00:00:12
validating higher order constructs in
00:00:14
this session we're going to talk about
00:00:16
how to validate higher order
00:00:18
constructs before we do that let's
00:00:21
quickly recap as to what we are doing
00:00:24
now here is the model that we want to
00:00:26
test now as part of step one we are
00:00:29
assessing our measurement model based on
00:00:32
measurement model since we've got higher
00:00:34
order constructs and lower order
00:00:36
constructs
00:00:37
here these six are your lower order
00:00:40
constructs and this internal marketing
00:00:42
is higher order construct with these
00:00:45
three and these four lower order
00:00:47
constructs now the first step is you
00:00:50
assess all the lower order constructs
00:00:52
and this is what we did here we designed
00:00:56
this model with all lower order
00:00:58
constructs no matter if they had a
00:01:00
higher order construct or not all lower
00:01:03
order constructs were part of this
00:01:06
analysis
00:01:09
PLS Sam algorithm was run
00:01:13
and Factor loadings reliability and
00:01:16
validity was assessed this was done in
00:01:19
previous sessions the video will be
00:01:21
shared in the
00:01:22
description now moving on since we've
00:01:25
got two higher order constructs here
00:01:27
internal marketing and internal service
00:01:29
quality now once lower order constructs
00:01:31
are validated the next step is to
00:01:33
validate the higher order constructs in
00:01:35
this case I'm taking internal
00:01:37
marketing as reflective formative and
00:01:41
internal service quality as reflective
00:01:43
reflective higher order construct now
00:01:45
what do we mean by reflective formative
00:01:47
and reflective reflective and what is a
00:01:51
higher order construct now we did
00:01:52
discuss the difference between
00:01:54
reflective formative constructs in the
00:01:56
last session but we did not discuss them
00:01:59
at higher level in this session we're
00:02:02
going to talk about higher order
00:02:04
constructs and how do we validate
00:02:06
reflective reflective higher order
00:02:11
construct validating higher order
00:02:13
constructs now the concept of
00:02:15
hierarchical component model or higher
00:02:17
order constructs higher order constructs
00:02:20
also known as heroical component models
00:02:22
in context of plsm provide a framework
00:02:25
for researchers to Model A construct on
00:02:28
one or more abstract dimension that is
00:02:30
your higher order construct and it's
00:02:32
more concrete subdimension so in higher
00:02:35
order construct you've got a construct
00:02:38
at the second level or at a higher level
00:02:41
and then that construct has got 2 3 4 or
00:02:44
whatever number of
00:02:46
dimensions in this study we've got
00:02:48
internal marketing that is a higher
00:02:50
order construct with these three
00:02:51
dimensions and each Dimension is
00:02:54
measured using multiple
00:02:56
indicators now this is a lower order
00:02:59
construct
00:03:01
this construct is directly measured
00:03:03
using different indicators same as for
00:03:05
this same as for this same as for this
00:03:07
and same as for this but this construct
00:03:10
here is a higher order construct with
00:03:12
these dimensions and each Dimension is
00:03:14
measured using different indicators same
00:03:16
is the case with internal service
00:03:18
Quality Moving
00:03:21
On HCM refers to more General construct
00:03:24
that is measured at a higher level of
00:03:26
abstraction while simultaneously
00:03:28
assessing several subcomponent that is
00:03:30
dimension as is the case with internal
00:03:32
service quality and internal Marketing
00:03:34
in this
00:03:35
example hence by specifying lower order
00:03:38
construct SCCM can cover concrete rates
00:03:41
of more General conceptual variable of
00:03:44
interest moving on these are the
00:03:46
different types of higher order
00:03:48
construct now in this example I've got
00:03:50
reflective reflective and reflective
00:03:51
formative now what is reflective
00:03:54
reflective now we did assess or
00:03:57
understand the reflective construct in
00:03:58
the last session and the arrows are
00:04:01
pointing towards the indicators from
00:04:04
this latent variable this latent
00:04:06
variable and this latent variable but
00:04:08
have a look at this higher order
00:04:09
construct now these three lower order
00:04:12
constructs are reflected in this higher
00:04:13
order construct and the arrows are
00:04:15
pointing towards the lower order
00:04:17
constructs now this is reflective here
00:04:21
and reflective here just as internal
00:04:23
service quality in this case the
00:04:26
properties that we
00:04:28
identify for reflective constructs in
00:04:30
the last session still stand at a higher
00:04:32
level now these lower order constructs
00:04:35
in this case are interchangeable and
00:04:37
even if you remove one of them you're
00:04:39
still your construct will still be there
00:04:42
it won't lose its
00:04:44
identity reflective formative now at the
00:04:48
first level they are reflective that is
00:04:49
the arrows are pointing towards the
00:04:51
indicators at the higher level they are
00:04:53
formative because these lower order
00:04:55
constructs form the higher order
00:04:58
construct in this case internal
00:05:00
marketing now Vision development rewards
00:05:03
vision is reflective at to level
00:05:06
development is reflective at lower level
00:05:08
and rewards is reflective at lower level
00:05:11
now these three constructs form higher
00:05:13
order construct of internal
00:05:16
marketing if you remove any one of them
00:05:19
then there is no concept of internal
00:05:21
Marketing in this case we have got two
00:05:25
constructs internal service quality
00:05:28
reflective reflective that is reflective
00:05:30
at the first order and reflective at the
00:05:32
second level reflective formative that
00:05:35
is reflective at the lower level and
00:05:37
formative at the higher level with the
00:05:39
arrows pointing towards the construct in
00:05:41
this particular session I'm going to
00:05:44
guide you on how to validate a
00:05:46
reflective reflective construct that is
00:05:48
your internal service
00:05:51
quality the procedure is the same as we
00:05:53
follow it for reflective lower order
00:05:55
construct procedure doesn't change you
00:05:57
have to assess the reliability and
00:05:59
validity
00:06:00
however for higher order construct we
00:06:02
are going to use this joint two-stage
00:06:04
approach now once you have validated the
00:06:06
lower order constructs we are going to
00:06:08
generate scores for lower order
00:06:11
constructs there are other approaches
00:06:14
like embedded two stage approach however
00:06:16
the preferred and most used is disjoint
00:06:19
two-stage approach and the steps are
00:06:22
step one you have to generate latent
00:06:24
variable
00:06:25
scores and you have to copy them to a
00:06:28
data set that is your data set that you
00:06:30
are using in your study then import that
00:06:32
data set and then your loc's will now
00:06:36
serve as indicators for higher order
00:06:38
construct how now let's have a
00:06:42
look here is our model now the first
00:06:45
step we have to generate latent variable
00:06:47
score now each of these dimension of
00:06:49
internal service quality is measured
00:06:51
using these number of
00:06:54
indicators
00:06:56
here now these are actually the
00:06:59
indicator
00:07:00
of internal service quality
00:07:03
reliability Assurance empathy and
00:07:06
responsiveness now these are the
00:07:08
indicators of internal service quality
00:07:10
as mentioned in the model
00:07:14
here so in internal service quality has
00:07:17
got these
00:07:19
indicators but if we assess the
00:07:21
construct at lower level this construct
00:07:23
is measured using three or four
00:07:25
indicators so I need to change these
00:07:28
scores
00:07:30
that is measured using multiple items
00:07:32
into a single score so that they can
00:07:34
serve as indicators like this here so
00:07:37
how do we do this to do so we have to go
00:07:40
to
00:07:42
reports if you haven't run the model yet
00:07:44
just go to run go to calculate PLS
00:07:48
algorithm all good start and your
00:07:51
results are here now let's go to reports
00:07:54
and we need our latent variable score
00:07:57
report and where is your latent variable
00:08:00
score if you look at final results here
00:08:03
this is your latent variable scores just
00:08:05
click scores and here it is now you can
00:08:09
copy it export to CSV it's copied now
00:08:12
where is your data file it's always a
00:08:13
good idea to put it in the same data
00:08:15
file so here is my data file now I'll
00:08:18
put it at the end here have a
00:08:21
look
00:08:23
here you can delete this case ID now
00:08:27
it's the same number here
00:08:30
341 in
00:08:33
total here it is 340 because it starts
00:08:36
with zero so 341 rows the same number of
00:08:40
respondent that you had in your original
00:08:42
data now what happens
00:08:44
is your score for your lower order
00:08:47
construct that is measured using
00:08:49
multiple indicators is transformed into
00:08:51
a single item because now these items
00:08:55
are measuring here reliability Assurance
00:08:56
empathy empathy and responsiveness now
00:08:59
previously at the lower order level
00:09:01
liability was measured using four or
00:09:04
five items now those five four or five
00:09:07
items now are combined into a single
00:09:09
latent variable score now these four
00:09:12
will serve as indicators for the higher
00:09:15
order
00:09:16
construct let's save it and now let's
00:09:19
import it again now go back back yes we
00:09:24
want to save
00:09:25
it right click import data file
00:09:30
so it's saved let's close it here it is
00:09:36
import go back and here it is Data one
00:09:41
now let's look at our measurement model
00:09:44
again now we've got multiple data sets
00:09:46
so you have to select the data set that
00:09:48
you want and have a look here now Vision
00:09:52
that was previously like this now these
00:09:55
four indicators are used to generate
00:09:58
your latent variables source and here is
00:10:01
the latent variable score for each of
00:10:04
these factors or
00:10:06
Dimensions now since internal service
00:10:09
quality is measured using these four
00:10:10
dimensions and now these four dimension
00:10:12
will serve as indicators for internal
00:10:16
service quality that is measured at the
00:10:17
higher level how to do this let's go
00:10:21
back yes and let's duplicate
00:10:25
it copy resource
00:10:29
and right
00:10:31
click paste resource so let's say it's
00:10:33
the copy of measurement
00:10:37
model
00:10:39
and let's say it is reflective for
00:10:42
internal service
00:10:43
quality now
00:10:46
Save open
00:10:48
it and now I do not need
00:10:51
these
00:10:53
yeah yeah all of this is
00:10:56
deleted
00:10:57
now let's add
00:11:01
reliability Assurance empathy and
00:11:03
responsiveness all of them selected and
00:11:06
dropped here and this is your internal
00:11:09
service quality press enter now look at
00:11:12
this the four indicators for internal
00:11:16
service quality as in the
00:11:18
model here and you got these indicators
00:11:22
by generating latent variable score but
00:11:25
look at the arrows pointing towards the
00:11:26
indicator so this is reflective
00:11:28
reflective
00:11:30
now what about these this is also a
00:11:33
higher order construct so Vision
00:11:35
development and vs make up internal
00:11:36
marketing but this is reflective
00:11:38
formative we are going to discuss how to
00:11:41
validate a reflective formative scale in
00:11:43
coming videos now since it's red you
00:11:46
cannot run the model so let's link it so
00:11:49
internal service quality is linked to
00:11:50
the dependent variable and it is linked
00:11:53
to
00:11:55
your IV as well now that your model is
00:11:58
complete complete this is reflective
00:12:00
reflective all other lower order
00:12:02
constructs will remain the
00:12:05
same this is called disjoint two stage
00:12:08
approach because you do not use the
00:12:11
latent variable score for the other
00:12:14
latent construct you use them at
00:12:17
indicator level so you add the
00:12:19
indicators for each of the construct
00:12:22
just as we have done just we are going
00:12:24
to use the latent variable score for
00:12:27
these subdimensions of internal SW
00:12:29
service quality why because internal
00:12:32
service quality is reflective reflective
00:12:34
higher order construct and now we are
00:12:35
assessing internal service quality at a
00:12:38
higher level because we are assessing
00:12:40
the validity or the reliability that is
00:12:43
measurement model assessment for
00:12:45
reflective reflective higher order
00:12:47
construct now again calculate PLS
00:12:50
algorithm
00:12:52
[Music]
00:12:53
start
00:12:56
now here is your output this is your
00:12:59
graphic output if you look here let me
00:13:01
increase the size a bit
00:13:05
here look at
00:13:07
this RQ looks good Alpha looks good
00:13:12
reliability looks good A looks really
00:13:14
good now look at the report now look at
00:13:16
the outer loadings first for your
00:13:19
construct here look at the outer
00:13:20
loadings all good let's improve the zoom
00:13:23
here all good now look at the
00:13:26
reliability and validity
00:13:29
look at the internal service quality all
00:13:32
good yes now you can similarly look at
00:13:36
your discriminant
00:13:38
validity again for internal service
00:13:40
quality it's all good same is the case
00:13:43
for for and Loa all good this one is
00:13:46
higher and all these are
00:13:48
low So within construct variance for
00:13:53
internal service quality is higher than
00:13:55
its shared
00:13:57
variance now you see we do not change
00:14:01
the way we assess the reflective
00:14:03
reflective higher order construct we
00:14:05
assess it the same way we did it for
00:14:07
lower order reflective constructs and we
00:14:10
assess the same output we assess this
00:14:14
the outer loadings we assess the
00:14:15
reliability we assess the validity just
00:14:17
as we did for the lower order constructs
00:14:19
you do it for the higher order
00:14:21
reflective reflective construct as well
00:14:24
I hope the session would have helped you
00:14:26
understand the disjoint two-stage
00:14:28
approach that we can use to assess the
00:14:31
measurement model that is reflective
00:14:34
reflective higher order construct thank
00:14:37
you very much
![William Shakespeare — Biography by A&E [HIGH QUALITY]](https://ruhnrawpgxrzdrgaohnf.supabase.co/storage/v1/object/public/images/video/ITc13WnBWX4/thumbnail.jpg)
