S01E04-1-2c SPSS Basics - Entering ranking and pick from questions
HTML-код
- Опубликовано: 21 авг 2024
- Instructional video on how to enter two types of survey questions into SPSS: a question that asks people to rank a few items, and questions that ask people to pick a fixed amount from a list (e.g. pick two from the ten listed).
Visit the companion website at PeterStatistics...
How would you analyze the results of the rank questions?
I'd probably use a Friedman test. I discuss this test on my site at peterstatistics.com/CrashCourse/5-ThreeVarPair/ordinal/MultipleOrdinal3a.html, it also has links to RUclips videos.
Hope this helps.
can i do something like this in SPSS: ruclips.net/video/eza1XbeD2Hc/видео.html
Is this what you are looking for: ruclips.net/video/djBtLpW0Bvk/видео.html ?
Thank you very much! It is better than what I was looking for!
Glad it helped.
On a side note, RUclips has cancelled my partner program since I don't have enough subscribers, so it would really help if you subscribe (and all your friends) 😉
What is meant with the missing values column and the 999?
it's the value to enter if someone for example skipped the question. The 999 is just my own choice, you can chose any value you like, or leave it blank. I explain it in this video: ruclips.net/video/2cDujlIDXgc/видео.html , or on my site at peterstatistics.com/CrashCourse/SPSS/SPSS-2.html.
Hope this helps.
Great video!
I've been trying to find out how to analyse the data for my ranking question and hope that you may be able to help...
I used text box ranking with 10 answer options and the respondents were supposed to rank their Top 4 while leaving the other fields empty. Using custom tables I can see for each option how many voted it as TOP 1, 2, 3, 4 or didn't vote it, but with that data I'd like to have a final ranking. So which option ultimately ranks on TOP 1, TOP 2 etc.
I first thought I can use the mean, but that doesn't work as some options have only 5 votes and others have 50..
Can you tell me how to do that with SPSS?
Hmm interesting problem, and one of the reasons I don't like ranking questions :-). The mean would also have been my first thought, but if some options only have a few votes it seems a bit unfair. Perhaps add a, lets call it 'penalty score' if the option was not selected at all. So a 5 or even higher. Perhaps the number of votes regardless of which it was needs to have a certain weight. There is probably some sophisticated technique available, but unfortunately I don't know. Perhaps the people over at www.talkstats.com/ can help with this. A very simplistic solution would simply be to ignore the ones that have too few counts.
@@stikpet Yeah I already regret the way I designed certain questions in my survey :/ Thank you!
I had the idea to make the 'penalty score' 0 (just enough for each variable to reach 53, which is the max. votes one option got). This way it adds to the count but doesn't falsify the ranking score. Does that make sense?
Be careful if you use 0, since that would imply an even 'better' rank than 1. With ranking it is the lowest usually that is the 'best'. So the best is the one with the lowest average ranking. So either re-scale so that 1=>4, 2=>3, 3=>2, 4=>1 then you can use 0, otherwise use perhaps 5.
It might be a good idea to also show the average of the original. If there are big differences you can perhaps think about what might cause those.
Hi, i hope you see this. I am doing a questionnaire survey. i have a two part question in my questionnaire that asks the respondent to choose all that applies and out of the 28 choices rank any 2 in order of importance, 1 being the most important and 2 being the second most important, how will I input this data into SPSS? Thank you. Also, no respondent has all 28 choices ticked, the maximum number of choices selected I've seen so far is 6
I would split this in two. First create 28 variables, one for each option with 0 = not ticked, 1 = ticked. Similar as the example on ruclips.net/video/45nTNZYkYfw/видео.html at about 4:22. Then create two extra variables something like rank1 and rank2 with values 1 to 28 with each of the options. Then all your responses are nicely in SPSS.
Small tip: don't forget you can copy-paste, so especially with the 28 variables, create one, copy paste and then just adjust the name and label.
Hope this helps.
@@stikpet Thank you, Yes it is Very Helpful.
hi, your videos are really helpful, I have a question, I have asked participants to rate 1- 5 , 1 being less important, but there are 15 factors, obviously there would be some factors with no value, how do I show this? does this work in the same way you have shown above?
If I understand correctly, you asked participants about 15 things (factors) each to be rated on a scale of 1-5. Then each of those questions would be a separate variable in SPSS, with values 1 = fully disagree to 5 = fully agree (or whatever you used). You might also be interested on how to analyze those kind of 'list' questions. See my website at PeterStatistics.com and then under '3+ variables' and then '3+ ordinal'. Hope this helps.
thanks for replying, so I have entered the 15 factors as separate variables, and then inputted their ratings of 1 - 5, 1 being less important and 5 being most important ;across for each participant, I have been recommended to use factor analysis, but when I run this test an error has occurred, I have been recommended that selecting 5 factors out of 15 factors is what making the error occur and rather I should ask participants to rate all 15 i.e. 1- 15 ---1 being least important and 15 being most important, and then run the factor analysis,
do you think I'm running the wrong analysis or I should change my ranking scale to 1- 15?
I'm not too familiar with factor analysis, but it is used to see if for example 15 factors observed could be reduced to a fewer amount of unobserved factors. If that is what you want, then indeed you should perform a factor analysis. If you use a 5-point scale you probably have ordinal data, and most techniques for factor analyses require an interval or ratio scale. However the ordinal data is then simply assumed to be interval or ratio. More on this at stats.stackexchange.com/questions/215404/is-there-factor-analysis-or-pca-for-ordinal-or-binary-data. Changing the scale to 1-15 would make the leap from ordinal to interval/ratio a bit easier.
I'm sorry but as mentioned before I'm not too familiar with Factor Analysis (or Principal Component Analysis).
Note however if you are only interested to know which factors people tend to like more (or less), then you can 'simply' use the 1-5 scale as is, treat them as ordinal and use the Friedman test (as discussed on my site) to find out which score better (or worse), so then you won't need a Factor Analysis to my knowledge.