live_tv
Livestream Starting Soon
00
Hours
:
00
Minutes
:
00
Seconds
Up next in 10
Comparing two free statistics software packages, Jamovi and JASP, for ease of use, features, and usability in some common tests
Show More Show Less View Video Transcript
0:00
so hi my name is Dave uh I'm going to
0:03
guide you through this little tour of
0:05
jamovi and jasp where we compare the two
0:08
of them it's pretty quick skip any part
0:10
that you don't feel is important to you
0:13
or is necessary my background is mostly
0:15
in organizational change and
0:18
organizational research but I also teach
0:21
statistics as part of other courses so
0:24
let's get
0:28
started so here we are and you can see
0:31
that there are some things in Gray these
0:33
are the things that the SPSS file we
0:35
imported already had labeled as missing
0:37
values you can make everything bigger or
0:40
smaller you can change the color palette
0:42
and such which is nice and you can add
0:45
new modules which is nice so you see
0:48
these are modules that are already here
0:51
and then there's other ones that we can
0:52
add in these are bundled so we can go
0:55
out to the library and we can add a few
0:58
things so for example we can add an
1:00
Editor to run R code inside Jovi which
1:02
then gives you syntax let's just get out
1:05
of here and go back in it is really nice
1:07
to have the missing values here in the
1:09
data editor showing up as gray SPSS
1:11
doesn't do that at least the versions
1:13
I've used don't and in fact it's the
1:15
first time I've ever seen it like this
1:17
now if you want to see which values are
1:20
designated as missing you double click
1:22
on the variable name and here is the
1:25
variable description or variable label
1:28
as SPSS calls it and that's that's
1:30
imported directly from SPSS of course
1:32
you can set it here or change it here
1:34
yourself if you want to along with what
1:36
kind of measurement it is so this is
1:38
nominal which means that the numbers
1:40
aren't really
1:42
numbers and that makes sense because
1:44
it's basically a yes or no question here
1:47
uh should people be able to use health
1:48
insurance for abortion yes or no and
1:52
then you've got three things that are
1:53
labeled is missing as you see here 0 8
1:55
and N 0 is inapplicable eight is don't
1:59
know nine is no answer so these are all
2:01
uh missing values and you can click on
2:04
any of these once you've double clicked
2:06
in here and you see the same results
2:08
let's take a look now at uh transforming
2:11
and recoding basically recoding is here
2:14
under transform you might notice by the
2:17
way I should point out that it's showing
2:19
you value labels so the actual values
2:23
here are own numbers right now where it
2:26
says people should be able let's see
2:29
people should not be able is what it
2:31
says so that would actually be the
2:32
number two here so let's take a look at
2:36
this one here do you have a moral abort
2:39
uh a moral opposition to abortion if you
2:42
take a look then you find that uh one is
2:46
opposed two is not opposed three is it
2:48
depends normally we would want it to go
2:51
1 32 basically because we'd want it
2:53
depends to be in the middle so if you're
2:56
opposed you're on one side if you're not
2:57
opposed you're on the other and if you
2:59
say it depends you're in the middle so
3:01
really we have to swap this two and this
3:03
three well how do we do that is we
3:05
transform so how do we take this one and
3:09
three and swap them is we can either
3:12
compute a new variable with everything
3:15
set up correctly that turns out not to
3:18
actually work very well here or we
3:21
transform and we say that we will use a
3:24
new
3:26
transformation and so we add a recode
3:30
condition that if the source equals
3:33
notice the two equal signs two then we
3:37
use three we add a condition if it
3:40
equals three we use two now in SPSS if
3:45
you did this you would run into trouble
3:46
because it would convert all the twos to
3:48
threes and then all the threes including
3:51
the ones that used to be twos would go
3:53
to two but here you will notice here it
3:57
will actually work properly oh let's
3:59
also do a Source equals 1 we'll
4:01
basically copy those over and else we
4:04
will use zero we'll call it uh
4:09
swap two and
4:12
three and we will say that the
4:14
description is just
4:17
recoded to Three
4:20
swap and the suffix will be
4:24
fixed and uh we'll go back up and now we
4:28
should see that it is is fixed now of
4:30
course it's just showing the numbers now
4:33
if you take a look then you see that
4:35
where it says morally opposed it's one
4:37
just says it was where it says it
4:39
depends it's now two let's move this
4:42
over and not morally opposed is three so
4:45
it's now I apologize for doing the most
4:47
complicated thing nearly
4:49
first but recoding is of in handy so
4:53
let's say that we take a look at the
4:55
people who want to pay for abortion with
4:58
insurance or not and we will see how
5:00
they feel about abortion being moral
5:03
you'll notice that as I put things in
5:05
it's
5:06
already taking a look at the T tests
5:09
again grouping variable it has to be uh
5:12
one a binary variable a dichotomus
5:14
variable one where there's only two
5:17
possible values or else this will not
5:19
work you can't say like you can in SPSS
5:21
or pspp I'm going to put in this
5:24
variable of age groups and compare the
5:26
youngest and the oldest by specifying
5:29
that I'm going to to compare say groups
5:30
one with group four won't work uh I can
5:34
show you by will put AB moral in here
5:37
and it says must have exactly two levels
5:40
uh so here we see that there is a
5:42
violation of the Assumption of equal
5:44
variances SPSS will give you two sets of
5:47
results one where it assumes equal
5:50
variances and one where it doesn't here
5:52
you have to know to click on Welches or
5:54
man Whitney depending on the situation
5:57
I'm going to stick with Welches and you
5:59
see it's the same P value it's even the
6:02
same statistic different degrees of
6:04
freedom and you're not violating any uh
6:07
rules I always like to see the
6:09
descriptives so I can compare the means
6:12
by I and here they are do a quick look
6:15
at regressions so we'll do a linear
6:17
regression we'll try to predict uh
6:20
opposition or lack of opposition to
6:23
abortion with let's see adults in the
6:25
household age well that's an interesting
6:29
uh error that I've never seen so let's
6:30
take adults out and we'll just do age
6:33
let's see what cohort they're
6:36
in that's absolutely fascinating because
6:38
it was not doing this before when I was
6:41
testing it but it doesn't seem to be
6:43
giving me let's try
6:45
income all right it seems to like these
6:48
we'll just do age and income you might
6:50
notice that uh we don't have a choice of
6:54
what type of analysis to do it does
6:57
enter and only enter so let's do it with
7:00
two models we'll assume that age is the
7:02
most important and we'll add a new block
7:04
and we'll drag income in there and now
7:07
we see that we've got our two models it
7:11
doesn't label our two models which would
7:13
be nice but it does tell us that we can
7:16
explain about what is that 6% of
7:20
attitude towards abortion with age and
7:22
then another 0.1% or so with both age
7:27
and income and we can add in all sorts
7:30
of things like assumption checks of
7:32
various types uh we can check the
7:36
adjusted R squar and we can look at some
7:39
other things as well we can do the Inova
7:42
so it's basically most of what you get
7:44
with SPSS but in a somewhat different
7:47
way usually I've never seen an error
7:49
message in this before so it might be
7:52
something with this data set or it might
7:54
be a bug in the program now what about
7:56
copying the results to word you might
7:58
ask
8:00
well let's find out what that's
8:02
like copy
8:05
table and you see it pastes in as an
8:08
unnecessarily complicated table it's
8:10
still nice to do it this way because
8:12
when you do a correlation table it'll
8:14
come out quite
8:16
[Music]
8:18
nicely as you saw you can add in modules
8:22
and one of the libraries you can add in
8:24
lets you run our code it does run on R
8:29
and uh once that's
8:31
there if you know R code you can run it
8:35
I don't see any journaling which would
8:37
be nice and it would be very convenient
8:39
if the r code could be set up to print
8:42
out along with the analyses especially
8:45
if you're a faculty member trying to
8:46
figure out what your students were doing
8:49
but I don't see any way of doing that
8:51
there might be a way but I don't see it
8:53
so let's move on to jasp
9:00
[Music]
9:03
so here we are in jasp and I've just
9:05
uploaded the exact same file and you
9:07
might have noticed a couple of
9:08
differences one big difference is that
9:11
instead of having missing values grade
9:13
out they're completely blank again these
9:15
are all actually numbers in these
9:17
columns but it's showing you the value
9:19
labels so we can double click here we
9:22
get a very similar uh report where we
9:25
can put in custom values for the missing
9:28
values right now again missing values
9:31
are blank and it did that automatically
9:33
which you may or may not want usually I
9:36
have to admit I really don't care
9:38
because I don't you know they're missing
9:40
values I generally just leave them out
9:43
occasionally I find something else that
9:44
I want to make missing or if I'm
9:46
importing from Excel or tab delimited
9:49
format or something like that then I'll
9:51
want to set missing values myself so
9:54
let's take a quick look at recoding if
9:56
we can do that so let's try I go here to
10:00
data editing mode and you really don't
10:03
seem to be able to change data and
10:05
compute things you can compute with our
10:08
code uh and this is where I give up and
10:12
I say maybe you should do your recoding
10:14
somewhere else because this looks like
10:17
it's going to be incredibly hard to
10:21
do you might notice that there's now
10:24
basy in here to confuse your students as
10:26
well as classical uh let's go to
10:29
dependent samples we'll do the same
10:30
thing we did before grouping variable do
10:33
you think Insurance should be able to
10:34
pay now the problem here of course is
10:37
that if we use the morality of abortion
10:41
which is right here then we get
10:43
something that is probably accurate but
10:46
makes no sense so let's take a look at
10:50
age and we'll see what happens and again
10:53
as I put things in and out of these
10:56
windows the results come up instantly
10:58
both these program s are astoundingly
11:00
fast and there's no violation of the
11:04
that the variances are similar for both
11:07
groups so you don't get an error but if
11:09
we wanted to we could get the Welch test
11:12
just like that now let's go to
11:15
regression we'll do the same thing that
11:17
we did before we'll try to predict uh
11:20
let's see it's hard to predict morality
11:22
really we're going
11:23
to uh but it will be completely
11:25
meaningless because of the way it's
11:27
coded just remember that
11:29
so let's look at age and what else do we
11:33
look at before income there's income
11:37
16 and you see it immediately pops up on
11:40
the right it is absurdly fast let's try
11:43
something that might be more there we go
11:46
now we get two models you don't want to
11:48
use these methods in reality uh we're
11:50
just playing around right now I do have
11:52
other data sets that would be more
11:54
appropriate for this where I actually
11:55
know all the variables quite well here
11:57
we have an R squ of just about zero and
12:00
another one
12:02
of uh 2% which is really pretty small
12:06
trying to predict age using income and
12:09
ab from memory you notice the
12:11
description didn't come through it did
12:13
not import any of the
12:16
SPSS Variable labels which is a real
12:19
problem if you're importing SPSS files
12:21
it's not a real problem if you're
12:23
creating your own data files it's quite
12:25
nice really uh how fast this is and how
12:28
nicely form formed these results are it
12:31
gives you the Innova by default you can
12:33
shut that off if you want to or you can
12:35
add other things in as you see it's very
12:37
similar you can have the r squar change
12:40
I always think that should be in there
12:42
by default can add in descriptive so
12:45
that you can see how many people are in
12:46
each group and what the averages are
12:48
also handy to make sure you're not
12:50
making mistakes we have to uh copy and
12:53
paste to word so let's do
12:56
that and you see it's almost exactly the
12:59
same a lot of the underlying code for
13:02
jamovi and jasp is probably the same I
13:05
suspect that they are branches of the
13:07
same project and as time has go gone on
13:09
they've diverged more and more and more
13:11
they both use R
13:13
underneath and the sad thing is that
13:16
jasp is really although it's very very
13:19
fast and it's very convenient it also
13:22
takes up over two gigabytes of disk
13:24
space whereas Jovi only uses about let's
13:27
find out for sure there's this very nice
13:29
site code Mac stats. org where I write
13:32
down things that I've learned so here it
13:34
is under
13:36
free stats and you can see that I
13:38
compared these here we go so the
13:41
download for jasp is 1.3 gabt and for
13:44
jamovi it's 415
13:47
megabytes when they're on the
13:50
disc jasp ISS 2.2 GB and jamovi is 940
13:55
megabytes now one other nice thing is we
13:58
were talking about syntax and journaling
14:01
so your R console gives you the uh
14:05
syntax and here it is now so you can
14:08
type in R code and you might be
14:13
able to save things in an editor so
14:16
these are the standard missing values
14:18
across your entire data set you can
14:20
change these if you want to uh you can
14:24
use a spreadsheet Editor to do your
14:26
recoding that's their way around any
14:28
problems with Cod
14:29
you can change through you have more
14:31
control over the what the results look
14:33
like including showing the r syntax when
14:36
you do things and that both helps you to
14:37
learn and it helps you if you're having
14:41
your students do things then you know
14:42
what variables they were using very
14:44
easily and you can see exactly what they
14:47
were doing so here you see it happened
14:49
retroactively so if we want to do this T
14:52
Test in the console we can just paste it
14:55
right here in theory
15:01
ah okay and that is my summary these are
15:04
both very very good programs it's
15:06
amazing what you get for free and they
15:08
are somewhat different once you get
15:10
inside but the basic techniques are very
15:13
very similar from one to the other I
15:15
expect them to keep on diverging in some
15:17
ways and also to keep on getting better
15:19
but you see that it's been doing this
15:21
rather instantly and this data set here
15:24
is not small it's got over 2,000 uh
15:28
cases in it and an awful lot of
15:32
variables I go through and there's
15:34
hundreds of variables in this
15:37
thing so I wish you luck with your