inexpensive 'home' version?
Locked
123
123
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
 
|
Marta
Have you thought about joining their academic
author program? it won't help your university, but it might give you a
free licence, on condition that you don't use it for personal gain and send
copies of everything to SPSS. How do you think a retired guy on a
fractional pension like me manages to get access to SPSS? If they pull the
plug I'll be in cold turkey for ever!
John
|
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
|
Typed this right as John's email came through regarding the "academic author program":
I know that the SPSS/PSAW software package thrives on the licenses it sells, and that they share a nice piece of the education market (at least from the universities I have visited in recent years). However, IBM has been pretty smart in the past about letting educational entities have 'free' or 'discounted' equipment/software in the hope of training the next generation of users/buyers. Take for instance IBM Mainframe - from my understanding the full system can cost a pretty penny (as well as the upkeep etc) and from what I heard (maybe just rumor) but IBM equipped my university with a free one... why? Because, our university trains computer technicians to work on them. I think that this has been the reasoning/impetus behind the student/grad pack versions of SPSS, but even those can be relatively expensive (especially because some students are cajoled into the student version, then the grad version, and then the full version - and let's not forget the renewal fees); overall SPSS has been generous in some regards, but with the recent acclimation of SPSS into the IBM-fold, I hope it follows suit and they develop a stronger/cheaper and more competitive SPSS package for the education system. John, I'll be looking into it myself to get the details, but what exactly can you do with the academic author license/program (this is the first time I've heard of it)? i.e. who is it designed/made for? Does it include the full SPSS suite? Can you do research with it? etc. J. R. Carroll Grad. Student in Pre-Doc Psychology at CSUS Research Assistant for Just About Everyone. Email: [hidden email] -or- [hidden email] Phone: (916) 628-4204 2010/8/2 John F Hall <[hidden email]>
|
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
|
|
Justin
I was put on to this by George Argyrous (one of the
authors on the list of recommended SPSS textbooks on my site) when my 5-year licence for SPSS11
expired in 2006. Check it out on http://www.spss.com/media/collateral/academic-authors-program.pdf
and http://www.spss.com/vertical_markets/education/textbooks.htm
Basically I offered to review Julie Pallants'
SPSS Survival Manual for the Social Research Association (UK)
but when it arrived I discovered to my horror that it was for SPSS for Windows
and only used the GUI. I had only ever used a mainframe version in syntax
and only had a DOS based PC with WordStar4. My son lent me a PC with
Windows and Word: Major Lester (SPSS UK) wangled me an evaluation version
of the Windows version of SPSS 11 via SPSS France. I learned Windows, Word
and PowerPoint in double quick time and got the 5-year licence free from
SPSS France on the strength of my extensive critical review of
Pallant 2001 (there's a different review of her 2005 edition there as well):
long story, but it's all in my Old Dog, Old
Tricks presentation to ASSESS (European SPSS users) at York University in
2006.
Finding myself with a spare 5 years of SPSS 11, I
started to convert and update all the teaching materials from the
postgraduate (part-time evening) Survey Analysis Workshop I designed and taught
from 1976 to 1992 when, fed up with vindictive and malevolent senior management
and no longer willing to spend any more years banging my head against a brick
wall, I took early retirement. All my course materials needed to be
converted from WordStar4 to Word and from SPSS-X 3 for a Vax cluster to SPSS for
Windows on a PC. Hence the offer to do the Old Dog thing at York.
When licence renewal time came (it's every six months) SPSS couldn't or wouldn't
give me a renewal code foor 11, so I had to have 15. When that expired
after two or three cycles, they gave me 18, but luckily 15 is still maintained
so I managed to get a renewal code for that as well. Thus I find myself
with both 15 and 18, but although I have an extensive set of tutorials up on the
site now, I still have a lot to do.
The academic author program is for authors of
textbooks, but they seem happy to accept my on-line tutorials as well.
They're all syntax-based in preference to menus, but many of the examples are
also shown with the GUI as well (thus demonstrating the clear superiority and
speed of syntax). Moreover they're written with no maths and gentle
step by step progression, in research logic order, and have full colour
screenshots at each step: no textbooks do that yet. You can check the
sequence on the Summary
guide to SPSS/PASW 18 tutorials
The rest of the site is given over to material on
the main surveys I have been involved in over the years, especially the Quality
of Life in Britain series I did with Mark Abrams at the SSRC Survey
Unit, and includes various (not easily available if at all) working papers,
reports, questionnaires etc.
That's probably enough for now: nearly bed-time
here. I've been on the computer nearly all day and need to relax a
bit (ie fall asleep in front of TV leaving best part of a pint of English ale to
go cling-filmed in fridge overnight).
Best
John
----- Original Message -----
|
Interaction between SPSS v18 and ArcView 10 -- python issue
|
|
In reply to this post by SR Millis-3
Don't know if anyone on this list can help - but it is worth a try.
I've got SPSS v18 with the python plug in installed. Python 2.6 is the version that SPSS has installed I believe, in the c:\python26 subdirectory. ArcView v10 installs python 2.6.5 and be default places it in the 'python26' subdirectory. This SPSS does not like, and when running a command file that invokes python, I get a 'begin program' message and the processor just sits there. I've uninstalled the python 2.6.5 that ArcView installed and SPSS works with no problem. I then reinstalled ArcView and specified a different subdirectory for python (python265) but again have the same problem with SPSS. Is there something I can look at/for which anyone on this list might suggest? Thanks for any assistance! Dan Bibel Massachusetts State Police ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
|
|
In reply to this post by Marta Garcia-Granero
From: Marta García-Granero <[hidden email]> |
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
Administrator
|
For those who are in a university setting, the prices can be a lot lower than that. If your university is part of the GradPlan, you can get a perpetual license for Stata IC/11 for $179 (USD). For more info, see: http://www.stata.com/order/new/edu/gradplan.html
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
|
|
In reply to this post by Yves_Therriault
Yves
Je suis complètement d'accord. mais je crois que
c'est IBM qui s'en fichent, pas SPSS. Ils ont leurs actinnaires à soigner,
pas nous.
(I agree completely, but I think it's IBM who
don't care, not SPSS. They ahve their shareholders to think of,
not us.)
John
|
Re: inexpensive 'home' version? - a suggestion for SPSS-IBM
|
|
In reply to this post by Yves_Therriault
|
|
|
In reply to this post by Bibel, Daniel (POL)
Hi all,
I will like to ask if the following variable can be classified as a likert scale variable for 2-indepedent t-test or correlation usage. 1-Strongly Disagree 2-Disagree 3-Somewhat Disagree 4-Somewhat Agree 5-Agree 6-Strongly Agree If this variable is not considered as one, can I recode the somewhat disagree and somewhat agree into a group neutral for analysis? Thanks in advance. Regards Dorraj Oet |
|
|
Looks OK to me: it's ordinal, but we all treat this sort
of stuff as scale as well. Depends if you're looking fo structure in a lot
scales or comparing sample groups on the measure.
|
Re: 6 point variable
|
|
Hi!
I do not agree with John. Just because you encode the data it does not mean that you get "full benefits". If you Google on this, it is difficult to find a statistician who would recommend you to calculate mean or standard deviation of ordinal data. Tips: http://en.wikipedia.org/wiki/Level_of_measurement All the best Wilhelm (Wille) Landerholm +46-735-460000 Queue/STATB Making sense of data BOX 92 | 162 12 Vallingby | Sweden 2010/8/4 John F Hall <[hidden email]>
|
|
Administrator
|
In reply to this post by Jarrod Teo-2
A single item does not constitute a Likert scale, regardless of whether there is a "neutral" point in the middle. A Likert scale is the mean or sum of several Likert-type items. E.g., http://psychology.about.com/od/lindex/g/likert-scale.htm IMO, using parametric procedures on individual Likert-type items is far more likely to be problematic than is using them on proper Likert scales (i.e., sums or means of several such items). But having said that... In another recent thread, I mentioned a guideline I like to use: If it is fair and honest to use means and SDs descriptively, then parametric procedures should also work reasonably well. So if I was trying to decide whether I could use a t-test on your item, I'd probably start by plotting a pair of histograms, one for each group. If I saw two distributions that looked pretty similar, apart from a possible shift in location, then I'd feel a bit more comfortable with a t-test. But if things other than a location shift were leaping out at me (e.g., skew in opposite directions; unimodal vs bimodal), I'd be less comfortable with the t-test, and would look for an alternative that treated the variable as ordered categories. HTH.
--
Bruce Weaver bweaver@lakeheadu.ca http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." PLEASE NOTE THE FOLLOWING: 1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. 2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/). |
|
|
In reply to this post by Wilhelm Landerholm | Queue
Not to belabor the point but in some
respect whether one's variable is
nominal, ordinal, interval, ratio,
absolute or whatever is somewhat besides
the point. The more fundamental
question is:
What is the nature of the population
distribution from which this sample
has been drawn, if one is assuming this
is a sample from a population of
values?
One can always calculate a mean or a
standard deviation or any statistic
one wants on a group numbers but the
quetion is what does such a
statistic mean? Frederic Lord in
his classic 1953 article on the "statistical
treatment of football numbers" showed
that one can reach valid
conclusions about different samples of
numbers even when they appear
to be only nominal (for those unfamilar
with Lord's paper, the problem
is that a group of college freshman go
to teacher of statistics to determine
whether they are receiving football
jerseys with numbers systematically
lower than that of upper classmen --
the null hypothesis suggests that
if the numbers are assigned at random
than the mean of the freshman
jerseys should not be significantly
different from the mean jersey number
of the upper classmen; Tchebycheff's
inequality was used as the statistical
test and it showed that the freshman
mean was significantly lower).
The real problem, I think, is that many
people would in fact treat the
6 point scale as an interval scale and
then use a t-test or ANOVA or
multiple regression analysis that make
assumptions way beyond one's
data is interval (e.g., follow a normal
distribution). For purposes of
descriptive analysis, one should not
just get the mean but all statistics
that bear on the shape of the
distribution of values. Is the sample
symmetric around a central point (in
which case the mean, median, and
mode will all be similar in
value). If the distribution is skewed, why is
it skewed? Does the question that
one is asked causing respondents
to favor one side of the rating scale
over others? And so on.
Finally, what sorts of inferences
and/or conclusions does one want to
make. If one is simply describing
a sample of values with no interest in
a possible population from which it may
have been drawn, then the sample
is the population. Describe it
adquately. But if one is using the sample
to draw inferences about a population
from which it was drawn, then which
population is being used? What
characteristics of the population is one
interested in? How will one
confirm that one's sample of values is actually
drawn from the assumed
population. And so on.
There are many questions to be asked
and answered and both statistical
and graphical analysis may have to be
done. Stevens' distinctions may
be somewhat helfpul rules of thumb or
statistical heuristics but they don't
substitute for thinking about and
knowing the the data that one is working
with. That is, be smart about
your data.
-Mike Palij
New York University
Reference
Frederic M. Lord (1953). On the
statistical treatment of footbal numbers.
American Psychologist, 8,
750-751.
|
|
|
Hi all,
Thank you all for your valuable responses. Can I just summarize the responses as the following?
I shall give further details of my analysis. My client has wanted me to do a 2-independent T-test and ANOVA for a 6-point respond against gender and education level respectively. My hunch is to use a CHI-square due to the normality issue of the 6-point respond. Do feel free to further comment on this. Thanks so much in advance. Dorraj Oet Date: Wed, 4 Aug 2010 16:34:11 -0400 From: [hidden email] Subject: Re: 6 point variable To: [hidden email] Not to belabor the point but in some respect whether one's variable is
nominal, ordinal, interval, ratio, absolute or whatever is somewhat besides
the point. The more fundamental question is:
What is the nature of the population distribution from which this sample
has been drawn, if one is assuming this is a sample from a population of
values?
One can always calculate a mean or a standard deviation or any statistic
one wants on a group numbers but the quetion is what does such a
statistic mean? Frederic Lord in his classic 1953 article on the "statistical
treatment of football numbers" showed that one can reach valid
conclusions about different samples of numbers even when they appear
to be only nominal (for those unfamilar with Lord's paper, the problem
is that a group of college freshman go to teacher of statistics to determine
whether they are receiving football jerseys with numbers systematically
lower than that of upper classmen -- the null hypothesis suggests that
if the numbers are assigned at random than the mean of the freshman
jerseys should not be significantly different from the mean jersey number
of the upper classmen; Tchebycheff's inequality was used as the statistical
test and it showed that the freshman mean was significantly lower).
The real problem, I think, is that many people would in fact treat the
6 point scale as an interval scale and then use a t-test or ANOVA or
multiple regression analysis that make assumptions way beyond one's
data is interval (e.g., follow a normal distribution). For purposes of
descriptive analysis, one should not just get the mean but all statistics
that bear on the shape of the distribution of values. Is the sample
symmetric around a central point (in which case the mean, median, and
mode will all be similar in value). If the distribution is skewed, why is
it skewed? Does the question that one is asked causing respondents
to favor one side of the rating scale over others? And so on.
Finally, what sorts of inferences and/or conclusions does one want to
make. If one is simply describing a sample of values with no interest in
a possible population from which it may have been drawn, then the sample
is the population. Describe it adquately. But if one is using the sample
to draw inferences about a population from which it was drawn, then which
population is being used? What characteristics of the population is one
interested in? How will one confirm that one's sample of values is actually
drawn from the assumed population. And so on.
There are many questions to be asked and answered and both statistical
and graphical analysis may have to be done. Stevens' distinctions may
be somewhat helfpul rules of thumb or statistical heuristics but they don't
substitute for thinking about and knowing the the data that one is working
with. That is, be smart about your data.
-Mike Palij
New York University
Reference
Frederic M. Lord (1953). On the statistical treatment of footbal numbers.
American Psychologist, 8, 750-751.
|
|
|
In reply to this post by Mike
Common sense by all means (no pun
intended).
Apologies to all those statisticians out there, but
I would always put substance above technique. As my old boss at the SSRC Survey
Unit, the late Dr Mark Abrams, used to claim, "If it's worth saying, you can say
it in percentages!" He kept a slide-rule on his desk and once, in
exasperation at a slight delay in courier delivery of printout, yelled,
"Computers! Computers! It's quicker to count them by foot!"
However, I think he would like the barcharts now available in
SPSS.
Back to Likert. I would never rely on a
single item to measure an attitude, but I have no problem with generating a
correlation matrix (en route to selecting items) or
calculating an attitude score from carefully selected ordinal
items (regardless of the shape of their distributions).
On my website,
3.5 Derived variables is a set of
tutorials with fully worked examples (and critical comments) on the use of COUNT
and COMPUTE to construct scores on simple attitude scales to measure
"attachment to status quo" and "sexism" using data from a
real survey (self-completion survey of all pupils aged 15 and 16 in
one school attending on a single day). I don't go into the methodology of
how the ordinal items for "sexism" were selected (Pearson matrix, eyeball
clustering, check alpha for reliability) but the results make
sense. The other (reputedly Guttman) scale was replicated from work by the
late Prof Hilde Himmelweit (LSE) but I can't find a reference for it anywhere:
it may have come from a draft questionnaire she used in the 1970s when she was
working on young people's political attitudes.
5.1 An introduction to COUNT and COMPUTE [Page
Header]
5.2 Teenage Attitudes (Tutorials) [Page Header] 5.2.1 COUNT and COMPUTE - Preliminary notes [doc: 1.9 mb] 5.2.2 Data checks 1 - Status quo [doc: 0.99 mb] 5.2.3 The COUNT command 1 - Attachment to status quo[doc: 1.29 mb] 5.2.4 The COMPUTE command 1 - Attachment to status quo [doc: 1.28 mb] 5.2.5 Data checks 2 - Sexism [doc: 1.95 mb] 5.2.6 The COUNT command 2 - Sexism [doc: 2.33 mb] 5.2.7 The COMPUTE command 2 - Sexism [doc: 1.44 mb] All materials from the survey (facsimile questionnaire, raw data, SPSS
saved file) are on Fifth Form survey
if you feel you can extend the
analysis.
John Hall
|
|
|
In reply to this post by Wilhelm Landerholm | Queue
NEVER take means of ordinal data Get proprotions for each response, then transform to normal, or logistic THEN take means, then ptransfor back to proportions Best Diana On 04/08/2010 19:22, "Wilhelm Landerholm | Queue" <wl@...> wrote: Hi! Professor Diana Kornbrot email: d.e.kornbrot@... web: http://web.me.com/kornbrot/KornbrotHome.html Work Centre for Lifespan & Chronic Illness Research, CLiCIR School of Psychology University of Hertfordshire College Lane, Hatfield, Hertfordshire AL10 9AB, UK voice: +44 (0) 170 728 4626 Home 19 Elmhurst Avenue London N2 0LT, UK voice: +44 (0) 208 883 3657 mobile: +44 (0) 7855 415 425 fax: +44 (0) 870 706 4997 |
Re: 6 point variable
|
|
Hi again,
Unfortunately, there is not so much written (easily reading) on the problems around the ordinal scales. There are some around how to measure pain that is quite readable; but, I would recommend those interested to take a look at Fiorenzo Franceschini et al article Qualitative ordinal Scales: The Concept of ordinal Range Link: http://staff.polito.it/fiorenzo.franceschini/Pubblicazioni/Qualitative%20ordinal%20scales%20the%20concept%20of%20ordinal%20range.pdf Simple and easy to understand Is there anyone out there who has a good article or tip on articles dealing with CLT and ordinal scales? All the best Wilhelm (Wille) Landerholm +46-735-460000 Queue/STATB Making sense of data BOX 92 | 162 12 Vallingby | Sweden 2010/8/5 Kornbrot, Diana <[hidden email]>
|
|
|
In reply to this post by Marta Garcia-Granero
Maria,
I can understand how you feel being forced to abandon a tool that you are proficient in using. But I believe that once you started using Stata seriously you will find that in many respects (perhaps not all) it is superior to SPSS, and on the long run you won't regret it. I used SPSS since about 1980 (Mainframe SPSS-X, SPSS-PC, and versions 6 up to 15) and switched to Stata because I attended workshops for advanced methods that were not possible to use with SPSS (gllamm, analyzes of count data, survey analyses). Since about two years I'm using SPSS only occasionally, and when I do I am happy that I made the move to Stata. Many of my colleagues are still using SPSS, but more and more are moving to Stata, too. To my mind R is very interesting and promising but still not really an alternative because it takes much more programming skills than most people can afford to develop or invest. Of course, being forced to switch is a bad thing - it kills curiosity, enthusiasm, pleasure, and its a waste of your expertise. Although this should not happen, there is also a bright side to it: You will become more flexible in choosing the tools that are suited best for the problem at hand. I recommend to read the (to my mind) fair comparison of SAS, SPSS, and Stata (with a short excursion to R) - although some years old there are some very good arguments to be considered seriously when making a choice between different packages: http://www.ats.ucla.edu/stat/TechnicalReports/number1_editedFeb_2_2007/ucla_ATSstat_tr1_1.1_0207.pdf Yours, Dirk Marta García-Granero wrote: >Hi everybody: > > Perhaps SPSS 15 could be saved from extinction and turned into that > cheaper home version some people is asking for. > > How I wish it was that way! University of Navarra will abandon SPSS > absolutely (not even one individual license) by June 2011. It looks like > being charged a lot for SPSS 15 (without any technical support) for > years, and the fact that we will have to upgrade (like it or not), even > though PASW/SPSS 18 doesn't run OK on campus computers, was simply too > much. We will switch to Stata. > > I have always insisted in this forum that students don't need to use the > newest and brand new version of SPSS. SSPS 15 is reasonably stable, and > powerful enough for basic statistics teaching (descriptive, Student t > test, oneway- twoway- ANOVA, correlation, regression and contingency > tables analyses). > > If SPSS keeps on loosing universities, they will finally loose a lot of > graduated people, that will decide to use other statistical software in > their jobs (since they did not learn how to use SPSS at their > universities). > > I'm really sad to have to stop using SPSS after 2011. I have been a SPSS > user for 20 years, but I will not be able to afford one individual > license for myself after the University switches to Stata. I really wish > there was a inexpensive home version of SPSS I could by for private use. > SPSS 15 deserves being saved from extinction. > > Best regards, > Marta GG > > (soon a Stata user) ======================================== Dr. Dirk Enzmann Institute of Criminal Sciences Dept. of Criminology Rothenbaumchaussee 33 D-20148 Hamburg Germany phone: +49-(0)40-42838.7498 (office) +49-(0)40-42838.4591 (Mrs Billon) fax: +49-(0)40-42838.2344 email: [hidden email] http://www2.jura.uni-hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Enzmann.html ======================================== ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
A defense of SPSS pricing (was: inexpensive 'home' version?)
|
|
In reply to this post by Peter A. Neenan, Ph.D.
The following message by Peter A. Neenan, Ph.D. created a flurry of
responses: > Are there any inexpensive 'home' or 'basic' versions of SPSS for > individuals who are not affiliated with an academic institution? I > know there is a 'student' version--or there used to be--but one had > to be enrolled or otherwise affilliated with an academic institution > to purchase it. Any suggestions? Many of the messages in response to Dr. Neenan's message have been critical of the pricing model for SPSS and have suggested a special much lower price for individuals. In the spirit of debate, let me offer a contrarian viewpoint. I don't think that any of us on this list have an appreciation of the big picture. In the world of software where things change very rapidly, we have seen continual and regular success for SPSS (and SAS for that matter) over a span of more than four decades. The pricing structure for SPSS has been pretty consistent, and they make a ton of money. The problem with a special low price version for people like you and me is that it would cannibalize all their current sales. You can argue, what about Stata? What about R? Those are different products, and they should have different prices. If IBM adopted the Stata pricing policy, how do you know that they wouldn't take a very successful product and drive it to ruin? Oh sure, cut the price and sales go way up. But do any of us understand the market for statistical software well enough to predict whether the increase in volume will make up for the loss in revenues. I don't think so. The folks at SPSS (and at IBM even more so) are survivors who have continued to succeed in a mercurial market. Maybe they are making the wrong call here, but given their access to marketing data and their previous track record, I am extremely reluctant to second guess them. Don't get me wrong. I'd be thrilled if SPSS lowered their prices. And I use R largely because of its price. You can't ignore price when picking a package (or for that matter, ignore the difference between a flat out purchase versus a yearly license fee). I just think that some of the arguments that SPSS should lower their prices are a bit naive. It's sort of like telling Jaguar that they should sell all their cars at the price of a Honda Civic. You can also make the argument that Stata is superior, but I find the two packages almost impossible to compare. I use both. Stata is better for some people and SPSS is better for other people. I have not bought SPSS for my private consulting business, but I am very grateful that at my part-time job at UMKC, that they have made the commitment to pay for SPSS, and I recommend SPSS to a lot of people at UMKC. I'm thinking about writing a second book, * http://www.pmean.com/10/SecondBook.html and if I can convince a publisher to print it, I will use SPSS in all my examples. Other than how it affect me directly, I don't know if the SPSS pricing model is good or bad. I don't think any of the rest of you do either. The only comment I will say with certainty is that it is a mistake for any one of us to become too dependent on a single statistical package. If arguments about statistical packages amount to a religious fervor, count me in among the polytheists. Steve Simon, Standard Disclaimer Sign up for The Monthly Mean, the newsletter that dares to call itself "average" at www.pmean.com/news "Data entry and data management issues with examples in IBM SPSS," Tuesday, August 24, 11am-noon CDT. Free webinar. Details at www.pmean.com/webinars ===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD |
|
|
In reply to this post by Jarrod Teo-2
buy no buy St agree f11 f12 Agr f21 f22 Neutral f31 f32 Disag f41 f42 Str dosage f51 f52 Then perform a chi-square test This will tell you which responses result in most buy decisions. It will also tell you whether you gain more by converting agree to strongly agree, or neutral to agree In my view you would be better off with strongly like, like, neutral, dislike strongly dislike than the ghastly agree format Best Diana
Professor Diana Kornbrot email: d.e.kornbrot@... web: http://web.me.com/kornbrot/KornbrotHome.html Work Centre for Lifespan & Chronic Illness Research, CLiCIR School of Psychology University of Hertfordshire College Lane, Hatfield, Hertfordshire AL10 9AB, UK voice: +44 (0) 170 728 4626 Home 19 Elmhurst Avenue London N2 0LT, UK voice: +44 (0) 208 883 3657 mobile: +44 (0) 7855 415 425 fax: +44 (0) 870 706 4997 |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |