PSYC 5110

Statistics and Research Design in

Psychological Science

Dr. Kitty Y. F. Fung

Why do Psychologists Conduct
Experiments?
Psychology is the scientific study of behavior and
mind

a systematic way to use evidences to examine

testable theories
~ Should I expose my infant to music?
~ Is this drug effective for depression?
~ Do people make rational decisions?

even if you don’t want to be a psychologist,

scientific method is still important for you to think
critically and rationally
 Different Types of Questions
Value judgment
e.g., Is the place beautiful?

Logical questions
e.g., If a>b, and b>c, then is a>c?

Empirical questions
e.g., Does the sun rotate around earth?
Is red wine good for the heart?

This 3rd category is our main focus

Different Ways of Knowing
Relying on authority

Use of reason

Experience

Scientific method
 Relying on Authority
accept the validity of information from a source we
judge to be expert

often a good idea

e.g., it is the earth that rotates around the sun,
not the sun that rotates around the earth

problems
~ authorities can be wrong
~ when you are the expert
 Use of Reason
Intuitive reasoning of human mind is often flawed

We have difficulties with:

~ estimating magnitudes
~ calculating probabilities
~ considering base rates
~ many other things
 Estimating Magnitudes
the world population is approaching 7 billion, is that a
lot?

all living humans, placed end to end, would extend to

the moon and back about a dozen times

the weight of a small mountain would easily exceed

the weight of all living humans
 Calculating Probabilities
Suppose you’re on a game show, and you’re given a choice
of three doors. Behind one door is a car; behind the others
are sheep

you pick a door – Door 1

the host, who knows what’s behind the doors, opens
Door 3. And then he asks you,
“Do you want to switch to Door 2?”

Should you switch?

(a) Yes (b) No (c) Does not matter
 Considering Base Rates
Only 1 in 10,000 university students have brain tumor. A
test for the tumor is fairly accurate. If you have a tumor,
the test comes up positive 99% of the time. If you don’t
have it, the test gives a false positive only 2% of the time.
If a test comes out positive, what’s the chance that you
actually have a tumor?

A) >99%
B) 50-99%
C) 10-50%
D) 1-10%
E) <1%
 Experience
The conclusions drawn from experience could also
suffer the flaws of human intuitive reasoning

A 6
Vowel Even number
Which cards must you see to verify this rule? If a card has a vowel on
left side, it must have an even number on right side

E 2 B 4 S 5 A 7
 Confirmation Bias
~ a tendency to search out and pay special attention to
information that supports one’s belief, while ignoring
information that contradicts a belief

has an especially large impact when dealing with

ambiguous descriptions and decisions

it may indeed be an important reason why fortune-

telling and astrology are appealing to many people.
After hearing a description, we tend to search for facts
that confirms this description
 The Power of Coincidence
If you take any two people at random, and spend
enough time, you could find just as many odd
coincidences

What makes more sense?

~ there’s a mystic connection between two dead
presidents …?
~ or that these are merely random co-occurrences…?

The power of coincidence comes about because of

our failure to consider the probability that something
happens by chance
 number of times something did happen

hits
Probability =
hits + misses

number of times something could have

happened, but didn’t

we remember the “hits”

but we often forget all the “misses”
we need both to calculate a probability
 Science as a Way of Knowing
Science assumes determinism
~ statistical determinism in Psychology
Science makes systematic observations
~ pseudoscience often rely on occasional observations
Science produces objective knowledge
Science is data-based
Science produces tentative conclusions
Science asks answerable questions
Science develops falsifiable theories
The Goals of Research in Psychology
Description
~ identify regularly occurring sequences of psychological
events (e.g., behaviors, thoughts, emotions, etc.)
Prediction
~ psychological events follow certain ‘laws’ that are regular
and therefore predictable

Explanation
~ psychological events are explained in terms of their
relationship to other factors; causal explanations are ideal

Application
~ science informs real-world applications of psychological
events
 Theory
Deduction Induction

Experiments
Hypotheses/
Observations
Predictions
Observation events that happen in the world
Hypothesis/ specific predictions for specific
prediction situations generated from a theory
Theory interrelated set of concepts used to
explain a body of data and make
predictions
Generating Good Research Questions
Surprising
~ could the results be surprising / counterintuitive, and
challenge well-established beliefs?

Theoretically interesting
~ will the question have important implications on lots of
other important issues?

Practically applicable
~ can it be applied to practical settings?
Generating Good Research Questions
Research from observations
~ recognize odd events that deviate from the general
trend

Research from theory

~ apply the theory in a novel way to find unique and
counterintuitive predictions
Research from theory (example)
Aristotle's theory of gravity has been around for
more than 1000 years

but Galileo made a unique prediction from this

theory which leads to a breakthrough
Theory: heavier objects fall at a
faster speed

Predictions:
If two objects are locked together as a
whole, would the new object’s falling
speed be

faster than the heavy one

because it is even heavier?

between the two because they

balance each other out?
 Balance
Very
each
heavy
other
Generating Good Research Questions
Research from other research
~ have a good scholarship about the situation of
the field
~ literature search and review
~ PsycINFO
~ Google scholar
~ Science Citation Index at Web of Science
Web of Science Starting with a keyword
To choose a paper. It is helpful to consider:
Times cited
You can actually sort the papers by that
The journal (Impact factor)
To look for relevant paper. It is helpful to consider:
Papers that have cited this paper
Papers that have been cited by this paper
Papers that share citations
 Treeing Forward
look for papers that have cited the known papers, this
allows you to find the most recent progress in a field

Seitz et al (2009)

Tong et al (2006) Tononi et al (2008)

Tsuchiya & Koch (2005)

 Treeing Backward
look for papers that have been cited by the known papers,
this allows you to find the earlier classic studies in a field

Tsuchiya & Koch (2005)

Blake et al (2002) Lumer et al (1998)

He et al (1996)
a combination of these 2 strategies can organize these
papers as a network
you can start with one paper and find all the relevant
papers for a field
Main Sections of a Research Paper
Title Page
Abstract
Introduction
~ statement of the problem
~ review of related literature
~ statement of the hypothesis
Method
~ participants
~ instrument
~ design
~ procedures
Results
~ present the results in the order of the hypotheses
Discussions
References
Tables and Figures
 Elements of an Experiment
Operational definitions
~ define theoretical constructs in terms of operations
entailed in manipulating or measuring the constructs
~ link the concept to an observable event

Independent variable

Dependent variable

Does variation in the IV cause variation in the DV?

 Why statistics is important in
Psychology?
Statistical determinism
~ a formal and more reliable way to make decisions in
scientific research

Question: one of your clients, Mary, has been depressed

for a few days. But how do you know whether
this is some systematic disorder, or just fluctuations
of a normal person?

Rule: based on day-to-day records, try to figure out if Mary’s

level of depressions is significantly higher than normal
people (baseline)
 To see whether the dots are systematically above the
black line (baseline). Do you have a decision yet?
Level of depression
 The right point
Too early Too late
Days
Level of depression 6 12 20 30 60
What we learned from this question?
• Statistics is not a weird mathematical game

• It helps you to decide whether an observation is reliable

or not

• You will often encounter these kind of situations in all

areas of psychology as well as in daily life
 Basic concepts
Variables: characteristics that can take on any number
of different values
Values: possible numbers or categories that of a
variable can have

Scores: a particular member’s value on a variable

Variables Values Scores
Age 1, 4.5, 30,…… Mike is 30 years old

Gender Male, Female Mike is male

Ranking 1st, 2nd, 8th,…. Mike ranks 3rd in his class

 Different Types of Variables

Nominal/Categorical
Rank-order
Equal-Interval
 Nominal/Categorical

• Values are categories

e.g., Gender, Nationality

• No arithmetic, no comparison
(male and female are different, no calculation can
be made)
 Rank-order

• Values are the “rankings”

e.g., ranking 1st, 3rd, 15th

• No arithmetic, but comparison allowed

(1st > 4th, but the difference between 1st and 2nd
is not necessarily equal to the difference between
3rd and 4th)
 Equal-Interval

• Values on “even” scales

e.g., Height, Weight, Age

• Arithmetic and comparison both allowed

(e.g., 1.5 meter – 1 meter = 0.7 meter – 0.2 meter)
 Comparison Calculation
(Arithmetic)
Nominal/ No No
Categorical
Rank-order Yes No
Equal-Interval Yes Yes
 Discrete Variables take only values in discrete
sets, such as the integers. For example, a score
on a 100-points system could be 65, 66, but not
65.34

Equal-Interval

Continuous Variables do not have a minimum

unit. For example, your weight could be 123.48
Question: What are these variables?
Nominal
Brand of computers (categorical)

Number of pets Rank-order

Size of home in m2 Equal-interval

(discrete)

Ranking in this class Equal-interval

(continuous)
Two Branches of Statistical Methods
Descriptive statistics
~ summarizing and describing a group of numbers
such as the results of a research study
(e.g., Mean, Standard deviation)

Inferential statistics
~ drawing conclusions and inferences that are
based on the numbers from a research study,
but go beyond these numbers
 Distribution
a basic way of summarizing and describing a
group of numbers

How would you describe the distribution of ages of

this group of people?

Tom 17 Mike 18 Annie 19

John 18 Jack 18 James 20
Patricia 18 Linda 16 Robert 20
Barbara 18 Elizabeth 19 Jennifer 20
 Frequency Table
list the number of people for each age

Age Number of people for this age

16 1
17 1
18 5
19 2
20 3
Frequency Polygon (i.e., line graph)
a typical way to show a distribution

6
Frequency

0
16 17 18 19 20
Age
 Histogram
another typical way to show a distribution
Sometimes a distribution will become too complex.
You can see a trend, but there are too many levels
to allow a clear presentation

40

35

30

25

20

15

10

0
1 11 21 31 41 51 61 71
Solution: Merge them into a small number of levels

40
30
20
10
0
1 11 21 31 41 51 61 71

200
150
100
50
0
1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80

This is a grouped frequency table (graph)

 Shapes of Distributions
Unimodal: (1 peak)

Bimodal or Multimodal:
(2 or more peaks)

Rectangular: (no peak)

Symmetrical vs. Skewed Distributions
Symmetrical distribution: approximately equal
number of observations on both side

Skewed distribution: one side is more spread out

than the other, like a tail

Left-skewed Right-skewed
 Ceiling and Floor Effects
Ceiling effect
~ it occurs when scores can go no higher than an
upper limit and “pile up” at the top

Left-skewed

Floor effect
~ it occurs when scores can go no lower than a
lower limit and “pile up” at the bottom

Right-skewed
 Matching the Graphs
Right-skewed
Distribution of
Life expectancy

Symmetrical
Distribution of IQ
(i.e., intelligence)

Left-skewed
Distribution of
household Income
 The Normal Curve

Particular characteristics
~ bell-shaped
~ unimodal
~ symmetrical
~ average tails

Normal curve is frequently used in both natural and

social science
 Kurtosis
~ the extent to which a distribution differs from a normal
curve
Heavy-tailed: more scores in the tails and more peak
(Large kurtosis) than the normal curve

Light-tailed: fewer scores in the tails and more flat

(Small kurtosis) than the normal curve
Misleading Graphs
Misleading Graphs
Misleading Graphs