SPEAKER 0

This material is made available to you by or on behalf of the University of

SPEAKER 1
On your knee. Hi, Aha. Good afternoon. How are you?

SPEAKER 2
Yeah, good, thanks. Do you need my camera on?

SPEAKER 1
Oh, it'd be good if you can. Um, I am recording, so I will upload this recording
for people. So if you want to keep your camera off for that reason, that's fine.
But just for interactivity, it might make it a bit easier if you can turn your
camera on. That would be appreciated.

SPEAKER 2
Yeah, and um, Cause the thing that I am. Like, um, kind of confused is I think
we're working on the table in Chi square, right? So there's like, um, a chi square
thing happening in jazz and so um. Uh, there's another table we'll be working on in
Word document. So I think these two basically contain the same information, am I
right?

SPEAKER 1
Uh, do you, I think what you mean is just creates a table, and you also create a
table in your report, and even though they have the same information, they are
different tables. Is that what you mean?

SPEAKER 2
Yeah, because I realised that these two tables actually have the same data. So I
was afraid that I'm doing it wrong.

SPEAKER 1
Oh, for your hypothesis 1 and hypothesis 2, do you mean? Or it's just, um, for the
same hypothesis, the one

SPEAKER 2
from Jasp and the one from the Word document actually had the same information.

SPEAKER 1
Yes, that's correct. They, they will have, um. Maybe, do you want to share a screen
with me to show me what you're referring to, because there are multiple tables that
we could be talking about right now, and I don't want to give you the wrong advice.
And then we'll get started after that.

SPEAKER 2
Yeah, um, Thank you. So, I'm have to scroll down to that section.

SPEAKER 1
And just for everyone else while you're doing that, we haven't started yet. We've
got about 11 people that are coming in, so I'm just helping you and you with this
problem, and then we'll jump into the workshop proper after that.

SPEAKER 2
So my, my laptop's lagging is actually like, OK, um. How do I share the, the
document like directly?

SPEAKER 1
Uh, down the bottom, oh no, it's better if you share screen than share directly so
that it captures on the recording and everyone knows what you're referring to. But
down the bottom of zoom, if you hover your mouse over zoom, there's a green button
that says share.

SPEAKER 2
OK. So Yeah.

SPEAKER 1
Can you all see that? Now we can, yep, so this is this Jasp, this is, yeah, so this
is actually the, the one from

SPEAKER 2
Jasp and also a chi square one. So, um, and I'm basically doing one from Word
document following that video you've uploaded, and this is actually like this. So I
think this one, so I mean this and this one looks the same like.

SPEAKER 1
Yeah, that's exactly what you're saying that. That's really helpful. Thank you. You
are correct. The, the table that you're building that that video helps you build,
uh, is the table that goes in your report, and it has all the same information as
that top information in JaSP. The only reason you build it yourself is so that you
can get it into APA formatting, because that Jasp table, even though it has the
same information, it is not APA formatted.

SPEAKER 3
Uh, OK.

SPEAKER 2
So we're basically doing adjusting APA.

SPEAKER 1
Yeah, that's right. And then that second table there, that chi square test table
that's on, on the screen, you don't put that in your report as a table. Instead,
you spell you say that in, in the text.

SPEAKER 2
Oh, so this one's actually converted into the text.

SPEAKER 1
Yeah, yeah, that's right.

SPEAKER 2
Thanks so much. That's like, that's confusing me, you know, for thinking about it
for days.

SPEAKER 1
I'm really glad you asked that, that was really helpful. Thank you. I was wasting
the time, sorry, I'm so sorry. No, no, I don't think that was a waste at all. Um, I
think that's going to be very useful for other people who couldn't make it here,
um, who are wondering the same thing. That was good. Thank you, thank you. So, um,
I thought I would jump in, there's only 11 of us, uh, signed up for the session,
and there's only 6 of you here, so, I'm gonna get your vote on this. Do you want me
to go through each of your specific 6 or 5 questions, one by one? Um, or Do you
want to keep it all anonymous and work together answering those questions? Cause
everyone's everyone wrote me a question, right, ahead of time. Do you want to keep
it anonymous and then just work through those questions together as a group, or do
you want me to go through one by one each person and have a chat with you about
your specific thing? Completely up to you, I don't mind either way. I want to do
what's gonna work best for you. OK, Small humans running around, uh, in the
background is gonna make it a little bit tough to have discussion. So maybe what we
will do. Uh, if I pop. No ices. OK, I can't. That's all right. I'm just gonna set
up my screen so that I can share something with you. pop that over there.

SPEAKER 3
And She.

SPEAKER 1
I'm gonna share a link with everyone in the chat. Welcome XZ. Do you want to pop
your real name into your title of who you are, or? Um, turn your camera on, one of
those two. I do have someone attending the workshop with XZ, but just so that I
know exact uh as their initials, but just so I know exactly who I've got on this
link and that we don't have someone zoom bombing us. Would you mind putting your
real name into your Zoom name? I'm hesitant to continue unless you do that. It says
not contributing and not responding to me. I'm just putting a message on their
screen to change their name, which we can all see at the moment. Like I said, if
you don't do that, I'm gonna have to kick you out of the zoom and continue so that
people here feel safe and free to talk about what's happening for them with their
square. And then when you can put your new name in, you can come back and join us,
OK? OK. All right, I'm sorry about that. Uh, I'm just gonna share. I think with
you. This is a link to a um. Google Drive? So I've put everyone's questions that
people asked ahead of time into this Google Drive. I've also put some resources at
the top there that when I was reading through the questions, it occurred to me that
the answers could be found in these documents from my perspective as a teacher, but
it's possible that my perspective as a teacher is wrong, and I think that the
answers are there, and you working through those resources have not been able to
find the answers. So, what I'm gonna do. Is we're going to go through each of these
questions together, and I've given you access to edit this as well, so that when
you jump in, And we go through, if you see a question there that you know the
answer to, you're very welcome to type your answer here into the type your answer
here space, or share the resources that you think would help the person who's asked
that question there, OK? So, we'll go quickly through them now, I'll introduce you
to all of the questions. And then I'm going to take a moment to show you through
some of the resources in slide one. While I'm doing that, feel welcome to add your
answers or thoughts to these questions as we go through. I'll show you how to
calculate the chi square, and I'll show you what you're looking to end up with as
your final product for your report. Um, and then we'll come back to these specific
questions and talk them through one by one. How does that sound? That's OK, cool.
Thank you, Andre. OK, so first question, I'm confused about how to use Jas, I don't
know how to do it, and I don't know how to get the data I want. It's a really good
question. 3 people had questions similar to this. Um, I'll, I'll I'll show you how
to do that in this workshop. It's also being recorded so that you have all the
answers, but also for hypothesis one, the answers are in lab 3. So hopefully doing
this today gives you enough practise that you can have a go at that for your second
hypothesis. Um, how do we introduce tables or interpret their data as they relate
to the hypotheses? There's resources available for that, but have a think, if you
know of any, you can pop them in there. Do we write our chi square results from
Jasp into word descriptions in the results? Have a think about where you might find
the answer to that. Do we mention if the observed counts were fewer or greater than
expected, even if the standardised residuals are not significant? Uh, there were
213 cases excluded from our full number of people who accessed the survey. Um, is
that a limitation, does that affect the sample size of people over 25? These last
three are a little bit more getting into the nitty gritty. So if this worries you,
don't worry, these are just questions that people are asked that are flipping them
up, but they're not necessarily going to affect your report. Um, is it correct that
the standardised residuals do not imply an association between variables, but
instead they reflect differences between expected and observed counts? And how many
combinations of hypothesis one do we need, uh, in order to show the results? So,
I'm gonna take you through Jasp itself, um. I'll show you how the data set works.
This is your actual data. In lecture 7 I used a fictional data set, this will be
your actual data. I'll show you how it works. Um, feel welcome to unmute at any
time or to post a message, uh, I will. Move my message chat over here so that I can
see if you post a message and ask a question as we go, OK? Um, the other thing I'll
be using or that you can use as we go through this as well if you want, is the
guide to calculating square. So does everyone on the call at the moment know where
to find that? The guide to calculating square, Andrea's not in uni, you know where
that is. Yep, Emily knows. Asher, have you seen this as well? Asta has. This is in,
if you go into. The canvas site, and go to either lab 3, the assessment itself, or
if you go to, uh, the week 5 activities page, or even the modules page, you'll find
something called the guide to calculating Chi Square. Uh, and it looks like Yes Did
you find it? Um, We wrote this. Uh, once we closed the survey, we downloaded the
data, and we cleaned up the data set, and we've given you as much information here
as we can so that you know how to run, uh, your analysis, so that you can do it for
hypothesis one, and then so that you can do it on your own for hypothesis 2.
There's all sorts of other information in here as well, so that person who asked
the question about the 213 cases that were removed, they're getting that
information from here. Um, so What happened was, we had, uh, 892 people access the
survey at some point while it was open. Some of those people opened it multiple
times over. Some of those people opened it, sped through it, didn't answer
anything, and then closed it. Some people opened it and left it open for weeks at a
time, telling us that they weren't genuinely answering the questions. Um, some
people answered it and gave us answers that were fictional, like being 110 years of
age. So for all those cases, we delete them. We don't count them as legitimate
data. We don't count them as, as, um, authentic data, and we remove them from the
data set. They don't count as a sample at all. Um, that's normal. That happens in
every data set, every survey you've ever been a part of. The researchers will
download it and delete the things that they don't think are authentic responses. So
we're just letting you know that that happened. Uh, on average, people had the
survey open for 3 hours, but it took them, most people completed it with, um,
within 4 minutes. That's what the mode tells you, the highest group of people who
had the same amount of time. Um, there are 4 variables in the data set that we
converted from continuous data. We converted the, we hoped to give you the age
continuous variable, so that you could calculate your own mean and standard
deviation of the age, but we weren't able to do that because we had a few people
under the age of 18 who were minors, and we had a few people over the age of. I was
around about 35, where we started to see unique ages, so only one person of that
age. Um, Because of that, that made the minors vulnerable, first of all, and the
people with unique ages, only person in the cohort at that age, they were
identifiable. And we have to protect the privacy of those people. So, instead of
giving you the mean score where you can find each of those people, instead of
continuous data, we created categories, and we put the minors, people of 17 years
of age, 18 and 19, together in this group called Under 20. And we put people over
the age of 30 in this group over 30. We combine people who are 25 to 30 in this
group, 25 to 30, and in doing that, we're still showing you the distribution of the
ages, but it's categorical, not continuous. Is that OK with everyone? That's a
little bit different from what we wanted to do. Uh, Identity status worked the way
we thought it would work. We had beautiful normal distributions, picked the
midpoint of the distributions, had people lower than the midpoint and higher than
the midpoint, um, to work out if they had low or high exploration and commitment,
which tells us whether they were achieved, um, foreclosed in moratorium or in
diffusion. So that worked really nicely. Friendship quality was a hot mess of the
category. Um, most people said that their friendships were supportive, so the mean
score out of 15 points was 9. Most people said, My friend is very supportive. Uh,
and most people did not acknowledge any conflict in that friendship at all. So out
of that same scale, 0 to 15, the mean score was 2. So our original plan, and as
we've done in the past, is to compare those two scores and say, well, anyone with a
score of support that's higher than conflict, uh, we're gonna call them a
supportive relationship. Anyone with conflict height and support, we'll call it
conflicting relationship, and so on. We can't do that with this data set, because
nearly everyone who said their friend is supportive. So what we did instead is we
Keyword is this one, we transformed the data, this is also normal, it's a normal
thing to do with skewed data. We transform it so we have the mean score, the middle
group, we have people lower than that middle group and people higher than that
middle group. And then we compared those categories. So if someone had higher,
higher support than everyone else, and higher conflict than everyone else, we said
they had a mixed, uh, friendship. If they had higher conflicts than everyone else
but lower support than everyone else, we said they had a conflicting relationship.
So the key in telling you all that is that it didn't work out the way that we
wanted it to, so we had to transform the data. That's normal. We still end up with
our categories, but they mean relative to the rest of the sample,
rather rather than, uh, being in a conflicting relationship overall. I hope that's
not too overwhelming for you. I'm giving you a little bit more than um we gave
everyone else. It's all here, I'm just saying it verbally. OK, is that OK so far?
Alright, great. So we have all this here. We also have, by the way, our mean age
and our standard deviation for age for the whole sample. So in your method, when
you're describing your participants, this was the overall mean and standard
deviation of age for the whole sample. You're going to need that. OK, um, then we
get into how to access the data file. Is anyone here having trouble accessing the
data file itself? No, OK dokey, great. Um, then we get into Jasp. So I'm gonna jump
into jas, which Looks Like this. This is a data set in Jas. Um, I got here, I've
opened this up before just to make sure I could, by clicking the menu, clicking
open, and then navigating to where I had it located on my computer, that's all. Uh,
4. I'll turn these off for the moment and just show you what the data set looks
like. Every row is its own person, so we have 679 complete cases here, so 679
participants in our sample. We've got their age, we've got their gender, we've got
their enrollment status, and so on, all the way through to the two variables that
we're interested in for our first hypothesis, identity status and friendship
quality. Every person has their own label in those columns. Uh, I won't dive into
the meaning behind all the numbers in the Kasquare statistic for the moment. If you
want that, we can talk about it, but it wasn't anyone's burning question coming
into the workshop, so I won't do that. I'm just gonna work straight into
calculating hypothesis one. So we know for hypothesis one, that we want Everyone in
our analysis to be under the age of 25, right? So to do that, we can either click
this little black philtre here, that button means philtre, or I can double click on
the variable that records people's age. Um, sometimes it comes up looking like that
and looks a bit confusing. That's OK, just click the other tab and go to label
editor. Um, so I know I want to get rid of or exclude everyone who is over the age
of 25. So if I just tick next to these two, see how they turn into crosses. So I'm
saying to Jas, include everyone with all these ticks, exclude the people who have
these, um, these two categories thanks to them. And you can see that if I just
close that variable window for a second. When we scroll through the data set, see
how anyone over the age of 30 is greyed out now. So JAS is not going to include,
you grade 25 to 30 grade out, just won't include these people in our analysis. And
we also know that because it tells us down here, we've got 679 rows, only 621 have
met the requirements that just needs. Tell me if I'm going too slow or if I'm going
too fast. Yeah, Andrea.

SPEAKER 3
Sorry, I can mute now because I sent them to the park, um. I can't see what you're
sharing. Oh, thank you.

SPEAKER 1
Thank you. No, let me try again. Has that been the whole time? Yeah. Oh my gosh,
thank you for telling me, Andrew. I've showed you through the guide. I was talking
you through the guide. Boy oh boy. I really appreciate you telling me. We could
have gone through the entire thing without. Um, I don't want that. OK, so thank
you, thank you for telling me. Let me show you, first of all, just for the sake of
the recording, the Kai Square guide, so that, I think it was Asher, maybe, who
didn't put their hand up knowing where it was. Can you see now, the Kai Square
guide? Yeah, I, I actually know where it was. Um, I might have missed the question.
I'm so sorry. That's OK, that's all right. I was, that whole time I was talking, I
was just scrolling through this, and you were very politely listening along, um,
and then I opened up Jasp. To show you, here's the menu for how you open data on
the site there. Sorry to do this so quickly. Uh, and then here's the little black
philtre, can you see that now? You can see my just window? Yeah, thank you, great.
Um, here's the age. Uh, variable, I've put two little crosses there and there, so
I've told Jas not to include those two categories. Is that OK? Or am I still not
sharing the right screen? I am.

SPEAKER 3
That's right, yeah, yeah, OK, thank you, great, great, great.

SPEAKER 1
OK, so. I've told it, don't, don't include anyone in the category 25 to 30, and
don't include anyone in the category 30 and over, and now you can see their grade
out. So Jasp knows not to include them. And at the very bottom down here, data has
679 rows. That's how many participants we have in total. Only 629 have passed
through the philtre. So I'm only including 621 participants who are aged 25 and
under. OK, so to run the analysis. Uh, I'm going to do a frequencies analysis, so
we're counting the frequencies. That's what Chi square does. We've got two sets
here. We're going to ignore the Bayesian stats. We're not doing Bayesian stats in
this subject. We're just doing classical stats. One of my tutors said to remember
it, by C for chi square, C for classical. I think that's quite nice. Uh, so
contingency table. Yeah. And it looks like it's all changed. All that's happened,
my data set is sitting back there. If I pick up this little tab and scroll it left
and right, I can hide or show my actual data set. Then it gives me this table. I'm
so self-conscious that I'm not recording it, um, the screen anymore. Please feel
very welcome to unmute if something goes wrong and you can't see what I'm referring
to. Um, this first little table in the middle here, this is our analysis tab, so we
can run as many analysis as we want here, I'll just make some other ones up just
for the fun of it. And each time, We're just recording it there, we can always come
back to something that we've run by clicking the little arrow next to it. We only
want for this analysis, a contingency table, so this is the one I'm gonna look at
here. The 3rd output, this one over here, this is our results tab. So later on when
we change a couple of the um features of results, it's changing the features of
this particular tab over here, and we can move these around as much as we want to
see whatever we want to see. OK, so, uh, we're gonna put friendship quality in the
columns, just to be consistent with the lecture, and identity status in the rows.
It doesn't matter, you can switch those around, but just for simplicity, I'm gonna
put them that way around. We say to Jas, I want you to do some extra things in the
cells here, right? We have our observed counts. We know we also need expected
counts and the residual counts. So in cells of my chi square, I'm gonna turn on the
expected count, and I'll turn on the Pearson count. So Pearson is Jasp's way of
referring to the standardised residual. It's a long story for another day, if you
want it. Um, but see how, as I did that, we've got our 621 as our total table
count. We've got our observed count still. We've got our expected count. We've got
our Pearson residual, and then down here we have our square statistics. So lecture
seven will give you much more about what each of these statistics actually means.
What that looks like in lab 3. With this. Lab 3 took students through the Chi
square guide, step by step, and then, Put this up on the slide for uh, for
students, because this is the output for hypothesis one. You've got our 621
participants, we've got our 4 statuses, we've got our 3 friendship qualities. Uh,
we've got the observed count, the expected count, and the piercing residual. Now,
back to your news question right at the very start, this is not APA formatted. What
we want for, uh, your lab report. Um, looks like this. This is from just from the
ALM, right? Completely different topic from you, but still a chi square. We want
tables. This is probably not a great example because it's not APA formatted. Let me
open up the other one. Got two high square. Um, examples, where's sample 2. Yeah.
For some reason as a word document on PDF, but that's OK. What we need is APA
formatted tables. If I just push that down so we can see it on one page for a
second. That look like this, right? So you'd have your statuses in the two rows,
this is 2 rows, you're gonna have 4 rows. You have your friendship quality in the
columns there with your observed count, your expected count, and your standardised
residual. It does look quite different from the jasp table, but as you pointed out,
it's the same information, just Formatted slightly differently, formatted as APA.
So that's the kind of thing that we're looking for, um, so you use your, Just
output. Yeah. Uh, create the table for yourself and convert these numbers into that
APA formatted table. OK, so that, that's probably as quick as I can do it while
talking you through it. Do you want to come back to the workshop, um, uh, Google
Drive and answer some of those questions now? OK doke, alright, I'm gonna put your
faces like this so I can see everyone's names. Um, Now, it looks to me, 12345. Is
everyone here, was everyone here when I shared the link to the Google Drive? Yeah,
OK, great. OK. So, Let's have a look through uh some of these questions now. The
first one, I'm confused about how to use Jap. I don't know how to do it to get the
data I want. I don't, I don't want to call anyone out and say if this was your
question, did I, do you need, do you still need help? But maybe you can say, here,
um, If, if A stu a fellow student asked you if they needed help, where would you
send them for support or for information? There's one place I would send, oh, you
go, you go.

SPEAKER 3
Uh, probably the God. The guide to calculating password and then the, um, lab notes
as well. Yeah that's.

SPEAKER 1
Yeah, I that's what I was gonna say too, the Chi square guide. I feel a little bit
biassed though, because I wrote the Chi square guide. So if the person who wrote
this question, and to be fair, 3 people wrote this question, um, doesn't find that
guide helpful, I need to know why, and I need to know how. So if you have further
questions about that, yeah, there we go. Oh, that's so nice. I don't know who wrote
this, but it's very, very nice. I'm gonna lift this up and show you. Starting with
Jas can feel overwhelming. What proved helpful for me was delving into some sample
data folders and exploring the data set without the pressure of making risky
alterations. That's such a good idea. The JAS website itself does have pretend data
sets, um, Fictional data sets that you can play with and do all sorts of things
with and learn how to use just, that's a really good idea. Um, I really like who,
who wrote that. That's really, really helpful. Thank you. Uh, I'm gonna add that
there. Less resources to play with the data. The second question, how do we
introduce the tables and interpret the data as they relate to the hypothesis? I'm
gonna jump in and answer this, um, to show you what I've shared so far, and if it's
not enough, feel welcome to speak up and say what you think someone else might need
to answer that question. But my first thought in answering that was lecture 7. Um
Towards the end of lecture 7, I showed some examples of, How to write up the
descriptive section. Right? So, first of all, the examples in the ALM, those
example lab reports, show you, you need to write your descriptive statistics, you
need to introduce the table, say the main features of the table. You need to state
that your analysis, I'll jump onto the last pages a bit quicker to see. Um, state
whether state your analysis and whether it was significant or not, and then explain
the standardised residuals for the cell that you, or the standardised residual
singular for the cell that you were predicting. So all that text there is what you
need for each analysis. Um, the, the slides just above that, give you examples of
bad through to best. Examples of how to do all of those things. And I invited
people to tear these apart on the discussion board, and no one took me up on that
offer, which I think just shows that we have a very polite and nice cohort that
wasn't so keen to, um, critically tear apart different examples. So the key point
then would be that the best example there shows you the level of detail you need to
go when introducing your table and describing what's in the table. That to me would
be the best resource for an example of how you go about writing in words and then
showing the table itself. Um, any other ideas of what we can use? Or why that
wouldn't be enough and what else you would need to, to know how to write that up.
It's OK, no worries. I'm gonna leave that with lecture 7 then. We can come back to
that if if the person who asked that wants more help, we can come back. Um, do we
write? And I am behind in lectures, so that was probably

SPEAKER 3
a silly question.

SPEAKER 1
No, no, not at all. Not at all, especially if you're behind, you don't know that
that's happened yet. That makes sense. So now you know you can shortcut to lecture
7.

SPEAKER 3
I was definitely having trouble with the, um, The P value and then the pier and,
and significance versus insignificance and, um, You know, whether a large number is
better or a smaller number, and then, yeah, that was very confusing to me. Oh,
good.

SPEAKER 1
Oh, you're gonna like lecture 7 then we get into that. We get into why you can't
just say a chi square of 21 is significant. Um, I don't know if anyone else who has
seen the lecture, but the Spoiler alert is it depends on the size of the table. So
your degrees of freedom is telling the reader the size of the table. You say, with
a table of this size, high square value of that, which is the amount of variation
in the table, essentially, those two things together tell us whether it's
significant or not significant.

SPEAKER 3
Yeah, um, well, I've got you. With the mean and standard deviation of the sample
ages. If we choose to change the sample to just females, Um, do we just report what
they mean, like, use the descriptives and just to Change what's it's, it's a

SPEAKER 1
good question. It's one that the tutors were just talking with me about before this
session. The your full sample is 679 students. Your first analysis focuses on 621
of those. Your second analysis might focus on how many females do we have in total?

SPEAKER 3
Um, uh, I think there was about 500 and something when I ran it. OK, cool, cool.

SPEAKER 1
So your, your first analysis focuses on 621 people within that sample of 679. Your
second analysis focuses on 500 and something people within that 679. So, your full
sample is still the 679. So in your method section, in your participant section,
describe the distribution of age for the full sample. And then describe the
categories that led you to each of those, so like, um, Um, I don't give you the
full thing, but just as an example. You want to talk about the age variable,
because the age variables how you decided who would be in analysis one. And you
want to talk about the gender variable, because the gender variable is how you
decided who would be in hypothesis 2, but you don't then need to give me a mean age
distribution for hypothesis 2. Cause it's still the full sample that you're looking
at. The, the only tiny exception to that, and tutors were even hesitant to include
this, is if you think there might be a limitation of including just female
participants, such as they're all, let's say, under the age of 19 or something
ridiculously and, um, made up. Then you would want to include in your method
section 500 and something females, most of them, or all of them under the age of
19, and then in the discussion, say, my results didn't do what I wanted. It's
because I think I've got a really young female cohort here. Or something like that.
So the only reason you'd report. Characteristics of each of your subgroups is if
you think that's going to be a limitation coming down the track. OK. Does that make
sense?

SPEAKER 3
Yes. And in the tutorial, um, We're a bit unsure how, like with the tutor, we're a
bit unsure how to run. Uh, so we filtered out females. No, wait, does that make
sense? So we only had the female data.

SPEAKER 1
Oh, so filtered out the males, so you just female. Yeah, got you.

SPEAKER 3
And Let's do that. But Charlie ran it. Different way as well. And he came up with a
totally separate. Um, contingency table just for females. And so I saved that, and
I was like, well, I have to put both of those tables into my results section. But I
also ran it again using this method, and it all came up in one. So now I'm confused
as to Which I need to do. Ah, you need, if I understood you correctly.

SPEAKER 1
Um, you need You will need 2 contingency tables total in your results section. One
of them should address hypothesis 11 should address hypothesis 2. Um, I've just You
got it.

SPEAKER 3
I had 2 too. So he ran the first contingency table for just the interactions, not
interactions, um, between Uh, two of the variables, and then One for those
variables, just including females. I wish I could share it on my screen, but I
don't need. Do you, do you want to or do you want

SPEAKER 1
to talk me through it because um If you end up with two contingency tables for your
second hypothesis, that's not going to bode well for you because you only

SPEAKER 3
need. Yeah, but I think in the end. When I came out of my tutorial, I went through
the guide again because it's very quick in the tutorial. So when I came out, I ran
it again the way you said. And I ended up with just one. But, yeah, during the
tutorial, I think he was just confused as to how to philtre. contingency table just
to show. Um, because the It was still showing up with the total sample rather than
the total of females. Does that make any sense. All right, let's just do a separate
table just for females to show their total.

SPEAKER 1
So I think I know what he's done. I'm gonna I think I, I'm gonna try and replicate
what he's done, because I think I know. I think he started with the full sample,
and then he ran a descriptives. With gender. Or he could have done it as a
contingency table as well, but this will get you there as well. And then he said,
make this a frequency table. So when you do that, you see that there are 543
females.

SPEAKER 3
Yeah, he also had the, uh, variables as well there, the relationship, and yeah, I
don't know, which I think
SPEAKER 1
you can also do that here.

SPEAKER 3
Yes, that looks like what it is.

SPEAKER 1
Yeah. The issue with doing it that way, just give that a second to catch up. It
will catch up in a second.

SPEAKER 3
Yes, that, that looks more like it, yeah.

SPEAKER 1
Uh, The only, the issue with doing it that way is you're creating a massive table
with lots and lots of contingency cells. If I make this simpler for us and I turn
off the expected and pier and I've only got one number in each cell. Still catching
up with me.

SPEAKER 3
Yeah, so down the bottom there you've got the 2nd contingency table. Now it's got
the categories. So that's what I had. Yeah, yeah.

SPEAKER 1
Yeah, um, I've thrown it out anyway.

SPEAKER 3
I'm like, yeah, it's, it does get you there, but

SPEAKER 1
it's actually working off a different contingency table so it risks Sea Square
being. Yeah, wrong. I don't think it would be here. I would, I recommend going this
way and turning, excluding these people. I'm just going to do this so that you can
see it and check that this is what you have as well.

SPEAKER 3
Sorry to take this on a tangent to everybody else.

SPEAKER 1
No, that's, that's OK. It's all fun to play with, it's all good. Do do do. And if
we leave those two variables in, then you have your 543. And you've got your chi
square, and you know that that chi square is referring to only these 543 people and
only these variables, so.

SPEAKER 3
Yeah, yeah, cool. Thank you for taking that.

SPEAKER 1
That's right, no worries, no worries. Right, we'll come back to our questions. Um,
do we write our classical results from Jasp into the word descriptions in the
results section? We've actually already answered that at the start for you and you.
That was, you calculate it all in Jas, but then you need to write it up yourself
in, in Word, and I would use the advice in lecture 7 as a guide. The only caveat
with lecture 7 is my table at the end is not APA formatted, um, but that's OK.
There's other places to get APA formatted tables. Uh, do we need to mention the
observed counts for fewer or greater than expected, even if the standardised
residual is not significant. I'm gonna jump in again and and point whoever said
this to lecture 7. Oh my God, you like 7. I. Preempted this question. Do I include
my results if the cell that reflects my hypothesis has a non-significant
standardised residual? Yes, report it. And the reason why you'd report it. Is
because remember, this is an assessment. If this was a paper that you're
publishing, you probably wouldn't report non-significant findings, but in an
assessment, you're demonstrating to your reader that you know how to read the table
and that you know how to interpret the analysis. So what we're looking for when
we're marking is, have they told us the Chi square statistics? Have they
interpreted them correctly? Have they found the right standardised residual, and
have they interpreted that correctly? So if your standardised residual for the cell
that you're interested in is not significant, you still need to report that to us
in your lab report. I hope that was helpful. I probably explained that a little bit
better in the lecture. Again, feel very welcome to unmute yourself. Thank you,
whoever wrote yes, that was really helpful. And then that's lecture 7, gorgeous,
thank you. Um, so, 213 cases were excluded. Is that a limitation and doesn't affect
the sample size of people over 25? Part of where this is coming from is there seems
to be a very common second hypothesis that is the same two variables, identity and
friendship quality, but instead of looking at people under 25, they're looking at
people over 25. And when you do that, you have 58 people. But the more exciting
thing is, you don't have anyone with achievement or moratorium. So, of all of our
58 people over the age of 25, none of them answered the questions in a way that
indicates achievement or moratorium, which is really interesting, really
unexpected. Um, so some people are interpreting that as, well, maybe that's just
because our sample's not big enough, maybe we don't have enough people. Um.
Unfortunately, that's not correct. 58 people is definitely enough people within a
chi square table to find a relationship. Uh, and the reason we know, if I create,
Maybe I'll do it off the one in just that we've got now. In fact, I'll just do it
if you don't mind, if you can bear with me for a second. I'm gonna hide everyone.
That he's under the age of 25. In fact, if this was your question, or if you want
me to look at a different relationship with the over 25s, let me know, and I'll use
those. Variables instead, otherwise I'm going to use identity and friendship
equality. I'll just come back to gender and turn on. All the other variables, all
the other categories, my mistake. Um, And I wait for my mouth to unfreeze. Mm.
Friendship quality and identity status and let this light up for a second.

SPEAKER 3
The statistics.

SPEAKER 1
OK, so here's our 58 people. The only expectation, there's lots of assumptions we
got in all of our stats, right, like we can't run um correlations where we've got
categorical data. There are certain assumptions that need to be met with Kai
Square. Most of them are very easy to meet, and we've met most of them
automatically. For example, one of them is that our categories cannot be dependent
on each other, um, meaning that, uh, a person, uh, actually, the yeah, the
categories can't be joined together. So everyone in our sample has to have their
own label of, let's say, conflicting friendship, mixed friendship, or supporting
friendship. They can't have both a conflicting and a and a mixed relationship, they
can only have one of these labels, that's one of the assumptions. The second
assumption is that the variables have to be independent from each other. In other
words, the way that people answered about their identity status had to be a
different set of questions from the way that they answered about their friendship
quality. So there's no, um, built-in bias into the survey. So that was met. The one
that's tricky is this one around expected counts, which is that the the the
expected counts within the table can be lower than 5, but it can only be lower than
5 for 25% of the cells. So in this case, we have 6 cells, right? 25% of 6 is 1.5,
which means we need, No fewer than 1.5 cells, let's call it 1 cell, to have an
expected count lower than 5. If more if more cells have an expected count um under
5, then we don't have enough power to pick up a chi square. What that means for
sample size is we can say, well, If 25% of our cells can be under 5, that means 25,
75% of our cells can be over 5. I did work up an example to show you, which is a
bit easier, I'm gonna show you that, um. Do excuse me, there we go. So, in this
imaginary table, we've got 4 contingency cells, it's a little bit easier to work
with than 9. We know that we want that. 25% of 4 is 1, so only one of our cells can
have a number lower than 5. Everything else can be 5 and higher, and if that's the
case, we're good to go. Which means the minimum number of people we need is 5
people in 3 of our 4 cells. Right, so in a 2x2 contingency table, the minimum
number of people we need is 5 people. If we add an extra category here. Petre. Now
we've got A 3 by 2 table. Um, so Let's, let's keep our numbers easy, shall we?
Let's go 4 by 2. So I don't like doing 25% of 6. We've got a 2 by 4 contingency
table. So we've got 8 cells in the middle here. 25% of 8, a quarter of 8 is 2. So 2
of our cells can have a count lower than 5. Every other cell has to have an
expected count higher than 5. Following on that. It's easier with a table, right,
than trying to just verbalise it. That means that our total minimum number that we
could have in a table this big is 5, 1015, 20, 25, 30. If we've got 30 people or
more, we're good to go with Kai Square. So When we're looking at this table, which
is 2 rows by 3 columns, so we've got 6 cells there. We know we're going to need.
Um, 20 75% of those cells to have over 5 people in them. Which I think works out to
somewhere around about 40. But I'm just, my maths isn't with me right now cause my
brain's a little bit tired. Um, but the number that we need is definitely fewer
than 58, so we know we've got enough people in this table to have power. That's not
the problem. The problem Which I think is pretty interesting. I if I turn on the
expected counts. Now, if you, granted you haven't seen Lake for 11, Andrea, I'm
conscious of that, but when you watch it, I talk about what the expected counts
mean, but they basically mean if we've got, um, Uneven or unbalanced variables, so
we've got heaps of people in one category and not many in another, that's going to
affect how many people are in our columns and our rows. So our expected count is
telling us, if these two variables are not related at all, Then the numbers in our
contingency cells are just going to be a consequence of the number of people in the
row and in the column. Like, there's not many people in this row, so I wouldn't
expect many people in the cells of that row. There's heaps of people in this row,
so I'd expect these numbers to be higher. The reason I'm pointing all that out is
that the expected counts are almost exactly the same as the observed counts here.
So there is just no relationship between identity status and friendship quality
for, The people over the age of 25, that's what this is saying. There's just that
any variation that we're seeing is purely because of the number of people in the
cell. We're expecting 9 people here, we've got 8. We're expecting 10.5 people here,
we've got 10. We're expecting 9.8 people here, we've got 9. It's only because of
the size of the cells. To show you what I mean by that, I'm gonna go back to the
one that was significant and show you the difference between The observe and
expected account there, bear with me. I promise this is worth it. This is gonna be
an aha moment for you. OK, we're switching it around, we've got just the under 25s.
Bring it back over. Give it a second to catch up. OK, we're back to our 621. Got a
significant chi square. Now look at the differences between the observed and
expected counts. Right, they're much bigger. So the fact that they're bigger, that
there's more variation in the table tells you there's a relationship between these
two variables. So that's what's going on with that one. Whoever wrote, um, it's the
58 people, just a size issue. It's not a sample size issue. It is literally that
over the age of 25, these two things are not related to each other, which to me
makes this relationship all the more interesting. under 25, we do see a
relationship. But that's me, and Yuni's yawning, so I'm gonna move on to the next
question.

SPEAKER 3
Sorry, Abby. Yeah, I, I feel like I'm talking the whole time. Um. With the
assumptions that I and then like, Post-op tests, do we have to mention that we ran
any of those in our results?

SPEAKER 1
No, no, in a way, the standardised residual is a post hoc test, but you don't have
to treat it as that because we've built it into the hypothesis. Great. Good
question. Thank you. Um, so, is it correct that standardised residuals do not imply
an association between variables and instead relate to the difference between
expected and observed counts? I'm just gonna say yes to this, that is correct. I
can give you a whole other rant on why it isn't a relationship between Variables,
but I'll save you from that because I've just tired everyone out talking about
expected and observed counts. But yes, if you want more on that, lecture seven goes
into standardised residuals and how they're a measure of the difference between the
observed and expected count. Um, and how many combinations from the hypothesis one
do we need to show all the results or all the combinations? Um, I apologise for the
grammar there, I was trying to convert it into a question from the way the person
had worded it. Where this is coming from, and I think I'm saying this more for
people watching the video after the fact than for the people on the call, but I'm
gonna do this anyway. There is sometimes confusion around, if I show you this one
here, if, if I'm only predicting the, the difference between observed and expected
counts for, say, this top sell here, achievement, our people with identity
achievement, people with conflicting relationships. The question that gets asked
is, do I need to put the whole contingency table in? And the answer is yes. Because
if we philtre down to only the people with achievement status and only the people
with a conflicting friendship quality, we'll end up with a contingency table with
one cell, and it will have 19 people in it. And they'll all be in the same space,
and we won't be able to talk about variation between our variables, because we
won't have variables as such. So we report the entire contingency table, show it.
Show it all the categories that are relevant for your samples. So if you're doing
it over 25s, you won't have achievement moratorium, but you will have the other two
labels. So just put the whole table in for all the people in your interested group.
The chi square tells you the variation for that specific table. And then the
standardised residuals tell you the difference between observed and expected count
for your sale of interest. So you don't need to give a prediction for every cell
there, we just want to see a prediction for one cell. Um, I hope that was helpful
for whoever's watching this at home who needed me to ask and answer that question.
Um, we have 2 minutes left. I feel like I've been talking the whole time. Is there
anything that has been asked in these questions that you'd like answered or
anything else that's occurred to you that you want to talk through, that you want
to show me?

SPEAKER 2
So I'm, I'm just confused of the last sentence. I said no prediction for every
cell, just the ones you're measuring. What does that mean like prediction?

SPEAKER 1
Um, it means if you go back to I think it's lab one. I'm trying to think where the
best place to show you in the hypothesis itself. Oh, I know where the lab report,
the guide to Kaine. Class square, we'll show you, um. The hypothesis itself is made
up of two sentences. The first sentence confirms for the reader, oh, that's the,
that's the actual. Results, that's OK. Um, it doesn't matter. The first sentence
confirms for the reader, the sample that you're focusing on. And the two variables
of interest. So among participants under the age of 25, I predict there'll be a
significant relationship between identity status and friendship quality. That's the
first sentence. The second sentence is actually choosing one of these cells in this
table and saying more or fewer people will be in this cell than expected just at
chance level. So for example, more, I predict that more people with achievement
status will have conflicting friendships than expected by chance.

SPEAKER 2
Oh, that's actually like in the template for the hypothesis, right?

SPEAKER 1
Exactly, exactly. So so that second sentence is predicting one contingency cell in
this whole contingency table. It's saying there'll be a higher observed count than
the expected count or there'll be a lower observed count than the expected count.
Yeah, thank you. That's OK. Any other final questions? I've lost my little chat box
that I had. Apologies if you've been putting anything in the chat. Uh there's one
in the chat. Oh, that's all right. All right, well, if there's no other questions,
I'm gonna stop there and I'll stop recording. Uh, thank you so much for asking your
questions, though, and for setting up. And framing this workshop, I hope it was
you.