Please help me with a couple of points in my dissertation, I think i'm going mad

Watch
Wildcard123
Badges: 0
Rep:
?
#1
Report Thread starter 5 years ago
#1
Hi Guys,

I received the results of my Diss today, and unfortunately failed by 4 points.

I've read through my feedback and there are a couple of things I don't understand:

"Introduction would have benefited from moreup-to-date literature"

But there isn't any!

"The introduction could be clearer if the critiques of previous researchwere not presented immediately after the study details. It hinders the flow ofthe section and makes it less clear – especially if you are not addressing allof these critiques in your study."

I've been presenting them as in procedure, results, limitations/strengths. This isn't right? Where do I put the limitations then?

"It is not clear why the chosen analysis wascarried out. The results of the Mann Whitney test were reported incorrectly asbeing significant when they are not. The Chi Square analysis is not anappropriate test for the design or data in this study. "

Pretty self explanatory, but I have no idea what design I was supposed to use. I am absolutely rubbish at this part, so if someone help out here I would be eternally grateful.


Thank you guys, I feel completely and utterly lost!
0
reply
iammichealjackson
Badges: 18
Rep:
?
#2
Report 5 years ago
#2
It's hard to really make any comments without knowing what you did.

They're correct that you shouldn't write about things in your introduction which aren't directly related to your study (even if they are really smart critiques),and you definitely do not need to write the procedure, results, limitations/strengths of every study you mention in the intro.

The APA publication manual is a good guide to writing research.

From the APA publication manual on how to write an introduction:
Introduce the problem.The body of a manuscript opens with an introduction that presents the specific problem under study and describes the research strategy. Because the
introduction is clearly identified by its position in the manuscript, it does not carry a
heading labeling it the introduction.
Before writing the introduction, consider the following questions:
� Why is this problem important?
� How does the study relate to previous work in the area? If other aspects of this study
have been reported previously, how does this report differ from, and build on, the
earlier report?
� What are the primary and secondary hypotheses and objectives of the study, and
what, if any, are the links to theory?
� How do the hypotheses and research design relate to one another?
� What are the theoretical and practical implications of the study?
A good introduction answers these questions in just a few pages and, by summarizing the relevant arguments and the past evidence, gives the reader a firm sense of
what was done and why
0
reply
Wildcard123
Badges: 0
Rep:
?
#3
Report Thread starter 5 years ago
#3
The basic premise of my study is that:

Participants take part in a Paranormal Belief Scale (measures their level of paranormal belief)
Then they take part in a personality quiz, which generates a list of Barnum statements, and they are asked to rate how accurately the statements portray them.

I am trying to see if there is a correlation between level of paranormal belief and susceptibility to Barnum statements. Would a Pearsons Moment Correlation Coefficiant be a suitable test for this?
0
reply
iammichealjackson
Badges: 18
Rep:
?
#4
Report 5 years ago
#4
(Original post by Wildcard123)
The basic premise of my study is that:

Participants take part in a Paranormal Belief Scale (measures their level of paranormal belief)
Then they take part in a personality quiz, which generates a list of Barnum statements, and they are asked to rate how accurately the statements portray them.

I am trying to see if there is a correlation between level of paranormal belief and susceptibility to Barnum statements. Would a Pearsons Moment Correlation Coefficiant be a suitable test for this?


It depends, How many different groups were people categorized into? If you only had 3/5 groups on your scales then it might not be appropriate to use a correlation because that works best for continouos variables when your predicting a linear relationship between two outcomes.
0
reply
Wildcard123
Badges: 0
Rep:
?
#5
Report Thread starter 5 years ago
#5
I only had two groups - high paranormal belief and low paranormal belief!
0
reply
iammichealjackson
Badges: 18
Rep:
?
#6
Report 5 years ago
#6
(Original post by Wildcard123)
I only had two groups - high paranormal belief and low paranormal belief!
how was susceptibility to barnum statements grouped?
0
reply
Wildcard123
Badges: 0
Rep:
?
#7
Report Thread starter 5 years ago
#7
It wasn't really, they just gave an accuracy score of 0-5. Should I put them in groups?
0
reply
iammichealjackson
Badges: 18
Rep:
?
#8
Report 5 years ago
#8
(Original post by Wildcard123)
It wasn't really, they just gave an accuracy score of 0-5. Should I put them in groups?
I don't think its okay to do a correlation between a binary (two-group) variable and a 6 group variable. Correlations should be between continuous variables or binary groups, rather than an ordinal variable.
0
reply
Wildcard123
Badges: 0
Rep:
?
#9
Report Thread starter 5 years ago
#9
So would there be an alternative method? This is based on a study from 1949 but there was very little information on what statistical test they used.
0
reply
iammichealjackson
Badges: 18
Rep:
?
#10
Report 5 years ago
#10
(Original post by Wildcard123)
So would there be an alternative method? This is based on a study from 1949 but there was very little information on what statistical test they used.
I wouldn't take any statistical advice from a 1949 paper, lots of the tests we now use didn't even exist then. Out of the traditional statistical tests, I think you could use a t-test if you had a big enough sample. Mann-Whitney U test would probably be the best bet though.
0
reply
Wildcard123
Badges: 0
Rep:
?
#11
Report Thread starter 5 years ago
#11
I did originally use a Mann-Whitney test, and the only feedback I got was that I didn't report the outcome correctly, but I can work on that.

I also used a Pearson's Chi Test and was told it wasn't appropriate, so I guess i'll remove that.

So to sum up, there are two levels to the IV (level of Paranormal Belief: High and Low) and the DV is just accuracy score (of the barnum statements). I can definitely use Mann-Whitney for this?

In my feedback it says: "A group-based design (e.g. low, high) cannot beused to test correlation of two variables." Should I get rid of the High/Low thing altogether then?

I'm so sorry for all this. I found my dissertation difficult the first time around, and having to go back to it months later isn't any better.
0
reply
iammichealjackson
Badges: 18
Rep:
?
#12
Report 5 years ago
#12
(Original post by Wildcard123)
I did originally use a Mann-Whitney test, and the only feedback I got was that I didn't report the outcome correctly, but I can work on that.

I also used a Pearson's Chi Test and was told it wasn't appropriate, so I guess i'll remove that.

So to sum up, there are two levels to the IV (level of Paranormal Belief: High and Low) and the DV is just accuracy score (of the barnum statements). I can definitely use Mann-Whitney for this?

In my feedback it says: "A group-based design (e.g. low, high) cannot beused to test correlation of two variables." Should I get rid of the High/Low thing altogether then?

I'm so sorry for all this. I found my dissertation difficult the first time around, and having to go back to it months later isn't any better.
There isn't such a thing as pearson's chi test (either pearson correlation & chi-sqaure test are seperate things).

If you want to get technical, then you can do a correlation for the type of design you did. Pearson's correlation generally isn't used when your comparing two groups, but it gives you the same results as a t-test which is more typically used to compare groups! (see below).

In addition, even if you have non-normally distributed data (as you do), it doesn't matter because its the sampling distribution that has to be normal (http://stats.stackexchange.com/quest...rmal-when-n50;). In other words, as long as your sample size is large enough (e.g. more than 30-50) then the t-test is robust to having non-normal distributions. If you want to be safe you can use the Mann-Whitney U test because most psychologists are taught to use that for ordinal data, however its usually fine to use t-tests on ordinal data (e.g. http://link.springer.com/article/10....459-010-9222-y )

Pearson's correlation and student t-test give you the same results (see p values):
Lets say we want to see if scores differ between males and females (10 in each group):
MALES: 5 5 5 5 5 3 5 4 5 4
FEMALES: 3 2 2 4 2 0 2 2 3 5

Pearson's product-moment correlation
t = -4.3578, df = 18, p-value = 0.0003793
alternative hypothesis: true correlation is not equal to 0
95 percent confidence interval:
-0.8800069 -0.4011983
sample estimates:
cor
-0.7165108
Two Sample t-test
data: males and females
t = 4.3578, df = 18, p-value = 0.0003793
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
1.087577 3.112423
sample estimates:
mean of x mean of y
4.6 2.5
0
reply
Wildcard123
Badges: 0
Rep:
?
#13
Report Thread starter 5 years ago
#13
Wow, that is incredible! Thank you so much! If I remember correctly my group is really small (only around 15 participants I think - I ran out of time) so can I still use those tests?
0
reply
X

Quick Reply

Attached files
Write a reply...
Reply
new posts
Back
to top
Latest
My Feed

See more of what you like on
The Student Room

You can personalise what you see on TSR. Tell us a little about yourself to get started.

Personalise

Current uni students - are you thinking of dropping out of university?

Yes, I'm seriously considering dropping out (150)
14.78%
I'm not sure (44)
4.33%
No, I'm going to stick it out for now (303)
29.85%
I have already dropped out (26)
2.56%
I'm not a current university student (492)
48.47%

Watched Threads

View All