Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Tutorial 8 exercise: Hypothesis testing – the t-test

Surname NYAWOSE First name ZEKHETHELO
Student number 221009412 Tutorial group

Instructions:
 Read the questions listed below, and then, with these questions in mind, read
“Tutorial 8” (pp. 143-164) from Tredoux and Durrheim (2019).
 Then, attempt to answer the questions below.
 You must upload a copy of this tutorial exercise in MS Word format to
moodle@ukzn by 14h00 on Thursday 11 May 2023.

Declaration: By submitting this tutorial exercise for evaluation, I declare that,

although I may have consulted with my tutor and/or other students in my class
about the work in this tutorial exercise, I typed my own responses into this
document, and that the final product is therefore my own work. I understand that if
I copy and paste the work of another student into my tutorial exercise, that this
constitutes cheating and academic misconduct on both my part and the other
student. I also understand that making use of unethically leaked model
answers from previous years’ tutors and tutorials constitutes cheating and
academic misconduct. Students who cheat on this tutorial exercise will be
referred for disciplinary action as per UKZN student rules.

Key points recap: Tutorials 6 and 7

From Tutorial 6 (the sampling distribution of the mean), we learnt that samples do
not represent populations exactly. Therefore, sample statistics are not the same as
population parameters, and samples only provide an estimate of populations. In
addition, the Central Limit Theorem (see pp. 115-116) tells us that if we draw two
samples from a population and calculate the means from these samples, we can
expect to get two slightly different scores.
From Tutorial 7 (hypothesis testing: the z-test), we learnt how to use the z-
�−�
equation (� = � ) to test hypotheses about groups (e.g., KZN people [group 1]
n
drive better than people from other provinces [group 2]; children who have strict
parents [group 1] are more disciplined than children who do not have strict parents
[group 2]). However, using the z-equation requires knowledge of the population
(i.e., � and �), and researchers seldom have this information. If we know the
population mean and standard deviation by the Central Limit Theorem, however,
we can determine the mean and standard error of the sampling distribution of the
mean. It is then possible to transform the sample mean, (�), into a z-score, and
determine the probability of observing such a mean if the null hypothesis were
true. Where we do not know the population parameters, we estimate these values
from samples, and do this via adapting the z-equation into a t-equation (i.e., the
purpose of Tutorial 8).
Section 1: t-test concepts

1
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

The t-test is used to determine whether the means of two samples are sufficiently
different to conclude that they are in fact draw from two distinct populations, or
whether the scores suggest that both samples come from a single population.
Therefore, the t-test is used to compare two estimated population means.
Imagine three possible situations (see Figures 1, 2, and 3 below) with two normal
distributions (Sample “A” and Sample “B”) – see below (the means of each normal
distribution are denoted by the dotted vertical lines).

Q 1. Which of the above figures is most likely to show two sample means (“A” and
“B”) that come from the same population? [1]
FIGURE 3

Q 2. Which of the above figures is most likely to show two sample means (“A” and
“B”) that come from two different populations? [1]
FIGURE A

Sample means are not exact representations of the population mean, but differ
slightly, on average, by an amount indicated by the standard error – the standard
deviation of the sampling distribution of the mean. The variation in sampling
means is the reason we cannot simply look at the sample means when we want to
see how population means differ.

The t-test allows us to determine the degree to which two distributions overlap by
taking into account the distance between the means of the distributions, and also
considering the width of the distributions (their variances).

�−�
The general form of the t-equation (� = � ) is very similar to the z-equation
n
�−�
(� = � ).
n

Q 3. What is different about the t-equation and z-equation? [1]

THER T EQAUTION USES THE SAMPLE STANDARD DEVIATION ,THE Z

2
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

EQAUTION USES THE POPULATION STANDARD DEVIATION

As implied above, the main difference between a t-test and a z-score

transformation is that the population parameters (� and �) do not have to be known
for the t-test. Therefore, the t-test can be used when we only have information
from our sample. With the z-formula, the population standard deviation (�) is used
to calculate a z-score, which in turn allows us to calculate a probability. When we
don’t have the population standard deviation (�) and/or mean (�), we have to use a
sample-based estimate (i.e., the standard error) in our t-test formula.

Section 2: t-Test assumptions

Q 4. In order to do a t-test, your data should meet certain assumptions. In the table
below, the four assumptions of t-tests are identified. Fill in the descriptions in
the three empty cells below. [3] (no description is needed for how to test for
independence)
t-test
How can you test the
assumption Description
assumption?
of . . .
ALL SAMPLES THA YOU ARE  Formal tests;
ANALYSIS HAVE BEEN  Consider the skewness and
DRAWN FROM kurtosis coefficients;
POPULATIONS THAT ARE  Draw a histogram and
normality NORMALLY DISTRIBUTED examine the shape of the
distribution (if bell shaped,
then the data is probably
normally distributed)
Each participant only provides
independence one score (with the exception
of a repeated measures).
No scores should be drastically USE BOXPLOTS AND
no significant
different from the rest SCATTER PLOTS TO TEST
outliers
THE ASSUMPTION
THE  Levene’s test of homogeneity
POPULATIONVARIATIONSOF of variance
TWO OR MORE SAMPLES  Check if variances differ by a
ARE CONSINDERED EQUAL factor of less than 4. If the
homogeneity of difference if less than 4, then
variance the variances are homogenous
enough. Basically you divide
the larger variance by the
�21
smaller variance (� = )
�22

3
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Section 3: Three sub-types of t-tests

There are three sub-types of t-tests. Examine the flowchart below which depicts
the steps you could follow to choose which t-test to use.

If you are comparing a population mean to a sample mean, then use a one-sample
t-test. However, if you are comparing two sample means, you first need to ask
yourself if the two means come from the same sample. If the two means come
from the same sample (e.g., you tested your sample twice on a measure), you
should then use a repeated-measures t-test to compare your sample means.
However, if your two means come from two different samples (e.g., men and
women), you then first need to check whether the variances of the two groups
differ by a factor of more than 4. In other words, you need to calculate if the
variance of one group is more than 4 times the variance of the other group. If the
two variances differ by more than a factor of 4, you should use an independent
samples t-test with separate variances. However, if the two variances differ by a
factor of less than 4, then you should use an independent samples t-test with
pooled variance estimates.

Use the following scenario to answer Qs 5-10

It is known that the average British man spends 100 minutes a day doing housework.
A group of British researchers want to find out if British women do more daily
housework than British men. These researchers sample 100 women at random from
the population and find that these women, on average, do 180 minutes of housework
a day with a standard deviation of 20 minutes. Tip: draw two normal distributions
(one for British men, and one for British women) (like in Figures 1, 2, and 3 above) to
help you visualise this scenario.

4
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Q 5. Choose and write out the correct null and alternative hypotheses. (3)
�0 : �1 ≠ �2 � 0 : �1 = �2
a b
�1 : �1 > �2 � 1 : �1 ≠ �2
c
�0 : �1 = �2 d
� 0 : �1 = �2
�1 : �1 > �2 � 1 : �1 < �2

Which is the correct D IS CORRECT

null and alternative
hypothesis (a, b, c, or
d)?
How would you write Null hypothesis: BRITISH WOMAN DO AS MUCH DAILY
the correct null and HOUSEWORK AS BRITISH MEN
alternative Alternative hypothesis: BRITISH WOMEN DO MORE
hypotheses in words? DAILY HOUSEWORK THAN BRITISH MEN

Q 6. Because the researchers do not know the population standard deviation for
how much daily housework British men do, they cannot determine a z-score
(zcalc), which they could then use to assess whether their sample mean for
British women falls within the rejection region or not (see Tutorial 7). The
researchers therefore need to use a t-test. Because the researchers are
comparing their sample mean of British women to a population mean (of
British men), a one-sample t-test is needed (see the flowchart above). As
indicated earlier in Tutorial 8, the formula for the general t-test (and the one-
�−�
sample t-test) is � = � . Using this formula, calculate t (tcalc). (3)
�
tcalc
�−�
�= �
�

180 − 100
�=
20
100

80
�=
2

t = 40

5
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Q 7. Calculating the critical value for t (tcrit), against which the researchers can then
compare their t-value (tcalc), depends on the degrees of freedom (df).
Calculate the degrees of freedom for this sample. [1]
df =N-1
=100-1
=99

Q 8. Are the researchers using a one-tailed or a two-tailed test and why? [2]
ONE TAILED TEST ,THE ALTERNATIVE HYPOTHESIS IS DIRECTIONAL ,KT
CLAIMS THATH WOMEN DO MORE DAILY HOUSEWORK THAN BRITISH MEN

Q 9. Using an alpha of 0.05, your degrees of freedom, and the t-table (Table A1.2
in Appendix 1 of your statistics textbook), determine the critical t-value (tcrit). [1]
tcrit = 1,6602

Q 10. We now need to decide whether to reject the null hypothesis or not. We do
this by comparing tcrit with tcalc to determine whether tcalc falls within the
rejection region. Use the table below to come to a decision whether to reject
the null hypothesis or not. [2]
Rewrite tcrit 1,6602
Rewrite tcalc 40
Does tcalc fall in the rejection region? YES
Delete the incorrect statement We reject the null hypothesis.
What conclusion can we make? BRITHISH WOEM DO MORE DAILY
HOUSEWORK THAN BRITISH MEN

Use the following scenario to answer Qs 11-19

A researcher wants to know whether there is a difference in maths results between
15-year-old boys and 15-year-old girls. The researcher collects a random sample of
maths scores from a variety of schools. These are the descriptive statistics of the
data she collects:
Descriptive statistic Boys Girls
� 54.34 60.25
�2 75.22 73
� 117 124

Q 11. Choose and write out the correct null and alternative hypotheses. (3)

a
�0 : �1 ≠ �2 b
� 0 : �1 = �2
�1 : �1 > �2 � 1 : �1 ≠ �2
�0 : �1 = �2 � 0 : �1 = �2
c d
�1 : �1 > �2 � 1 : �1 < �2

6
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Which is the correct B IS CORRECT

null and alternative
hypothesis (a, b, c, or
d)?
Null hypothesis: THERE IS NO DIFFERENCE IN MATHS
How would you write RESULTS BETWEEN 15 YEAR OLD BOYS AND 15
the correct null and YEAR OLD GIRLS
alternative Alternative hypothesis: THERE IS A DIFFERENCE IN
hypotheses in words? MATHS RESULTS BETWEEN 15 YEAR OLD BOYS AND
15 YEAR OLD GIRLS

Q 12. Why would the researcher use an independent samples t-test to find whether
there is a difference in maths results between 15-year-old boys and 15-year-
old girls? [1]
THE INDEPENDENT SAMPLES T TEST IS USED TO FIND A DIFFERENCE
BETWEEN THE MEANS OF TWO INDEPENENT VARIABLES

Q 13. Calculate � to determine if the variances of the maths scores for the boys and
girls are homogenous. [2] (round off to two decimal places)
�21
�= 2
�2

75,22
�=
73

� = 1.03

Q 14. Your answer to the question above should be below 4, which means your
variances are homogenous. You should therefore use the pooled variance of
the two samples when calculating the standard error. Calculate the pooled
variance of the two samples. [3] (round off to three decimal places)
2 2
�1 − 1 �1 + �2 − 1 �2
�2� =
(�1 + �2 − 2)

(117 − 1)75,22 + 124 − 1 73

�2� =
(117 + 124 − 2)

8725,52 + 8979
�2� =
239

�2� = 74.077

7
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

Q 15. Now using the formula (Equation 8.6) for an independent samples t-test with
pooled variance estimates, calculate t (tcalc). [3]
�1 − �2
�=
1 1
�2� +
�1 �2

54,34 − 60,25
�=
1 1
74,077 +
117 124

−5,91
�=
241
74,077
14508

t = -5,327

Q 16. Calculating the critical value for t (tcrit), against which the researcher can then
compare her t-value (tcalc), depends on the degrees of freedom (df). Calculate
the degrees of freedom for these two samples. [1]
df = (N1+N2-2)
=117+124-2
=239

Q 17. Why would the researcher use a two-tailed test? [1]

TO DETERMINE IF TWO MEANS ARE DIFFERENT FROM ONE ANOTHER

Q 18. Using an alpha of 0.01, your degrees of freedom, and the t-table (Table A1.2
in Appendix 1 of your statistics textbook), determine the critical t-value (tcrit). [1]
tcrit = 2,6259

Q 19. We now need to decide whether to reject the null hypothesis or not. We do
this by comparing tcrit with tcalc to determine whether tcalc falls within the
rejection region. Use the table below to come to a decision whether to reject
the null hypothesis or not. [2]
Rewrite tcrit 2,6259
Rewrite tcalc -5,327
Does tcalc fall in one of the rejection NO
regions?
Delete the incorrect statement We reject the null hypothesis.

8
 Psychology 201 Tutorial 8 exercise: Hypothesis testing – the t-test

What conclusion can we make? THERE IS IN MATHS RESULTS

BETWEEN 15 YEAR OLD BOYS AND 15
YEAR OLD GIRLS

Total marks _______ [35]