SPSS - Stacked bar chart of multiple variables (via restructure)
HTML-код
- Опубликовано: 30 июн 2017
- Tutorial on how to create a stacked relative bar chart of multiple variables that each use the values. This alternative method restructures your data, and then uses the chart-builder or cross tab to create the chart. If you have Custom Tables a faster method is explained at • SPSS: Stacked Relative...
Companion website at PeterStatistics.com
Donation welcome at Patreon: www.patreon.com/bePatron?u=19...
Without exaggeration one of the most useful things I have ever found on RUclips
thanks. glad the video helped.
that was great. thanks
I have the basis version of SPSS v.27 and your method just worked perfectly, thanks a lot!
Glad it helped
Thank you for this vid, it helped me greatly
YOU SAVED MY LIFE THANKS VERY MUCH
You're welcome and glad I could help. Good luck with your research/study.
Thanks so much, I learnt a lot from your video.
glad it helped.
Awesome video Dude! I was struggling to get this output :)
Glad it helped.
On a side note, RUclips has cancelled my partner program since I don't have enough subscribers (currently around 797, should be 1000 😞), so it would really help if you subscribe (and all your friends) 😉
Awesome video it solved all my problems
glad it helped. Good luck with your analysis.
This just solved my life goddamn, thank you!
No problem. Never expected to solve someone's life with this :-)
I love you so fucking much, man. This is *exactly* what I was looking for. Thank you and stay healthy!
Glad it helped. Good luck with your research/analysis/study, and indeed stay healthy.
Great video, thanks a lot! Just wondering, we are comparing to different groups (to different hospitals), is it possible to display results from the to different groups in the same graph?
Thanks.
What you could look into is in the chart builder, there is a tab 'Groups/Point ID', there you can either select the 'Rows panel variable', or 'Columns panel variable'. This gives a new option in the preview area where you can then move the grouping variable to (in your case the one that determines which hospital).
You might then also want to check which percentage you want to see, so check those again.
Hope this helps.
Very useful, thanks sir
You're welcome, and glad it helped.
really helpful!
glad it helped. good luck with your analysis.
In this video you restructured a few variables into 1 group but is it also possible to use the same technic to make multiple groups? Because I made 1 group and finally got the graph like I needed it but than I realised I still had to make 10 other groups for my other graphs so I tried the same proces and yet it didn't work. I just don't know how to get there.
Not really sure what you mean. It should work with different groups, but you'll have to repeat all steps each time. Alternative, you might have the option 'Custom Tables', in that case you could try the method described in: ruclips.net/video/djBtLpW0Bvk/видео.html
Hope this helps.
@@stikpet Thank you. I will try that.
Love this video, but I have a question ? Can you show percent and count on the bars when you create graph from the custom tables method ?
when you have the graph double click on it to open the chart editor.
Then click in the menubar from the chart editor on 'Elements' and then 'show data labels'.
In the properties window you can then move the count from the 'not displayed' to the displayed section (the percent should already be there).
Hope this works for you.
@@stikpet Thanks Peter, I can get it working in the Restructured method , but I am wondering how to get it working with the Custom Table method, I have a large number (20 or so ) of themed likert scale questions so I don't want to have to restructure 20 times. For this reason, I am using the Custom Table method, but when I use this method, I can only show Count or Percent, but not both. I am wondering if there is a way to show both using this Anyalyse /Custom Tables method ?
@@margaretforde1721 sorry, I also only seem to get either count or percent using the custom tables.
Good
Thank you!!!!!!!!!!
You're welcome. Glad it helped.
I used the custom table method for the frequencies. But for my thesis I have to compare the variables with different groups. Can I use this restructure method to do cross tabs with fisher test with for example male vs female?
If you have a binary variable (male vs. female) and an ordinal variable, you could use ruclips.net/video/P38-UDrzIgA/видео.html and for the test perhaps a Mann-Whitney U test. peterstatistics.com/CrashCourse/3-TwoVarUnpair/BinOrd/BinOrd-0-Introduction.html
@@stikpet Thanks. the problem is, my categories are from a multiple response question. The respondents had to give themselves a role and could have multiple roles per respondent. how do I compare this?
@@leondfeijen uhm so something like ruclips.net/video/A49xaW6rvxs/видео.html then?
@@stikpet almost. This only shows the count of the multiple response. I don’t know how to explain it shortly, so i will let it go for now.
a while ago someone else had a similar question where my reply was:
What you could look into is in the chart builder, there is a tab 'Groups/Point ID', there you can either select the 'Rows panel variable', or 'Columns panel variable'. This gives a new option in the preview area where you can then move the grouping variable to (in your case the one that determines which hospital).
You might then also want to check which percentage you want to see, so check those again.
Perhaps that's something for you as well.
My data is multiplied with 8. Is it possible to change it back?
Do you mean your original data as shown in the data view, or only when you see something in the Output?
In case of the first, then I recommend first to save your file under a new name. Then either try to use 'Transform - Compute variable'. You can then enter a new name for a variable, and in the numeric expression part move the variable and divide by 8. Unfortunately you would have to do this for each variable. It is possible that this might be possible as well to do for all variables but then by using SPSS syntax, and I'm not too familiar with that.
You could also then perhaps copy your data to Excel, divide all numbers there by 8 and then copy them back.
In case of the second, then it is probably that you somehow selected 'weight cases' somewhere, select 'Data - Weight Cases' and select 'Do not weight cases'.
Hope this helps. Perhaps if you still have the SPSS syntax (the log bits in the output) it's possible to trace back what went wrong and how to reverse it.