Welcome to PSYC 2321:

Analysis of Behavioural
Data

CHAPTER 7
INSTRUCTOR: NICOLE JENNI
Announcements
A NOTE ON EMAIL ETIQUETTE

Myself and the TAs all have a 24-48 hour response time
on e-mails
PLEASE don’t leave your work so last minute if you
anticipate needing help
Any content or assignment questions should be
addressed in OFFICE HOURS
Announcements
MIDTERM 1 AND ASSIGNMENT 1 GRADING
WILL TAKE ~ 2 WEEKS
We will go over these scores after reading break
Assignment 2 demo Thursday Feb 15
Assignment 2 DUE Thursday Feb 29
Ch 7 Achieve Activities due Sun Feb 25 (after reading
break)
We use the Normal Distribution in different ways

Precise “area under the

curve”

z score z statistic
Compare one sample’s
Compare one raw score to a
mean score to a distribution
distribution of other raw
of other possible sample
scores
mean scores

We worked on this last class

Using the Normal Distribution in Two Ways

Raw score z score Percentile

𝑋−𝜇 Normal
𝑧= Distribution
𝜎

Raw group Percentile for

z statistic
mean the Mean
𝑀 − 𝜇𝑀 Normal Distribution models
𝑧=
𝜎𝑀 Sampling Distribution of the Mean
What percentile is a z-score of -1.5???

When we can’t read

a percentile off a
chart….
We can look it up in
our

Z-TABLE!!!
z scores -3 -2 -1 0 1 2 3
 Introducing…

The z Table!
APPENDIX B-1 (4+ PAGES LONG)
ORIENTATION
Appendix B-1 (4+ pages long)
➢ Lists every potential
value of z

➢ Gives you the

corresponding area
under the curve

➢These two
columns will
always sum to
50%
Q2 What is the percentile for the Analytical
Writing raw score of 5.5 (z = 2.30)?

A. 1.07%
B. 48.93%
C. 51.07%
D. 98.93%

Need to add 50% to 48.93

In words, what does this result mean?

Q3. Which percentile will fall below 50%?
Why?
B. D.
A. Verbal C. Analytical
Quantitative Psychology
Reasoning Writing
Reasoning Subject
Range: 130-170 Range: 130-170 Range: 0-6 Range: 200-990

μ = 150.24 μ = 153.07 μ = 3.55 μ = 618

σ = 8.44 σ = 9.24 σ = 0.86 σ = 106

e.g., score of 160 e.g., score of 167 e.g., score of 5.5 e.g., score of 600

z = 1.16 z = 1.51 z = 2.30 z = -0.17

Percentile = Percentile = Percentile = 98.93 Percentile =

 Q4. What is the percentile for the
Psychology subject score of 600?
➢ Note: your z-table only presents
Psychology A. 93.25 positive values for z, but because
the normal curve is symmetrical we
Subject B. 56.75
can still figure this out!
Range: 200-990 C. 43.25 ➢ Protip: ALWAYS draw it out
μ = 618
D. 6.75
σ = 106
e.g., score of 600
.
z = -0.17 This person scored better than
43.25% of people who took this test in
Percentile =
the last five years
Q5. More difficult problem: What percent of
scores fall between z values of ±1.96?
➢ Step 1: DRAW IT OUT
A. 2.50% ➢ Step 2: decide what we want to look for
B. 5.00% ➢ the table ONLY converts positive z’s
➢ We need to decide if we want middle or
C. 47.50%
tail section
D. 95.00% ➢ Step 3: Now go find this number in the table

x2

-1.96 1.96
Q6. More difficult problem: What percent of
scores are more extreme than z=1.64?
A. 2.50% ➢ Step 1: DRAW IT OUT
➢ Step 2: note more extreme means in BOTH
B. 5.00% direction
C. 10% ➢ Step 3: Now go find this number in the table
D. 90%

-1.64 1.64
x2
We can also convert percentages back to
z Scores
From percentages to z scores
◦ Step 1: Use the z table in reverse, taking a
percentage and converting it into a z score.

◦ Step 2: Convert the z score to a raw score

using the formula.
 We can also convert percentages back to
z Scores
Psychology Q7. You need a psychology score in the 90th percentile
to apply for graduate school. What do you need to
Subject score on the Psychology subject test?
Range: 200-990
A. 1.28
μ = 618
B. 644
σ = 106
C. 754
Score:??? D. 0.25
z =???
Z scores: -3 -2 -1 0 1 2 3
Percentile = 90 Raw scores: 300 406 512 618 724 830 936
Percentile: 0 50 100
We can use this same
distribution to judge
whether a group mean
comes from a population?
WHY?

…Because of the Central Limit Theorem.

Central Limit Theorem

A distribution of sample means is a

normal distribution, even when the
population distribution is not normal
(as long as sample size >30)

◦ A distribution of means is less variable (smaller SD) than a

distribution of individual scores.
 Refresher: Sampling distribution of the mean

1. It’s a frequency distribution of means

◦ Those means are calculated by drawing an infinite number of n
sized samples, calculating the mean for each sample, and plotting it

2. Its mean, denoted 𝜇𝑋ത or 𝜇𝑀 , equals the mean of the

population (𝜇)
3. Its standard deviation equals the standard deviation of the
raw score population, divided by square root of the sample
size 𝜎
◦ Denoted 𝜎𝑋ത or 𝜎𝑀 , called the Standard Error of the Mean 𝜎𝑀 =
𝑁
4. It is normally shaped (depending on sample size and shape
of the population’s distribution)
Implications of the Central Limit
Theorem
As long as N ~≥ 25 or 30, sampling distribution of the mean is
normally shaped
Therefore, can use the Normal Distribution to approximate the
Sampling Distribution of the Mean
◦ Helps us find out probabilities of getting various sample means under the
null hypothesis distribution
 Practice: you are interested in measuring trait impulsivity in
high school students, as you wonder if this age cohort
resembles the average population
You are using a well validated scale to measure impulsivity. It is scored out of 100,
and in the population has a mean of 50 and a standard deviation of 16.
You randomly sample 64 high school students, and assess their impulsivity. You
record a sample mean of 55
1) What is the standard error of the mean (SEM?) for this sampling distribution

𝜎 16
𝜎𝑀 = 𝜎𝑀 = 𝜎𝑀 = 2
𝑁 64

Z -2 -1 0 1 2
X 46 48 50 52 54
 Q9. Practice: you are interested in measuring trait impulsivity
in high school students, to ultimately see if it can help predict
substance use
You are using a well validated scale to measure impulsivity. It is scored out of 100,
and in the population has a mean of 50 and a standard deviation of 16.
You randomly sample 64 high school students, and assess their impulsivity. You
record a sample mean of 55
2) What is the probability of obtaining a sample mean of 55 OR
higher?
Two Steps:
1. What’s the high school z statistic relative to
the whole population?
𝑀 − 𝜇𝑀
2. Look up the area under the curve 𝑧=
𝜎𝑀 Z -2 -1 0 1 2
X 46 48 50 52 54
 Q9. Practice: you are interested in measuring trait impulsivity
in high school students, to ultimately see if it can help predict
substance use
2) What is the probability of obtaining a sample mean of 55 OR
higher?

Two Steps:
1. What’s the high school z statistic relative to
the whole population?
2. Look up the area under the curve

Z= 55-50/2 = 2.5
Area under the curve = 0.6%
Z= 2.5
Did this sample (high school students) come from
the same population as the ‘general population?

Unlikely…. It maybe came from a different population

How rare is a sample mean this size, if it had been randomly

drawn from the population where 𝜇 = 50, and 𝜎 = 16?

If high school students really did reflect the same impulsivity as

the general population, we had a 0.6% chance of sampling a
mean of 55 or higher
We can use the sampling distribution of
the mean, to conduct a hypothesis test!

INTRODUCING HYPOTHESIS
TESTING!!!
What are the 6 steps to
Null Hypothesis
Significance Testing?
Six Steps to Hypothesis Testing
IN GENERAL
What are we
Determine the
comparing?
characteristics Determine the Make a
Choose the State the null
of the Null critical values Calculate the decision
correct and research
Hypothesis based on the test statistic. about the null
hypothesis hypotheses.
Sampling alpha level. hypothesis.
test and check
Distribution.
assumptions.

Comparing a Use
sample with a (Typically
mean on a scale population If α = .05, zobtained =
variable to a two-tailed) two-tailed Retain or
population where parameters: 𝑀 − 𝜇𝑀
the mean and H 0 : µ2 = µ 1 𝜇 and 𝜎 and zcritical = Reject
standard deviation
sample size 𝜎𝑀
are known. H 1 : µ2 ≠ µ 1 ±1.96
[Assumptions] (for 𝜎𝑀 )

APPLIED TO Z TEST
Research Scenario: you have a suspicion that people
who are very impulsive are more likely to engage in Summary
criminal activity, and you would like to put this to the 𝜇 = 50
test. You plan to measure trait impulsivity in 64 randomly σ = 16
M= 55
sampled prisoners from jails across the country N= 64

Reminder: Our impulsivity scale has a mean of 50 and a

standard deviation of 16, our sample of prisoners has a
mean of 55.
Step 1:Choose the correct hypothesis test and
check assumptions

The z-test is appropriate when we know

descriptive information about the distribution
of raw scores in the population

We must know two things:

1. 𝜇
2. σ
Q1. Step 2: State the null and research hypotheses.
What is our null hypothesis?
A. This sample of prisoners came from the same population as
the rest of Canada.
B. This sample of prisoners came from a different, more
impulsive population.
C. This sample of prisoners came from a different population
that is either higher or lower than the existing population.
Q2. If the null hypothesis is true, and prisoners
are the same in impulsivity, what else is true?
A. A random sample from of prisoners could have any mean
impulsivity score within the range of population raw
scores.
B. A random sample of prisoners probably has a mean
impulsivity close to 50
C. If we reject it, we could be making a Type I error.
D. All of the above.
 Step 2: Every null hypothesis has a
corresponding research hypothesis.
Q3. What is the two-tailed research hypothesis that corresponds
to that null hypothesis?

A. This sample of prisoners came from a different, more impulsive

population.
B. This sample of prisoners came from a different population that is
either more or less impulsive than the existing population.
Step 2: State the null and research
hypotheses.
• This sample came from the
Null same population.
Hypothesis • H0: µ2 = µ1 -OR - H0: µpris = 50

Two-Tailed • This sample came from a different

population with a mean impulsivity that is
Research either higher or lower than the original
population.
Hypothesis • H1: µ2 ≠ µ1 -OR- H1: µpris ≠ 50
Comparing one-tailed versus two-tailed hypotheses

You have to decide before looking at your data!

Usually default to two-tailed.

• This sample came • This sample came

Null from the same Null from the same or less
Hypothesis population. Hypothesis impulsive population.
• H0: µall = µpris • H0: µall ≥ µpris

• This sample came from a

different population with a • This sample came
Two-Tailed One-tailed from a different, more
mean height that is either
Research more or less impulsisve than Research impulsive population.
Hypothesis the original population. Hypothesis
• H1: µall ≠ µpris
• H1: µall < µpris
 Step 3: Determine the characteristics of the Null Summary
𝜇 = 50
Hypothesis Sampling Distribution. σ = 16
M= 55
We use the Normal Distribution to model the N= 64
Null Hypothesis Sampling Distribution of the Mean

𝜎 16
𝜎𝑀 = 𝜎𝑀 =
𝑁 64
𝜎𝑀 = 2

𝑧 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 𝑠𝑐𝑜𝑟𝑒𝑠 −2 −1 0 1 2

𝑟𝑎𝑤 𝑔𝑟𝑜𝑢𝑝 𝑚𝑒𝑎𝑛 𝑠𝑐𝑜𝑟𝑒𝑠 46 48 50 52 54

Step 4: Determine the critical values based
on the alpha level
The alpha (α) level determines our critical cut-off.

• Alpha is a value that the researcher can set PRIOR

to running their experiment,
• Alpha corresponds to our experiments type 1 error
rate
• Almost always in psychology: α = .05, two tailed
 Step 4: Determine the critical values based
on the alpha level
“Critical Values” exclude alpha (α) % of the curve.

Two tailed alpha = 5%

α2 = .05

Need to divide our alpha

into the two tails of the
distribution 2.5%

𝑧 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 𝑠𝑐𝑜𝑟𝑒𝑠 −2 −1 0 1 2

𝑟𝑎𝑤 𝑔𝑟𝑜𝑢𝑝 𝑚𝑒𝑎𝑛 𝑠𝑐𝑜𝑟𝑒𝑠 46 48 50 52 54

𝜇
𝜇𝑀
Step 4: Determine the critical values based
on the alpha level
Need to look up our critical z from our z-table

𝑧𝑐𝑟𝑖𝑡 = 1.96
Step 4: Determine the critical values
based on the alpha level
“Critical Values” exclude alpha (α) % of the curve.
-1.96 1.96

Region of Rejection Region of Rejection

𝑧 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 𝑠𝑐𝑜𝑟𝑒𝑠 −2 −1 0 1 2

𝑟𝑎𝑤 𝑔𝑟𝑜𝑢𝑝 𝑚𝑒𝑎𝑛 𝑠𝑐𝑜𝑟𝑒𝑠 46 48 50 52 54

𝜇
𝜇𝑀
 Summary
𝜇 = 50
Step 5: Calculate the test statistic σ = 16
M= 55
N= 64
Here we want to compute z-statistic 𝜎𝑀 = 2

-1.96 1.96
𝑀 − 𝜇𝑀
𝑧=
𝜎𝑀
Region of Rejection Region of Rejection
55 − 50
𝑧=
2
𝑧 = 2.5
𝑧 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 𝑠𝑐𝑜𝑟𝑒𝑠 −2 −1 0 1 2

𝑟𝑎𝑤 𝑔𝑟𝑜𝑢𝑝 𝑚𝑒𝑎𝑛 𝑠𝑐𝑜𝑟𝑒𝑠 46 48 50 52 54

 Summary
Step 6: Make a decision about the null 𝜇 = 50
σ = 16
hypothesis M= 55
N= 64
𝜎𝑀 = 2
Two ways of making our decision

Comparing Zstat to Zcrit Comparing p value to our alpha

• Zstat > Zcrit we can reject 𝐻0
• If Zstat < Zcrit we can not reject 𝐻0
 Summary
Step 6: Make a decision about the null 𝜇 = 50
σ = 16
hypothesis M= 55
N= 64
𝜎𝑀 = 2
Two ways of making our decision

Comparing Zstat to Zcrit Comparing p value to our alpha

• Zstat > Zcrit we can reject 𝐻0 • P-value < alpha we can reject 𝐻0
• If Zstat < Zcrit we can not reject • P-value > alpha we can not reject
𝐻0 𝐻0

More conventional to report p-

values in published work, p<.05
means a significant effect!
Statistical Significance
If we reject our null hypothesis, our results are statistically
significant

What does “statistically significant” mean?

◦ A finding is statistically significant if the data differ from what would be
expected by chance if there were, in fact, no actual difference.

◦ Statistically significant does not necessarily mean that the finding is important
or meaningful.
◦ If we do not reject our null hypothesis our sample is ‘not significantly’ different than the population
(our data are never ‘insignificant’ or unimportant