21' Writing Results Sections

1
21. Writing Results Sections

• Pelham Blanton Ch. 12
• Anderson, Linden Habra, 2005

2
Purpose

• The textbook chapter is a series of make-believe
  studies designed to give examples of how to
  report statistical analyses in scientific
  research articles.

3
Writing Results Sections

• Describe the statistics used and what they mean
• This depends on the stats you use (you will learn
  the appropriate tests in 218)
• e.g., These differences in reaction times were
  statistically significant (t(9) 3.78, p .004)
  
• Describe your findings
• e.g., The ADHD patients reaction times (M 1.4
  s, SD .3) were greater than those of controls
  (M 0.8 s, SD .5)
• Display your results visually
• graphs, figures, tables

4
Anderson, Linden Habra, 2005

• Examined the effect of hostility on response to,
  and recovery from, anger provocation
• Why?
• There seems to be a relationship between stress
  and disease development (particularly
  cardio-vascular disease)
• But, everybody experiences stressonly some
  develop illness
• Psychologists are interested personality
  characteristics that may account for the
  difference

5
TABP Hostility

• Type A Behaviour Pattern (TABP) first identified
  as increasing risk for CVD
• But not always Type Bs found to be more likely
  than Type As to die of second heart attack
• Refined analyses showed that hostility and anger
  are the toxic components in TABP

6
Proposed Mechanisms

• Hyperreactivity
• High hostiles show hyperreactivity to stress
• Hypertensives show same
• Normotensive children of hypertensives show same
• Strong model chronic hyperreactivity causes CVD
• Weak model hyperreactivity a non-causal marker
  of future disease development

7
Not the whole story . . .

• Stress response is normal and natural and healthy
• Prolonged departure from homeostasis is what may
  be harmful
• Should look at stress recovery, i.e., its not
  how high you go, its for how long that is
  important

8
Back to study

• 45 participants
• Complete the CMHQ, a standard hostility
  questionnaire
• Measure their SBP, DBP, HR over baseline, task
  and recovery (return-to-baseline) periods
• Between-within design
• Task serial 7s with harassment (nagging)

9
Hypotheses

• Highly hostile persons should (1) show
  hyperreactivity and (2) slowed recovery
• Bigger increases and stay high longer

10
Results
Describe results
State statistical test used and results
11
Where possible, show magnitude of the difference
observed
12
Figures
13
Complicated text description

• A better between-group comparison would be made
  by holding task values constant when looking at
  recovery data. We did this by entering the
  recovery data into a 2 (Time recovery 1,
  recovery 2)3 (Hostility low, medium, high)
  ANCOVA, co-varying out task values. This analysis
  revealed a significant main effect of hostility
  for SBP recovery ( F(2,41)5.87, p.006). The
  results are displayed in Fig. 3, with the high
  hostility group showing the highest SBP values
  and the low hostility group showing the lowest.
  Pair-wise comparisons (Tukeys HSD) revealed that
  the low hostility group had significantly lower
  SBP than the high hostility group ( p.001). The
  difference between the low hostility and medium
  hostility group approached significance ( p.07).
  Medium and high hostility groups were not
  significantly different. Analysis of the HR data
  showed no effect of hostility on HR recovery.
  Table 1 lists the adjusted recovery means for SBP
  and HR broken down by hostility.

14
Figures help
15
Tables too
16
Things to notice

• Describe what statistics used and why.
• Describe findings.
• Present findings in many different ways
• In text, figures, tables, etc.
• Give both p values and change score / effect size

17
Conclusion

• Measuring recovery is important