A 4 point Likert scale
Locked
 
|
Hi
I intend to use a 4 point Likert scale format in my survey. I want to know if this puts limitations on the type of statistical analysis tests conducted? Regards Mohamed |
|
Why do you want to use only a 4 point Likert scale?
_The more you restrict the variance of a variable the more you restrict its possible covariance._ As a rule of thumb you want items to approximate a continuous construct as much as possible given who your respondents are. An actual Likert item has 5 points SD D ? A SA. Do you have a Disagree to Agree construct for your response scale? If your scale has many items, the total (mean) score might not be too restricted. If you do not have scales, but are measuring a construct with a single variable the restriction of variance and therefore on covariance is even more problematical. In short, a 4 point response scale is usually inadvisable. Why coarsen your measurement any more than is really necessary? Art Kendall Social Research Consultants On 10/17/2011 1:13 PM, Mohamed wrote: ===================== 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 REFCARDHi I intend to use a 4 point Likert scale format in my survey. I want to know if this puts limitations on the type of statistical analysis tests conducted? Regards Mohamed -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4910676.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== 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
Art Kendall
Social Research Consultants |
|
|
Thank you for your help. The point is, I do not want to have a neutral position. This is not a construct. so do you suggest a 6 point scale , giving more stretch and eliminating the neutral status? However, I do have a construct [5 items]
that was originally measured on a 4 point scale and I intend to use it. Regards Mohamed From: Art Kendall <[hidden email]> To: Mohamed <[hidden email]> Cc: [hidden email] Sent: Monday, October 17, 2011 8:19 PM Subject: Re: [SPSSX-L] A 4 point Likert scale
Why do you want to use only a 4 point Likert scale?
_The more you restrict the variance of a variable the more you restrict its possible covariance._ As a rule of thumb you want items to approximate a continuous construct as much as possible given who your respondents are. An actual Likert item has 5 points SD D ? A SA. Do you have a Disagree to Agree construct for your response scale? If your scale has many items, the total (mean) score might not be too restricted. If you do not have scales, but are measuring a construct with a single variable the restriction of variance and therefore on covariance is even more problematical. In short, a 4 point response scale is usually inadvisable. Why coarsen your measurement any more than is really necessary? Art Kendall Social Research Consultants On 10/17/2011 1:13 PM, Mohamed wrote: Hi I intend to use a 4 point Likert scale format in my survey. I want to know if this puts limitations on the type of statistical analysis tests conducted? Regards Mohamed -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4910676.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== 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 Art Kendall
With what Art said in mind, if you’re
desirous of eliminating the neutral option, a six-point scale is far superior
to a four-point scale. The research on Likert and Likert-type scales is that
most people can discriminate up to seven levels of response, so you’re
safe with six. Brian From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Art Kendall Why do you want to use only a 4 point Likert scale? HiI intend to use a 4 point Likert scale format in my survey.I want to know if this puts limitations on the type of statistical analysistests conducted?RegardsMohamed--View this message in context: http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4910676.htmlSent from the SPSSX Discussion mailing list archive at Nabble.com.=====================To manage your subscription to SPSSX-L, send a message to[hidden email] (not to SPSSX-L), with no body text except thecommand. To leave the list, send the commandSIGNOFF SPSSX-LFor a list of commands to manage subscriptions, send the commandINFO REFCARD |
|
|
In reply to this post by Mohamed
In addition to these obserbvations, there is no middle point.
Ralph
In a message dated 10/17/2011 2:24:12 P.M. Eastern Daylight Time,
[hidden email] writes:
Why do you want to use only a 4 point Likert scale? |
Re: A 4 point Likert scale
|
|
Not to belabor this too far, but in addition to the other points, depending on the question many survey respondents avoid the extremes in likert-type scales. Thus a 4 point scale effectively can turn
into a 2 point scale.
Evan Harrington, Ph.D.
Department of Forensic Psychology
The Chicago School of Professional Psychology
325 N. Wells Street
Chicago, IL 60654
From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Ralph Grubb [[hidden email]] Sent: Monday, October 17, 2011 1:32 PM To: [hidden email] Subject: Re: A 4 point Likert scale In addition to these obserbvations, there is no middle point. Ralph
In a message dated 10/17/2011 2:24:12 P.M. Eastern Daylight Time, [hidden email] writes:
Why do you want to use only a 4 point Likert scale? |
|
|
In reply to this post by Mohamed
A six point would eliminate the middle point.
This is not a construct.You must be measuring _something_ that the respondent agrees with. Your stimulus is an operationalization of some idea (aka, construct). If there is a well-established scale that unfortunately used a 4 point response scale, then you are stuck with it. How was the reliability of the previous use(s)? If it is still developmental, you should consider using a longer response scale? What is so bad about a midpoint what are you measuring? Art Kendall Social Research Consultants On 10/17/2011 2:26 PM, Mohamed Fawzy Afify wrote: ===================== 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
Art Kendall
Social Research Consultants |
|
|
In reply to this post by Mohamed
Without knowing more about the construct you're trying to measure, it's difficult to provide more than some very rough guidelines: --Avoid odd numbers of response categories. --Avoid "not applicable" or "no opinion" response categories. --Consider starting with 6 categories. For more guidance: ~~~~~~~~~~~ Scott R Millis, PhD, ABPP, CStat, PStat® Professor Wayne State University School of Medicine Email: [hidden email] Email: [hidden email] Tel: 313-993-8085 From: Mohamed Fawzy Afify <[hidden email]> To: [hidden email] Sent: Monday, October 17, 2011 2:26 PM Subject: Re: A 4 point Likert scale Thank you for your help. The point is, I do not want to have a neutral position. This is not a construct. so do you suggest a 6 point scale , giving more stretch and eliminating the neutral status? However, I do have a construct [5 items]
that was originally measured on a 4 point scale and I intend to use it. Regards Mohamed From: Art Kendall <[hidden email]> To: Mohamed <[hidden email]> Cc: [hidden email] Sent: Monday, October 17, 2011 8:19 PM Subject: Re: [SPSSX-L] A 4 point Likert scale
Why do you want to use only a 4 point Likert scale?
_The more you restrict the variance of a variable the more you restrict its possible covariance._ As a rule of thumb you want items to approximate a continuous construct as much as possible given who your respondents are. An actual Likert item has 5 points SD D ? A SA. Do you have a Disagree to Agree construct for your response scale? If your scale has many items, the total (mean) score might not be too restricted. If you do not have scales, but are measuring a construct with a single variable the restriction of variance and therefore on covariance is even more problematical. In short, a 4 point response scale is usually inadvisable. Why coarsen your measurement any more than is really necessary? Art Kendall Social Research Consultants On 10/17/2011 1:13 PM, Mohamed wrote: Hi I intend to use a 4 point Likert scale format in my survey. I want to know if this puts limitations on the type of statistical analysis tests conducted? Regards Mohamed -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4910676.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== 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 |
Administrator
|
Interesting guidelines, Scott. Thanks for posting. However, re items 3 and 4 (see below), my sister (to whom I sent the link) wonders how would one distinguish between a deliberate choice to not respond and failure to notice the item?
Cheers, Bruce --- from the rules of thumb --- 3. Avoid "Not Applicable" or "No Opinion" response categories. It is far better to instruct respondents to skip irrelevant items than it is to offer them the opportunity in every item to seem to provide data, but without having to make a decision. 4. Avoid odd numbers of response options. Middle categories can attract disproportionate numbers of responses. Like "Not Applicable" options, middle categories allow respondents to appear to be providing data, but without making a decision. If someone really cannot decide which side of an issue they come down on, it is better to let them decide on their own to skip the question. If the data then show that two adjacent categories turn out to be incapable of sustaining a quantitative distinction, that evidence will be in hand and can inform future designs.
--
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 Mohamed
I've seen a number of messages about why 4-points is
not as good as 5 or 6, for reasons of precision and bias. However, none of that speaks to the original question about whether this "puts limitations on types of statistical analysis tests conducted." In a word, mainly, "No." You will use ordinary ANOVA or Pearson correlations for analyzing the scale totals, or for analyzing single items if you want to pursue single items. If you have large enough Ns, you might look into developing a scale by IRT or Rasch methods. Now, I am going to assume that you might have heard of some bad advice that was once promulgated by a few non-statisticians in psychology and education. That was about 40 years ago, when interest first grew in various so-called "robust" techniques. The bad message has not been completely stamped out. What you certainly do *not* want to do is retreat to rank- based "non-parametric" statistics, which are generally messed up by the excess of ties. (Also, using ranks costs you the ability to discuss simple means). -- Rich Ulrich > Date: Mon, 17 Oct 2011 10:13:57 -0700 > From: [hidden email] > Subject: A 4 point Likert scale > To: [hidden email] > > Hi > > > I intend to use a 4 point Likert scale format in my survey. > > I want to know if this puts limitations on the type of statistical analysis > tests conducted? > |
|
|
I agree the tests would be the same, but the
restriction on variance would provide less "room" for an effect to
to be seen.
Art Kendall Social Research Consultants On 10/17/2011 9:37 PM, Rich Ulrich wrote: ===================== 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
Art Kendall
Social Research Consultants |
|
|
In reply to this post by Bruce Weaver
Another issue is non-response. Sometimes when the respondent does not have an opinion on an item and they do not have a "no opinion" response option, hey will simply fail to respond to the item. As far as the analysis goes, it really depends on the distribution. If it is badly skewed, especially with small or unequal group sizes, then a standard analysis may give spurious results. In this case, I like to use bootstrapping.
Paul Dr. Paul R. Swank, Children's Learning Institute Professor, Department of Pediatrics, Medical School Adjunct Professor, School of Public Health University of Texas Health Science Center-Houston -----Original Message----- From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Bruce Weaver Sent: Monday, October 17, 2011 3:28 PM To: [hidden email] Subject: Re: A 4 point Likert scale Interesting guidelines, Scott. Thanks for posting. However, re items 3 and 4 (see below), my sister (to whom I sent the link) wonders how would one distinguish between a deliberate choice to not respond and failure to notice the item? Cheers, Bruce --- from the rules of thumb --- 3. Avoid "Not Applicable" or "No Opinion" response categories. It is far better to instruct respondents to skip irrelevant items than it is to offer them the opportunity in every item to seem to provide data, but without having to make a decision. 4. Avoid odd numbers of response options. Middle categories can attract disproportionate numbers of responses. Like "Not Applicable" options, middle categories allow respondents to appear to be providing data, but without making a decision. If someone really cannot decide which side of an issue they come down on, it is better to let them decide on their own to skip the question. If the data then show that two adjacent categories turn out to be incapable of sustaining a quantitative distinction, that evidence will be in hand and can inform future designs. SR Millis-3 wrote: > > Without knowing more about the construct you're trying to measure, it's > difficult to provide more than some very rough guidelines: > > --Avoid odd numbers of response categories. > > --Avoid "not applicable" or "no opinion" response categories. > > --Consider starting with 6 categories. > > For more guidance: > http://www.rasch.org/rmt/rmt203f.htm > > > ~~~~~~~~~~~ > Scott R Millis, PhD, ABPP, CStat, PStat® > Professor > Wayne State University School of Medicine > Email: aa3379@ > Email: srmillis@ > Tel: 313-993-8085 > > > ________________________________ > From: Mohamed Fawzy Afify <mohamedfawzy74@> > To: SPSSX-L@.UGA > Sent: Monday, October 17, 2011 2:26 PM > Subject: Re: A 4 point Likert scale > > > Thank� you for your help. > > The point is, I do not want to have a� neutral� position. This is not a > construct. > > so do you suggest a 6 point scale , giving > more� stretch� and� eliminating� the neutral status? > > However, I do have a construct [5 items] that was originally measured on a > 4 point scale and I intend to use it. > > > Regards > > > Mohamed > > > > ________________________________ > From: Art Kendall <Art@> > To: Mohamed <mohamedfawzy74@> > Cc: SPSSX-L@.UGA > Sent: Monday, October 17, 2011 8:19 PM > Subject: Re: [SPSSX-L] A 4 point Likert scale > > > Why do you want to use only a 4 point Likert scale? > _The more you restrict the variance of a variable the more you > restrict its possible covariance._ > As a rule of thumb you want items to approximate a continuous > construct as much as possible given who your respondents are. > > An actual Likert item has 5 points SD D ? A SA.� Do you have a > Disagree to Agree construct for your response scale? > > If your scale has many items, the total (mean) score might not be > too restricted. > If you do not have scales, but are measuring a construct with a > single variable the restriction of variance and therefore on > covariance is even more problematical. > > In short, a 4 point response scale is usually inadvisable. Why > coarsen your measurement any more than is really necessary? > > > Art Kendall > Social Research Consultants > > On 10/17/2011 1:13 PM, Mohamed wrote: > Hi I intend to use a 4 point Likert scale format in my survey. I want to > know if this puts limitations on the type of statistical analysis > tests conducted? Regards Mohamed -- > View this message in context: > http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4910676.html > Sent from the SPSSX Discussion mailing list archive at Nabble.com. > ===================== > To manage your subscription to SPSSX-L, send a message to LISTSERV@.UGA > (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 > ----- -- Bruce Weaver [hidden email] http://sites.google.com/a/lakeheadu.ca/bweaver/ "When all else fails, RTFM." NOTE: My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above. -- View this message in context: http://spssx-discussion.1045642.n5.nabble.com/A-4-point-Likert-scale-tp4910676p4911452.html Sent from the SPSSX Discussion mailing list archive at Nabble.com. ===================== 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 ===================== 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 Rich Ulrich
Hello Thank you for your help. I appreciate it. First of all I am novice in that and I wanted to avoid making fatal mistakes ! The reason I asked this is that I saw on the internet
someone's power point slides wheres/he warned against using 4 point and some tests. So for me it was the 4 point rather than Likert itself ! As I am interested in ruling out the so called 'fence-sitters' I wanted to use 4 points. However, to have a better covariance, I was advised to use 6 points ! In a day and night the whole issue got another radical dimension. Is it ordinal or interval? parametric or non-parametric.' Hamlet's unasked questions ! As you guessed I have already got the answers and as you know, beforehand, giving mixed messages. This was good opportunity to learn about that for a novice, though confusing. However, Now I know it is not my own problem. It is a 'global' issue !! Thank you very
much Mohamed From: Rich Ulrich <[hidden email]> To: [hidden email]; SPSS list <[hidden email]> Sent: Tuesday, October 18, 2011 3:37 AM Subject: RE: A 4 point Likert scale
I've seen a number of messages about why 4-points is not as good as 5 or 6, for reasons of precision and bias. However, none of that speaks to the original question about whether this "puts limitations on types of statistical analysis tests conducted." In a word, mainly, "No." You will use ordinary ANOVA or Pearson correlations for analyzing the scale totals, or for analyzing single items if you want to pursue single items. If you have large enough Ns, you might look into developing a scale by IRT or Rasch methods. Now, I am going to assume that you might have heard of some bad advice that was once promulgated by a few non-statisticians in psychology and education. That was about 40 years ago, when interest first grew in various so-called "robust" techniques. The bad message has not been completely stamped out. What you certainly do *not* want to do is retreat to rank- based "non-parametric" statistics, which are generally messed up by the excess of ties. (Also, using ranks costs you the ability to discuss simple means). -- Rich Ulrich > Date: Mon, 17 Oct 2011 10:13:57 -0700 > From: [hidden email] > Subject: A 4 point Likert scale > To: [hidden email] > > Hi > > > I intend to use a 4 point Likert scale format in my survey. > > I want to know if this puts limitations on the type of statistical analysis > tests conducted? > |
«
Return to SPSSX Discussion
|
1 view|%1 views
| Free forum by Nabble | Edit this page |