Agenda

1
Agenda

• Preparing data
• Analysis of data
• Reporting research
• Graphs
• Presenting research

2
1. Data preparation

• Create database
• 8 characters - identifiable (dont just use Q1)
  e.g. numdksit or drnksit number of drinks /
  sitting
• Be sure to make a shortened but fully
  identifiable label, because that is what will
  appear on your graphs.
• Number your surveys!!!!
• And create variable subnum or casenum and list
  the survey on same line as rest of data from
  that survey.

3
Data preparation cont.

• Coding
• Reverse coding or other Data transformation
• Self-esteem example
• I generally feel good about myself strongly
  agree 7
• Sometimes I feel like Im not worth much as a
  person strongly agree 6
• transform 6 2 R

4
Data preparation cont.

• Create new variables
• 1. Keep the raw data column as is.
• 2. Make a formula to transform the data to a new
  variable
• e.g., of drinks at 1 sitting
• If male 5 drinks or over 1 (binge drinker)
• If male less than 5 0 (not binge drinker)

5
2. Use SPSS 4 descriptive statistics

• Find out about your sample
• How many? What characteristics?
• Find out how your sample as a whole answered
  every question
• Select interval/ ratio (scale) variables - get
  frequencies, means, standard deviations
• Select nominal and ordinal variables and get
  frequencies

6

• Recommendation Print out using 2 - 4 pages per
  sheet
• Check data for accuracy (9 on 7 pt scale)
• PREVENT CATASTROPHE
• Save your work frequently.
• Save your work on a hard drive and on a disk!
• (The program works faster from the hard drive.)

7

• You can print out all your data grouped by a
  characteristic (frequencies, means)
• e.g., you can see how males vs females answered
  all of your questions
• or binge drinkers vs. non-binge drinkers
• those that cohabit vs. those that do not
• (Ask for help on how to do this
• Be sure you have numbered your surveys and cases
  in the computer, because their order will change)

8
Use SPSS to determine Correlations

• Select Analyze, Correlate, Bivariate,
• Pearsons and
• Choose all your interval / ratio (scale) data
• Tape it together to get Correlation matrix
• Circle all correlations that are statistically
  significant
• Evaluate whether these correlations are weak,
  moderate, or strong

9
After completing a study of the effects of high
density (apartment) living versus single-family
living on the well being of families, you find
that your two samples have different measured
levels of well-being. Thus, you decide to do a
test of significance. Your test of significance
determines how likely it is that the observed
difference is due just to __________?
Chance
How would you determine which test of
significance to use?
10
Hypothesis testing/ Inferential statistics

• Use appropriate statistical test
• Non-parametric-- Ordinal and nominal data
• Parametric -- Interval or ratio data
• You can test whether the mean differences between
  two groups (males females) are by chance or
  statistically significant (interval / ratio data)

11
To compare means of 2 groups

• In SPSS Click Analyze
• Compare Means
• Independent Samples t-test
• Grouping Variable
• Highlight variable and click on arrow
• Define groups, e.g., males 1, females 2 --
  however you have your variable coded
• Highlight all variables you want tested (must be
  INTERVAL OR RATIO scale data), click arrow to
  test variable section
• Click OK

12
Independent Samples t-test
13
If compare 3 groups

• Use ANOVA
• (ask for help)

14
Marginal Significance

• If your p-value exceeds .05 but is close to it,
  you can treat it as marginally significant
• For your project, treat it as
• marginally significant or approaching
  significance if your p-value is between .05 and
  .1.
• Always list the significance level p ___

15
Crosstabs (for nominal and ordinal data)

• Analyze
• Descriptive statistics
• Crosstabs
• Highlight variables for row
• Highlight variable for column
• Click statistics, click chi-square or correlation
• Etc.

16
Research Report Structure

1. Abstract
2. Introduction
3. Method
4. Results
5. Discussion
6. Reference

17
Introduction

• Problem
• Literature review
• Rationale for study
• Research question
• Hypothesis

18
Introduction

• Discuss problems about this topic
• Who is affected? How are they affected?
• Adequate literature review?
• Accurate, representative, unbiased?
• Describe what is already known about this topic?
• Clarify how your study study fits in.
• Rationale - reasons its important to do this
  study. Why is it needed? What will your study
  contribute?

19
Introduction (cont.)

• Clear research question(s)
• Hypotheses that logically follow from the
  literature review
• A hypothesis is an educated guess based on the
  literature. If your hypothesis is inconsistent
  with literature, you need to justify it
• Hypotheses clearly stated and testable

20
Method

• Research Design
• Sample
• Variables
• Instrument / Measure
• Procedure

21
Method

• Describe research design
• Experimental/ Non-experimental /
  Quasi-experimental
• Quantitative / Qualitative
• Independent groups / repeated measures
• of observations / pre- post,
• random assignment to experimental / control /
  comparison groups

22
Sample

• Participants / Sample
• sample size
• demographics (age, gender, number, ethnicity)
• sampling technique
• Use descriptive statistics -- central tendency
  dispersion.
• Consider using a table for demographics

23
Variables

• Describe how you measured your variables
• (If experimental design, then dependent variables
  and independent variables)
• Describe research instruments/measures
• .Discuss reliability
• Discuss validity of your measures
• Attach survey to appendix

24
Procedure

• Clearly describe how study was carried out
• Data collection methods
• Describe how you collected data (when, how,
  where, with what, by whom)
• You can mention problems you encountered during
  data collection here

25

• Describe use of SPSS
• Statistical tests used
• Any reverse coding used

26
ResultsDescriptive statisticsHypothesis
testingAdditional findings
27
Descriptive Statistics

• Report descriptive statistics (measures of
  central tendency and measure of spread /
  variability)
• Consider using a table to present descriptive
  statistics and mention key numbers in text

28
Example A Table
29
Results

• Your major findings (the story),
• Other highlights of your results
• (what was significant, what was correlated)
• Data that tests your hypotheses
• .Summarize your data in graphs, charts, and/or
  tables.
• Include statistics and p levels along with the
  graph, chart, etc. that illustrates those
  results.

30
Hypothesis Testing

• Restate hypothesis state test(s) used
• State both descriptive statistics (, mean,
  correlation co-efficient) and
• inferential statistics (Chi-Square, t-value, and
  their p-value, etc)
• Clearly state what test results mean and whether
  or not hypothesis is supported
• Consider using charts to present results of
  hypothesis testing

31
Structure

• Use tables and charts but always mention key
  statistics in text
• Give consecutive numbers (e.g., Table 1, Table 2
  )
• and clear titles to tables

32
Chart 1. Tourist Expenditure (per day?)
Mainlander vs. Japanese
Chi-Square7.34, df2, plt.001
33
Chart 2. Mean Tourist Expenditure Mainlander
vs. Japanese
t2.30, df189, plt.001
34
Discussion
35
Discussion Summary

• Tie the statistical results to your research
  question and hypotheses
• Indicate if your results support or disconfirm
  each of your hypotheses.
• Go beyond numbers. Integrate results from
  different statistics and talk about patterns and
  tendencies in the data

36
Discussion

• What is the meaning of your data?
• Your interpretation of results
• Discuss alternative explanations for your results
• Separate your speculations from the conclusions
  supported directly from results

37
Discussion

• Compare your results with previous research,
  theory, / the findings in the literature reviewed
  in the introduction.
• Your results consistent with previous findings?

38
Discussion

• Discuss external validity of your study
• In what ways did you answer your research
  questions or not?

39
Limitations

• No research is perfect
• Acknowledge limitations of your research methods
• Any sampling bias?
• Any questions that might have been misunderstood
  by subjects?
• What you would do differently if you could do the
  project again?
• Strengths of your study

40
Implications

• How might your data benefit others / how could
  your findings be used or applied?
• Any suggestions for future research on this
  topic?
• New questions raised from the study?

41
This is an actual pie chart redrawn from a
recent Honolulu Advertiser story about racial
diversity on the UHM campus.

1. What does the pie data tell you?
2. Can you identify a problem with the data?
3. In order for the data to be really useful, I
   think it should be contrasted with other data.
   What data would you recommend?

42
Can you spot the error in the following graph?
How would you redraw it to more accurately
reflect computer knowledge among High School
students?
43
What is wrong with this chart?