.A. and Pathak, B.,
  Author experiences with the IS journal review
  process, Communications of the AIS, 13(37),
  2004, p. 629-53.
• Dalal, N., Singh, S. and Lanis, T. Research
  concerns of IS faculty an exploratory
  investigation, Journal of Computer IS, 39(3),
  2004, pp. 18-25.
• Koh, C.E., IS journal review process A survey
  on IS research practices and journal review
  issues, Infor- mation Management, 40(8), 2003,
  pp. 743-56.
• Gill, T.G., Whats an MIS paper worth? An
  exploratory analysis, Database for Advances in
  Information Systems, 32(2), 2001, pp. 14-33.
• Tanner, J., Totaro, M. and Hotard, D. Research
  productivity and teaching effectiveness MIS
  faculty, Journal of Computer IS, 39(4), 1999,
  pp. 8-15.

4
You have a chance to read and evaluate published
articles

• For each assigned article, please answer
• How was the survey conducted?
• Mailed (paper-and-pencil), email, posted on
  website
• Who was the target population of survey?
• How many respondents answered survey?
• What analysis methods did authors use to
• Conduct convergent and discriminant validity?
• Test their hypotheses or theory?
• What was the response rate?
• What efforts to contact or remind
  non-respondents?
• Was the response rate better or worse than the
  average article that Sivo, Saunders mentioned?
• Do you think the authors used a good methodology?
• What should authors have done differently?

5
Review Multiple Regression in SPSS

• This is for interval/ratio dependent variable
• you can only analyze 1 dep. variable at a time
• Always select options then pairwise deletion
• Indep. variables can be ordinal, interval, ratio
• you can analyze many indep. variables at once
• you can use a nominal variable if dichotomous
• do not use nominal variable with gt 2 categories
• unless you first create special dummy variables
• dummy variables can only have values 0 or
  1
• House_1 dummy variable can be 0 or 1
• it can be 1 for all respondents in House 1
• it must be 0 for all other people (House 2, 3,
  4, 5)
• Use the Transform Compute function to create

6
Review Multiple Regression in SPSS

• More details about multiple regression
• Entering predictor (independent variables)
• you can enter all predictors at the same time
• you can enter some predictors first, then others
• this is called blocks of predictor variables
• If you choose the blocks method, then
• select statistics and also R-squared change
• this will show how R2 increases for each block

7
Multiple Regression Results in SPSS
8
Problems with Response Rates

• Sivo, Saunders, Chang Jiang (2006)
• authors are all at University of Central Florida
• A good response rate is necessary for
• Internal validity to support your hypotheses
• External validity to be able to generalize
• Recommended response rates should be gt 50
• Most academic research articles, it is too low!

9
Problems with Response Rates

• Sivo, Saunders, Chang Jiang (2006)
• Problems with survey studies
• measurement error
• Due to imperfect questionnaires (bad construct
  validity)
• sampling error
• Due to inadequate sample size or nonrandom
  samples
• coverage error
• Due to inability to contact some people in the
  population
• We will focus most on the last problem
• Also called non-response error
• some people are systematically not represented in
  the sample, because they are more likely to not
  be among the survey respondents
• Very busy people are less likely to respond to a
  survey
• People without email address will not respond to
  email survey

10
More sources of Response Rates

• Other sources of non-response error
• People without email address will not respond to
  email survey
• People without telephones will not respond to a
  phone survey
• Research shows that men are less likely to answer
  than women
• Prior review of survey studies showed problems
• Paper published by Pinsonneault and Kraemer
  (1993)
• Identified the following problems in surveys
• Low response rates (I.e., lt 50)
• Unsystematic (not random) or inadequate sampling
• Single method designs

11
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• Writing the Results section of paper
• Setting the stage
• provide details about your n and response rate
• also explain efforts to contact non-respondents
• also compare respondents to non-respondents
• you can provide descriptive demographic
  information
• here is where you justify your survey method
• if this were a laboratory experiment study
• this is where you would show the treatment effect
• this is called a manipulation check in lab
  studies
• Describe your research analysis methods
• describe the methods in sequence you performed
  them
• explain your analysis methods before showing
  actual results
• please see examples in the two papers that I
  published

12
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• Present the findings in detail
• It is often a good idea to repeat each hypothesis
  before you present results
• Use tables, graphs, pictures where it may helps
  the reader to understand results
• A picture is worth 1,000 words!
• Do you have the same expression in China?
• Textbook shows graph for sex-biased job ads vs.
  sex-unbiased and sex-reversed job ads
• Presentation of specific statistical results
• textbook provides detailed recommendations

13
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• Discussion section
• Your opportunity to explain what is means
• Also why the results are useful / important
• First, say what you have learned in study
• Second, you can explain how your results
• Are similar to (same as) findings in prior
  research
• Are different from findings in prior research
• Extra post hoc analyses
• For any very surprising (counter-intuitive)
  results, you should try to provide explanation of
  why
• In our study, there was no relationship between
  individual computer self-efficacy and IT usage
  in fact, this relationship was very slightly
  negative (I.e., -0.06). We believe that the
  reason why our results differ from prior research
  studies is because our subjects were ...

14
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• Discussion section
• State the implications for various people
• academic researchers
• business or government managers
• policy makers, politicians, university
  administrators
• teachers or students or citizens
• Maybe some special group (people searching for a
  job)
• State new questions for future research
• We suggest the following questions for future
  study
• Also state the Limitations (weaknesses) of
  study
• Survey (or experiment) was only with students
• Survey was conducted within just 1 city or
  province
• Low response rate (e.g., lt30), or small n
• It is not possible to prove causality from survey
  data

15
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• Comments about writing style
• Each author develops their own writing style
• But its important to follow the standard of the
  specific journal(s) where you want to publish
• Some examples of writing style
• How long should a paragraph be? (e.g., 3-10
  sentences)
• Active voice (I analyzed the data using
  regression)
• Passive voice (The data were analyzed using )
• Some journals prefer active others prefer
  passive
• Some journals prefer you say we even if single
  author
• Combine very short sentences together using
  conjunctions
• Examples of conjunctions because, since, but,
  however
• Split very long sentences apart into separate
  sentences

16
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)

• More comments about writing style
• Textbook suggests that you work from an outline
• Some word processing software has an outline
  function
• Use examples in your writing - especially if you
  introduce a new construct that readers have never
  heard of before
• examples can help your readers to understand much
  better
• Use direct quotes from other researchers
• If you think another author said something in an
  interesting way that will be compelling or
  entertaining for readers
• Do not simply use direct quotes just because you
  are lazy
• Never plagiarize another authors words or ideas
• This is a substantial difference between Asia and
  USA
• Students in Asia are taught to emulate (copy) the
  experts
• Students in US are never permitted to copy
  another author unless they put into quotation
  marks and cite the source

17
Summary of Contents from Chapter 19 pp
463-476(Ch. 20 in green book, pp. 521-536)