|
Results
This is the scary bit, but do not fear 'cos Little Gerry is here!
What you want to do is present your results in a meaningful way,
and then make some sense of them in the light of your null
hypothesis. To do this we use some descriptive statistics and an
inferential statistic called the t-test. Before you do anything
though, you need to make up and complete a Raw Data table like the
one below.
Raw Data Table
Participant |
Gender
Male / Female |
Condition A
IQ Estimate Mother |
Condition B
IQ Estimate Father |
1 |
Male |
102 |
106 |
2 |
Female etc. |
100 |
110 etc. |
3 |
|
|
|
4 |
|
|
|
5 etc. |
|
|
|
Make sure in any study you include a completed raw data table as an
appendix.
When you have filled this in you can then use statistics to make
sense of it. You could for example work out
1 the mean and
median values of
IQ estimate in each condition.
2 construct a histogram
and a line graph
for each participant's estimate for each parent. Individual
participants go from 1-XX along your
x-axis
while IQ estimates go along the
y-axis.
With a histogram two 'bars' are allocated to each participant's
estimates - one for the mother, and one for the father. Make sure
axes are labelled; the diagram is given a name, and a fig. Number
etc
3 You could also do the same thing with all female, and then male,
estimates of each/both parents IQ. This may throw up some
interesting patterns.
4 You may also want to use an inferential statistic called the
related t-test
that will allow you to accept/reject your null hypothesis, and
consequently reject/accept your experimental.
Remember that you have a level of significance of 0.05, which means
that whatever happens you accept a 1: 20 probability that your
results occurred by chance or random factor.
You are allowed to us the related t-test because
|
a) Your research
hypothesis predicts a difference between perceived IQ
scores based on a person's gender. N. B. H1:
'That gender has a significant effect on perceived
intelligence.' |
|
b) You have interval
data (estimated IQ scores) |
|
c) Your data is
related because you are using a
repeated
measures design |
It is always a good idea before you present a statistic in a
results section to say what each statistical term means e.g. what
is meant by the mean or median of anything,
and why is a particular inferential statistic like the
related t-test appropriate in this circumstance.
Its appropriate because our data meets the conditions for the t-test
above. Say so and you'll get buckets of marks - as long as it makes
sense!
This is how you use the t-test. First consult your raw data chart
and construct the following.
Related t-test calculation
Put the condition you expect to be higher first!
Participant |
Condition
B
IQ Estimate Father |
Condition
A
IQ Estimate Mother |
d |
d² |
1 |
106 |
102 |
4 |
16 |
2 |
110 |
100 |
10 |
100 |
3 |
|
|
|
|
4 |
|
|
|
|
5 etc. |
|
|
|
|
|
|
|
(d) |
d² |
Steps
- Enter in your
participants 1-XX in the Participants column
- Enter in his
or her estimates for each parent e.g. Participant 1 102 and 106
- Find the
difference (d) between theses scores for each participant e.g.
with participant 1 the difference is 4.
- Then square
this difference (d²). With participant 1 d² is 16.
Having worked out d and d² for each participant
you now
need to work out two last things;
- Find the sum
of the (d) column giving you (d),
and then
- Find the sum
of the (d²) column giving you d²
In statistics when you see the sign
(sigma), this means 'add up'. The t-test is as complicated as this!
Now put this data into a t-test formula and work out the answer. You
have everything you need to know e.g. N, (d),
and d²
: N is the number of participants who took part in your study. The
related t-test formula is
You would have this formula in your
Results section, and underneath it the answer for calculated t.
The t-test calculation should be in an appendix. Direct the
reader to this. Once you have worked out calculated t, you want to
know if it means anything. Is your result 'significant' . To do this
is quite simple. You consult a critical value table for the related
t test for which you need to know three things
|
a) Whether you
have a one-or two-tailed hypothesis. This example is two-tailed. |
|
b) The level of
significance set for your null hypothesis; being 0.05,
and; |
|
c) The number of
degrees of freedom you are operating at. Degrees of
freedom (df) are calculated as N-1. N being the number of
participants who took part in your study. |
Lets say I ran this experiment with 22 participants and worked out
calculated t to be 7.71. Consulting the critical value table for a
two-tailed test, with a level of significance of 0.05, with 21
degrees of freedom (df = N-1, or 22-1 = 21) I find that critical
(tabled) t is 2.080. The instruction at the bottom of the table
should now be followed. It says that 'calculated t must EQUAL or
EXCEED the table (critical) value of significance at the level
shown.' My calculated t of 7.71 does indeed equal or exceed the
critical, tabled value of t, being 2.080 for a two-tailed test, with
a level of significance of 0.05, with 21 degrees of freedom (df =
N-1). As a consequence I can say in my Results section that on this
basis the null hypothesis is rejected and the Research hypothesis H1:
'That gender has a significant effect on perceived intelligence' is
accepted.
The results section in using the mean, the median, histograms, line
graphs and the related t-test makes statistical sense of our data.
This is extremely useful in the
Discussion
section that now follows.
Click here to continue on to the Discussion
Or Here to go back to the Method Section
|
|