[tips] Likert scale graph/chart results presentation
Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. - --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5n=Tl=tipso=17270 or send a blank email to leave-17270-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
Re: [tips] Likert scale graph/chart results presentation
Hi Nancy- I would use a separated bar graph. Items on a Likert scale are rarely (if ever) on a true equal interval continuum. The bar graph will make it clear that these were discrete responses. -Don. - Original Message - From: drnanjo drna...@aol.com Date: Thursday, April 12, 2012 11:46 am Subject: [tips] Likert scale graph/chart results presentation To: Teaching in the Psychological Sciences (TIPS) tips@fsulist.frostburg.edu Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. - --- You are currently subscribed to tips as: dap...@shaw.ca. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13157.966b795bc7f3ccb35e3da08aebe98f18n=Tl=tipso=17270or send a blank email to leave-17270-13157.966b795bc7f3ccb35e3da08aebe98...@fsulist.frostburg.edu Don Allen Retired professor Langara College --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5n=Tl=tipso=17272 or send a blank email to leave-17272-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
Re: [tips] Likert scale graph/chart results presentation
If you want to be scrupulous, the convention (such as it is) is to use bar graphs (with spaces between the bars) whenever the values use along the horizontal axis are discrete, and a histogram (with bars touching each other) when the values along the horizontal axis are continuous. But the convention is violated so regularly, that it is only a convention in the minds of scrupulous statisticians. Chris --- Christopher D. Green Department of Psychology York University Toronto, ON M3J 1P3 Canada chri...@yorku.ca http://www.yorku.ca/christo/ == On 2012-04-12, at 2:46 PM, drnanjo wrote: Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. - --- You are currently subscribed to tips as: chri...@yorku.ca. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=430248.781165b5ef80a3cd2b14721caf62bd92n=Tl=tipso=17270 (It may be necessary to cut and paste the above URL if the line is broken) or send a blank email to leave-17270-430248.781165b5ef80a3cd2b14721caf62b...@fsulist.frostburg.edu --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5n=Tl=tipso=17275 or send a blank email to leave-17275-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
Re: [tips] Likert scale graph/chart results presentation
My advisor, Michael Friendly, literally wrote the book on *Visualizing Categorical Data* (of which Likert style responses are a subset, web linkhttp://www.datavis.ca/books/vcd/). It was originally written for SAS, but some researchers have taken many of the concepts and implemented them in R (package linkhttp://cran.r-project.org/web/packages/vcd/index.html ). For relationships between questions (or other categorical variables, like gender), mosaic plots (simple examplehttp://www.statmethods.net/advgraphs/mosaic.html) are a good option, especially with the options for shading based upon the size of the standardized residuals per cell. I am also a fan of using correlation scatterplot matrices for ordered-categorical data, where simple bar plots appear on the diagonal, scatterplots with LOESS lines on the lower off-diagonals (with bubble size indicating count), and correlation coefficients on the upper-diagonals ( example http://i.stack.imgur.com/hCzCn.png, codehttp://www.r-statistics.com/2010/04/correlation-scatter-plot-matrix-for-ordered-categorical-data/ ). But you are correct in that, traditionally, likert results are visualized using simple bar charts. Hope that helps, Matt -- Matthew J. Sigal, BA (Hons), MA Quantitative Methods, PhD2 Department of Psychology 262 Behavioural Science Building York University, 4700 Keele St. Toronto, ON, Canada M3J 1P3 (416) 736-2100 x66163 msi...@yorku.ca / www.matthewsigal.com On Thu, Apr 12, 2012 at 14:46, drnanjo drna...@aol.com wrote: Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5n=Tl=tipso=17277 or send a blank email to leave-17277-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
Re: [tips] Likert scale graph/chart results presentation
I think either is ok ..though my preference is histogram style. Even though (as Don says) the Likert scales are not necessarily equal intervals if we think about the relationship between our data and the true (unknowable) underlying function, I would consider the scales continuous as opposed to discrete (besides, can't a scale be continuous and _not_ equal interval?) . When we analyze the data we convert them to numbers and interpret the points _between_ two values ... this can't be done with blatantly discrete values such as political affiliations. You cant be half way between Democrat and Socialist if collected categorical style, but you _can_ be half way between agree and strongly agree, both as an individual response and a group average. I suspect (no firm data though) that most people, on most scales, treat them as continuous scales. I have seen enough people put their check mark half way two anchors on the scale. Also, when we write the scales we prod participants into responding as if it continuous with our verbal anchors such as _degree_ of agreement. Many of these scales have numbers and a continuous line prodding the participants to respond continuously as best they can. Finally, from the two wrongs don't make a right department, statistically we often treat the data as equal interval, not ordinal. And finally finally, if we don't want to worry about the relationship between our crude measurements and the true underlying variable (the IQ is what IQ tests measure attitude) much of the agonizing about this issue goes away .. I think (but its late, thinking is hazardous past 9 pm!). So there are lots of forces pushing those little numbers into the continuous category ... good question Nancy. == John W. Kulig, Ph.D. Professor of Psychology Coordinator, University Honors Plymouth State University Plymouth NH 03264 == - Original Message - From: Christopher Green chri...@yorku.ca To: Teaching in the Psychological Sciences (TIPS) tips@fsulist.frostburg.edu Sent: Thursday, April 12, 2012 5:23:02 PM Subject: Re: [tips] Likert scale graph/chart results presentation If you want to be scrupulous, the convention (such as it is) is to use bar graphs (with spaces between the bars) whenever the values use along the horizontal axis are discrete, and a histogram (with bars touching each other) when the values along the horizontal axis are continuous. But the convention is violated so regularly, that it is only a convention in the minds of scrupulous statisticians. Chris --- Christopher D. Green Department of Psychology York University Toronto, ON M3J 1P3 Canada chri...@yorku.ca http://www.yorku.ca/christo/ == On 2012-04-12, at 2:46 PM, drnanjo wrote: Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. - --- You are currently subscribed to tips as: chri...@yorku.ca . To unsubscribe click here: http://fsulist.frostburg.edu/u?id=430248.781165b5ef80a3cd2b14721caf62bd92n=Tl=tipso=17270 (It may be necessary to cut and paste the above URL if the line is broken) or send a blank email to leave-17270-430248.781165b5ef80a3cd2b14721caf62b...@fsulist.frostburg.edu --- You are currently subscribed to tips as: ku...@mail.plymouth.edu . To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13338.f659d005276678c0696b7f6beda66454n=Tl=tipso=17275 (It may be necessary to cut and paste the above URL if the line is broken) or send a blank email to leave-17275-13338.f659d005276678c0696b7f6beda66...@fsulist.frostburg.edu --- You are currently subscribed to tips as: arch...@jab.org. To unsubscribe click here: http://fsulist.frostburg.edu/u?id=13090.68da6e6e5325aa33287ff385b70df5d5n=Tl=tipso=17282 or send a blank email to leave-17282-13090.68da6e6e5325aa33287ff385b70df...@fsulist.frostburg.edu
RE: [tips] Likert scale graph/chart results presentation
From my limited reading about Likert and Likert scales, I'm not sure he would approve of reporting frequencies on individual items at all. It is my understanding that a Likert scale is always a collection of Likert items designed to measure some construct. Although it is common to report frequencies of responses and sometimes combinations of scores (such as the percentage of 4's and 5's) that is not how Likert intended them to be used. I have read (or heard) that Likert items/scales are technically only those which use a 5 point scale of agreement, all others are technically Likert-like scales. Those with labels other than agree/disagree are technically variations of semantic differential scales. Can anyone confirm or deny this? Finally, does the concern over interval become less of a problem when averaged or summed over a series of items, if the scale and its individual items have been developed properly? Doug Doug Peterson, PhD Associate Professor of Psychology The University of South Dakota Vermillion SD 57069 605.677.5223 From: John Kulig [ku...@mail.plymouth.edu] Sent: Thursday, April 12, 2012 8:45 PM To: Teaching in the Psychological Sciences (TIPS) Subject: Re: [tips] Likert scale graph/chart results presentation I think either is ok ..though my preference is histogram style. Even though (as Don says) the Likert scales are not necessarily equal intervals if we think about the relationship between our data and the true (unknowable) underlying function, I would consider the scales continuous as opposed to discrete (besides, can't a scale be continuous and _not_ equal interval?) . When we analyze the data we convert them to numbers and interpret the points _between_ two values ... this can't be done with blatantly discrete values such as political affiliations. You cant be half way between Democrat and Socialist if collected categorical style, but you _can_ be half way between agree and strongly agree, both as an individual response and a group average. I suspect (no firm data though) that most people, on most scales, treat them as continuous scales. I have seen enough people put their check mark half way two anchors on the scale. Also, when we write the scales we prod participants into responding as if it continuous with our verbal anchors such as _degree_ of agreement. Many of these scales have numbers and a continuous line prodding the participants to respond continuously as best they can. Finally, from the two wrongs don't make a right department, statistically we often treat the data as equal interval, not ordinal. And finally finally, if we don't want to worry about the relationship between our crude measurements and the true underlying variable (the IQ is what IQ tests measure attitude) much of the agonizing about this issue goes away .. I think (but its late, thinking is hazardous past 9 pm!). So there are lots of forces pushing those little numbers into the continuous category ... good question Nancy. == John W. Kulig, Ph.D. Professor of Psychology Coordinator, University Honors Plymouth State University Plymouth NH 03264 == From: Christopher Green chri...@yorku.ca To: Teaching in the Psychological Sciences (TIPS) tips@fsulist.frostburg.edu Sent: Thursday, April 12, 2012 5:23:02 PM Subject: Re: [tips] Likert scale graph/chart results presentation If you want to be scrupulous, the convention (such as it is) is to use bar graphs (with spaces between the bars) whenever the values use along the horizontal axis are discrete, and a histogram (with bars touching each other) when the values along the horizontal axis are continuous. But the convention is violated so regularly, that it is only a convention in the minds of scrupulous statisticians. Chris --- Christopher D. Green Department of Psychology York University Toronto, ON M3J 1P3 Canada chri...@yorku.camailto:chri...@yorku.ca http://www.yorku.ca/christo/ == On 2012-04-12, at 2:46 PM, drnanjo wrote: Hey gang: I am teaching an upper division research methods class for the first time in my life. As such, I want to be scrupulous about the guidance I give If one wishes to present likert scale results in pictorial form, would one do a histogram (continuous, with bars touching) or a bar graph (each point on the Likert scale represented by a bar? I am asking because the rules seem to be lose sometimes - for example, income is technically quantitative and ratio type data but some researchers divide income into classes and make a bar graph instead of a histogram or line graph. Thanks in advance for sharing the collective wisdom. Nancy Melucci (in this case) California State University in the Hills of Dominguez. - --- You are currently subscribed to tips as: chri...@yorku.camailto:chri...@yorku.ca. To unsubscribe click