SPSS for newbies - semantic differential scale data entry and analysis Video Lecture | SPSS: For Beginners - Data & Analytics

today's question is from Massimo hi Phil
can you explain to me how to analyze
semantic differential scales surely man
she is a likert one I can't find any int
or real about this it's five years since
I posted my first data entry video in
SPSS and it's amazing I still get
questions from you guys about how to
enter the data if you look at the
popular uploads most of it on my channel
are about entering data in SPSS less
questions are actually about doing the
statistics for those of you actually get
beyond the hurdle of entering your data
notice that I have now organized my
organize my videos SPSS by like sections
now these chapters so depending on the
stage you are at your new analysis you
can look under these headings for what
you need for methodology also notice
that I have a series of videos at bottom
about topics or regression for student
projects okay so just a plug there ok
semantic differential scales how you
handle this is being pretty much treated
in my videos but let's just kind of just
talk about semantic differential scales
very similar to Likert scales as you
said but the answers is like it's an
adjective and it's on a bipolar scale so
you've got one you got the non extreme
so for happiness you have unhappy versus
happy at the other extreme and the scale
can be like is up to you one two four
one two five one two then weather here
is seven one two seven as with Likert
scales you may there's a case for like
treating them as ordinal and interval so
really they're ordinal so let's just
recap the kind of main methods open to
you if you are doing newbie and you're
doing this analysis for ordinal four
measures for summary stats you're
looking to report the median and
interquartile range for measures of
location and dispersion if you're
looking to report correlation you'd be
looking you'd be reporting this
Spearman's Rho if you're looking to
analyze the pairwise relationships and
the strength in terms of like odds ratio
you bill
a cross tabulation analysis or test of
chi-square test of Independence if you
do no more kind of complex analysis
involving more random variables you
would be running some kind of a
regression now if your semantic scale is
going to be treated as a dependent
variable you need ordinal regression and
that's something more advanced courses
do I cover it
that's like usual and master's-level on
the other hand other otherwise yeah you
you just be run your usual regression
fact analysis is possible for testing
ordinal
you running nonparametric tests right on
the other side if you're going to treat
them as interval then for summer stats
you're looking to report the mean and
standard deviation for location and
spread correlation you'd be doing the
Spearman's Rho regression you're running
the usual regression test you'll be
running the parametric tests like t-test
f tests let's talk about entering it as
if it were interval data that's the
easiest here because let's just bring a
purse BSS say I've got this variable
called HP wife are happy okay this is in
variable view I leave everything as it
is pretty much I could set the decimal
place to 0 because I'm my entry value is
going to be whole numbers from 1 to 7
scale I'm going to measures I'm gonna
set to scale 4 so it's not categorical
and then I just start entering my
numbers okay first person along this row
entry for the first person might have
said that's to unhappiness scale and 6
for the other and so on so down this
column you can have the numbers running
from 1 to 7 right after reading around a
bit you might see that some software and
some authors use some other numbers like
what 0 1 2 3 - 1 - 2 - 3 the idea being
that we know when you calculate the
average across it's the 0 that means a
lot in between you are the more negative
you are towards minus 3 the more unhappy
you are more positive 3 the more happy
you are
okay yeah you can do that instead of on
the scale of one to seven but I have to
think about it and unless any of you
have any other ideas I think this one
one two seven is a better one in terms
of flexibility if you're going to do a
regression if you're doing other things
other than just summary statistics
things to think about here is okay
reverse coding so some of your questions
you might run a reverse code so you want
to get them all pointing in the same way
when you're doing your data entry and I
have a video under how you do the first
coding so it's there okay next suppose
that we're going to treat as ordinal so
it's like categorical now then if you
think about two Likert scales were had
like on a scale of you should say
something like very very unhappy and
happy it's different skills right but
because they're so refined here this
seven I'll just keep them as numbers
because otherwise you're gonna have to
find a word for this one which would be
very very very unhappy very very unhappy
very unhappy never happy or not happy
happy very happy very very happy which
is a bit long-winded and which case all
I would change here in the liquids if
you've treated all no sorry is I would
not complete this as I did for the
Likert scale but over here for the
measurement you can just remind yourself
that it is all know by going to
measurement and ordinal remember this
really only matters if you're going to
use a chart builder but other than that
it's comforting for some people have to
remind themselves that they've got
something which is ordinal or scale tell
us it reminds them what skill level of
measurement that the variable is if
you're going to treat it as categorical
you might find depending on the number
of respondents in each of these cells 7
cells you might need to combine groups
because you might find these are too
fine and some of these categories might
have hardly anyone check ticking on them
like 7 here you might very few people
ticking on it in which case you'd be
thinking about combining groups and and
I have a video on how to combine groups
as well ok so I think that's that's
pretty much it so go and explore my
videos on how to analyze semantic
differential scales