SPSS Statistics

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

I would much appreciate someone from SPSS team to answer or comment on my topic.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

Thank you for this. Please leave me the link here to the page of the issue #7628 that has been created.

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

Sure, Please find the page here, https://github.ibm.com/SPSS/stats_java_ui/issues/7628

Thank you for this. Please leave me the link here to the page of the issue #7628 that has been created.

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

How do I get there? It asks for some "w3id and password", but what's that? (I've got only my IBMid login, if I remember right.)

Sure, Please find the page here, https://github.ibm.com/SPSS/stats_java_ui/issues/7628

Thank you for this. Please leave me the link here to the page of the issue #7628 that has been created.

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

Yes its accessible only for IBMers internal..we sure would keep you posted on the updates after triaging and prioritising.

How do I get there? It asks for some "w3id and password", but what's that? (I've got only my IBMid login, if I remember right.)

Sure, Please find the page here, https://github.ibm.com/SPSS/stats_java_ui/issues/7628

Thank you for this. Please leave me the link here to the page of the issue #7628 that has been created.

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------

All right, Aruna. Have a nice day.

Yes its accessible only for IBMers internal..we sure would keep you posted on the updates after triaging and prioritising.

How do I get there? It asks for some "w3id and password", but what's that? (I've got only my IBMid login, if I remember right.)

Sure, Please find the page here, https://github.ibm.com/SPSS/stats_java_ui/issues/7628

Thank you for this. Please leave me the link here to the page of the issue #7628 that has been created.

Hi Kirill,

Thank you for bringing this to our attention.To look into this further and ensure the best possible experience, we've filed defect #7628 for the uneven rendering issue. We appreciate your patience and understanding.

There is an alarming feature of (panelled) dot plot in SPSS. Points of continuous-data cases that are all distinct and are equally-spaced - are rendered non-uniformly on the dot-plot. This is dangerous, showing incorrect data pattern.

In addition to the example syntax, I am appending a pair of pictures as well.

*Have a dataset with 10 cases.
*Create some equally-spaced continuous data.
*As an example, we will create Prorortion Estimate values of the 10 different values of x.
compute x= uniform(1).
RANK VARIABLES= x (A) /RFRACTION into FracRank /PRINT=NO /TIES=MEAN.
RANK VARIABLES= x (A) /PROPORTION into rankitPE /PRINT=NO /TIES=MEAN /FRACTION=rankit.
RANK VARIABLES= x (A) /PROPORTION into BlomPE /PRINT=NO /TIES=MEAN /FRACTION=blom.
RANK VARIABLES= x (A) /PROPORTION into TukeyPE /PRINT=NO /TIES=MEAN /FRACTION=tukey.
RANK VARIABLES= x (A) /PROPORTION into VWPE /PRINT=NO /TIES=MEAN /FRACTION=vw.
VARSTOCASES
  /ID=id
  /MAKE trans1 FROM FracRank rankitPE BlomPE TukeyPE VWPE
  /INDEX=Index1(trans1) 
  /KEEP=
  /NULL=KEEP.

*--NOW BUILD THE PLOT (it is the panelled dot-plot)--.
* Legacy dot-plot graph (asymmetric). Uneven spacing is seen.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= asymmetric
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (asymmetric). Uneven spacing is seen.
GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point.dodge.asymmetric(position(bin.dot(trans1*1*Index1)))
END GPL.
* Legacy dot-plot graph (flat). Correct, even spacing.
XGRAPH CHART=[POINT] BY trans1[s]
  /DISPLAY DOT= flat
  /PANEL ROWVAR=Index1 ROWOP=CROSS.
* Or via Chart Builder (flat). Uneven spacing is seen.
  GGRAPH
  /GRAPHDATASET NAME="graphdataset" VARIABLES=trans1 Index1 MISSING=LISTWISE REPORTMISSING=NO
  /GRAPHSPEC SOURCE=INLINE
  /FITLINE TOTAL=NO.
BEGIN GPL
  SOURCE: s=userSource(id("graphdataset"))
  DATA: trans1=col(source(s), name("trans1"))
  DATA: Index1=col(source(s), name("Index1"), unit.category())
  COORD: rect(dim(1))
  GUIDE: axis(dim(1), label("Fractional Rank of x"))
  GUIDE: axis(dim(3), label("Index1"), opposite())
  ELEMENT: point(position(bin.dot(trans1*1*Index1)))
END GPL.

Only the "flat" layout on the XGRAPH plot gave the correct rendering of poins. In "asymmetric" layout, both XGRAPH and GGRAPH showed it wrong. Given that asymmetric layout is the default in dot-plot in SPSS, that is a matter of strong concern.

------------------------------
Kirill Orlov
------------------------------