Federerr
Badges: 19
Rep:
?
#1
Report Thread starter 2 years ago
#1
Hi guys,

Basically, for my coursework I am looking at different opinions people have of an area using social media.

I would like to conduct a statistical test on the correlation or significant difference between the date of the post and its opinion (good or bad)

I really don't know where to start and what test would be the best to do in the situation. Hopefully one of you guys would be able to help me out.

Thanks
0
reply
chocochip_
Badges: 17
Rep:
?
#2
Report 2 years ago
#2
wait i’m confused on the aim, you want to see whether the age of a post affects a person’s opinion?
0
reply
Federerr
Badges: 19
Rep:
?
#3
Report Thread starter 2 years ago
#3
(Original post by chocochip_)
wait i’m confused on the aim, you want to see whether the age of a post affects a person’s opinion?
Yes - Just a bit of background context is the area has had significant changes in recent times through regeneration schemes.

So if I can use a statistical test to see if there is a significant difference between types of opinions before regeneration schemes and the opinions after regeneration schemes.

Which statistical test would be the best to do so and how would I go about doing this?

It's a pretty vague explanation, I apologise.
0
reply
chocochip_
Badges: 17
Rep:
?
#4
Report 2 years ago
#4
if that’s the case then you’re looking for a correlation, so i would do spearmans rank maybe? will you rank how they feel about the post on a likert or rating scale?
0
reply
chocochip_
Badges: 17
Rep:
?
#5
Report 2 years ago
#5
OHH right i got you, I would use Wilcoxon
0
reply
Federerr
Badges: 19
Rep:
?
#6
Report Thread starter 2 years ago
#6
(Original post by chocochip_)
if that’s the case then you’re looking for a correlation, so i would do spearmans rank maybe? will you rank how they feel about the post on a likert or rating scale?
Yeah I was thinking about doing spear-mans rank as well.

However, the data I get on social media is qualitative. So using my own perception i'm going to need to categorise each post as to whether it's a positive or negative opinion.

If I had 10 posts in maybe 2007 and judged the number of positive opinion to negatives. Then did the same for 2017. How could I implement this into a statistical test?
0
reply
Federerr
Badges: 19
Rep:
?
#7
Report Thread starter 2 years ago
#7
(Original post by chocochip_)
OHH right i got you, I would use Wilcoxon
How difficult is it to do so haha. I've never heard of this test in my life.

BTW thanks for the help, it's really appreciated.
0
reply
chocochip_
Badges: 17
Rep:
?
#8
Report 2 years ago
#8
You’re looking for a difference as your seeing whether a post displaying the image of an area after regeneration will have an impact on the opinion on the person viewing the image, (i’m assuming you’re going to have a directional hypothesis as you’re guessing that people will have a more positive view of the image after the regeneration schemes.) you’re using the same participants in both conditions so you would use a repeated measures design and that would produce related data so i would use a wilcoxon test
0
reply
chocochip_
Badges: 17
Rep:
?
#9
Report 2 years ago
#9
Ah I see! well if you’re doing a rating scale (which i might suggest that you do.) from say 1-10, 1 being strongly dislike 10 being strongly like/love 5 being neutral then that would make it quantitative and ordinal data and would allow you to do a wilcoxon test, however if you keep it as positive vs negative then you have categorical data (nominal) so would do a sign test
0
reply
chocochip_
Badges: 17
Rep:
?
#10
Report 2 years ago
#10
(Original post by Federerr)
Yeah I was thinking about doing spear-mans rank as well.

However, the data I get on social media is qualitative. So using my own perception i'm going to need to categorise each post as to whether it's a positive or negative opinion.

If I had 10 posts in maybe 2007 and judged the number of positive opinion to negatives. Then did the same for 2017. How could I implement this into a statistical test?
If you’re doing this you would do a sign test, if you use a ranking scale you would do wilcoxon, if you need help with these mathematically you can PM me
0
reply
chocochip_
Badges: 17
Rep:
?
#11
Report 2 years ago
#11
I don’t think you should do spearman’s rank because you’re looking for a difference in people’s opinion before and after the regeneration scheme
1
reply
Federerr
Badges: 19
Rep:
?
#12
Report Thread starter 2 years ago
#12
(Original post by chocochip_)
I don’t think you should do spearman’s rank because you’re looking for a difference in people’s opinion before and after the regeneration scheme
OMG you're being so helpful right now.

Let me just explain to you how i'm going to conduct this.

1) So using google reviews - I will be randomly gathering 10 reviews people had of an area in 2007 and will be doing the same for in 2017

2) Each post is obviously qualitative - So using my own judgement and perception I will categorise each post into whether its a positive opinion of the area or a negative opinion of an area.

3) Then i'll have number of positive compared to negative posts of an area in both 2007 and 2017.

4) Using what I now have what is the best form of statistical analyses to check for any different.

It's rather confusing because there isn't really a way of ranking the data.
0
reply
chocochip_
Badges: 17
Rep:
?
#13
Report 2 years ago
#13
Ah I see now Do you have to use google reviews? Couldn’t you use volunteer sampling or opurtunistic sampling and get people to come in and look at both images and state opinions on the images/rank the images, this was you can avoid ethical issues because random people on google reviews may not want you to use their data in your study, (this obviously isn’t a major issue but just letting you know just in case ethical issues is being marked in your essay!)
0
reply
Federerr
Badges: 19
Rep:
?
#14
Report Thread starter 2 years ago
#14
(Original post by chocochip_)
Ah I see now Do you have to use google reviews? Couldn’t you use volunteer sampling or opurtunistic sampling and get people to come in and look at both images and state opinions on the images/rank the images, this was you can avoid ethical issues because random people on google reviews may not want you to use their data in your study, (this obviously isn’t a major issue but just letting you know just in case ethical issues is being marked in your essay!)
Ah yes, that would have been perfect. However, I've already conducted data analysis on photos and the last part of my 10k coursework (lol) is the use of social media.

I also havent used any statistical test. So I was hoping, I could do the test on data I get from social media.

I'm free to use any type of social media - If you have any suggestions.

My teacher said to just hide the picture and name of the person and I should be good - Obviously I did highlight this in my evaluation.
0
reply
chocochip_
Badges: 17
Rep:
?
#15
Report 2 years ago
#15
Plus you need the same participants in both conditions right ? 2007 and 2017? How can you be sure that the people who reviewed the area in 2007 wrote a review in 2017 after the regeneration scheme?
0
reply
chocochip_
Badges: 17
Rep:
?
#16
Report 2 years ago
#16
Oh i’ve got you, i think you should use the sign test then 100%!
0
reply
Federerr
Badges: 19
Rep:
?
#17
Report Thread starter 2 years ago
#17
(Original post by chocochip_)
Plus you need the same participants in both conditions right ? 2007 and 2017? How can you be sure that the people who reviewed the area in 2007 wrote a review in 2017 after the regeneration scheme?
yup, I will speak about this in my evaluation. As it is almost impossible to get someone to post about a place in 2007 and 2017. Hence, why I will be randomly choosing 10 posts in each year (idk if that solves anything lol)

In this coursework most marks are in the evaluation, so should the statistical test not really work out. I'd still get marks for having it and doing it.
0
reply
chocochip_
Badges: 17
Rep:
?
#18
Report 2 years ago
#18
(you do have the same participants in 2007 and 2017 condition right?)^
0
reply
Federerr
Badges: 19
Rep:
?
#19
Report Thread starter 2 years ago
#19
(Original post by chocochip_)
Oh i’ve got you, i think you should use the sign test then 100%!
Alright thanks a bunch for that! I'll watch a few youtube videos to see how to do it. If I need any help, hopefully I can PM you?
0
reply
Federerr
Badges: 19
Rep:
?
#20
Report Thread starter 2 years ago
#20
(Original post by chocochip_)
(you do have the same participants in 2007 and 2017 condition right?)^
Nope
This is what's confusing me loool.

Do you think, It's best to do what you suggested? To get 2 photos. One in 2007 and one in 2017 and get 10 people to rank the photos?

It's the easiest method right?
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 (19)
19%
I'm not sure (2)
2%
No, I'm going to stick it out for now (28)
28%
I have already dropped out (3)
3%
I'm not a current university student (48)
48%

Watched Threads

View All
Latest
My Feed