Two-way mixed ANOVA on SPSS
Вставка
- Опубліковано 4 жов 2016
- Learn how to conduct a two-way mixed ANOVA using SPSS. In this example we ask: Who eats more - male or female monkeys? Do they prefer red or yellow fruits?
We look at how to report the results in accordance to APA guidelines. (Note typo in example: "p ⋗ .001" should be "p ⋖ .001")
Want to quickly and substantially improve the quality of your writing? If you are a student or professional in psychology/social science, it's likely that you will be expected to follow APA guidelines. Learn the 30+ most common mistakes with the APA Style and Referencing guide:
amzn.eu/55oGkHM
Are you an aspiring social or natural scientist? If so, an understanding of statistics is essential! Subscribe to the channel for videos on how to complete a multitude of statistical tests, and like our facebook page for even more content!
SPSS-Assassi...
![ОРБАН приехал к ПУТИНУ и "отчитался" о поездке в Киев 😁 [Пародия]](http://i.ytimg.com/vi/_0ukDovOWq8/mqdefault.jpg)
I've been working on trying to perform statistical analyses on a set of data for well over a week, over 10 hours probably. This 10 minute video done more than all of my uni research methods lectures. Simple, easy to follow and effective. Thanks a million!
Thanks, glad it helped!
Literally
You saved my ass, thank you! Assignment due in 16h and you explained it better than any of my lecturers. Thank you so much!
Thanks Beatriz, glad I could help!
Thank you so so much. You have explained this SO WELL. Saved my life & my lab report!!! (Thank you for explaining exactly where you have found the statistics and how to interpret them)
Really appreciated the hint that we should always report statistics to 2 dp as thats the APA format!
Thanks Antonia. Glad it helped!
Absolute lifesaver! Thank you for explaining in such a simple to follow way.
Thanks Alexandra!
Thank you! Extremely helpful. Keep up the good work.
Thanks Michael!
so helpful, thanks so much!
Thank you so much!!
This video is extremely helpful for my final report
Glad I could help!
Very helpful thank you!
Thanks for this bro, really helpful, do you have any for other statistical tests also: Mainly - T test (Paired & singular samples), (One Way, Two Way) ANOVAS, ANCOVAS/MANOVA's, Multiple Regression Analysis & Thematic Analysis?
Glad it helped! I've got videos about most of those on my channel: www.youtube.com/@DavidRobinsonPhD/videos
Great video. Thank you so much.
Thanks Shreya, glad it helped!
Thank you!!!!!!
Hi, what if you get a significant within-subjects interaction effect, but the between-subjects effect of the IV is non-significant? What does that mean?
Hi! Great video, thanks. I would like to know how to handle adding covariates to the model. When i add my covariates, I get an interaction between the within-variable and the covariates, which i do not want since the within-variable is my moderator. This is also influencing the significance of the within factor. Do you have any idea how to handle this?
Thanks Eline! Sorry, I haven't come across that situation before, so am not sure what to suggest. All the best, David.
What happens if the box plot sig. Is less than 0.05? Could I this two-way anova method for: 3 conditions independent variables (interviewer tone for children) and then 2 types of questions used in the interview (open or suggestive questions).
Hi Millie, thanks for your question. Some suggest that you can use an alpha of .001 for Box’s test rather than .05 (e.g., Pallant). I.e., they say that the assumption is only violated if the sig value is below .001. Yes, your IVs can have more than two levels.
Thank you so much! Is there an alternative to this test, if you fail the assumptions on Box's test (probably as my sample sizes in each group are quite different)?
Hi,
There's not an alternative that I'm aware of.
If Box's test is significant, though the p value above .001, you might be able to argue that this is not problematic as some researchers (e.g., Hahs-Vaughn, 2016) have argued that the test is overly sensitive when applied to large samples.
All the best,
David.
Hi, Your example was pretty good. But I have a question. Can I use this same concept but with four variables ? For example, age congruence, age incongruence, race congruence and race incongruence. And how would I run it ?
Hi Renan,
You can run four-way ANOVAs, though they’re rare because the interactions become difficult to interpret. However, are you sure that you don’t have two variables (age and race), each with two levels (congruent and incongruent)? If so, then you can do a two-way ANOVA (as in the example). How exactly you set up the file and run the analysis depends on whether you’re using just between-subject variables, just within-subject variables, or both (as in the example).
Best wishes,
David.
Dear David, thank you for your reply. Yes, I suppose in this case a two-way between-subjects model is a more accurate (and correct) method to use. The initial design of the research however is more of a mixed model, a 2 x 3 x 2 model with age (young, old), traits (friendly, grumpy, indifferent) and fondness of children and orphans (fond of children, fond of orphans). The first two IVs are, as have been specified in my previous comment, between-subjects factors while the third IV is a within-subjects factor. People rate on how much they like children and orphans respectively, with each person specifying both their fondness of children and orphans with two respective scales with similar index ranging from 0 (not at all fond) to 100 (extremely fond). Hence a person in this case might, for instance, not like children at all (0) but would be extremely fond of orphans instead (100), or visa versa. The dependent variable being observed is labeled "empathy toward younglings", in this case signified/measured by individuals' ratings of the two levels of the within-subjects factor IV, and is a continuous variable due to the rating system.
However due to time constraints and lack of resources, this particular model had to be modified slightly by removing one of the two levels in the within-subjects factor IV, leaving it with only one level, fondness of children. The dependent variable is still the same, whether you now want to call it empathy still or instead fondness of children to better suit the single level of the IV. I've been reading here and there and I suppose it's not possible for an IV serving as a repeated measures factor to have only one level as in my planned model, and must at least have two just as in my initial model. As such there is no such thing as a repeated subjects model with a "x 1" in its arrangement. Is this assumption of mine correct?
If you only asked about fondness of children (or fondness of orphans, i.e., not both), then that third IV doesn't exist (a variable is something that varies, which isn't the case if there is only one level), and you would just have the other two between-subjects IVs and a DV, so a two-way between-subjects ANOVA would be suitable.
@@DavidRobinsonPhD I see, thanks David! Your videos and responses have been of great help
@@DavidRobinsonPhD hello again, David. I have another question which I hope you can clear up for me. So I suppose for my final 2 x 2 between subjects model with age (young, old) x traits (friendly, grumpy), no control group is involved and both age and traits groups serve as experimental groups? Also, then the DV being fondness of children (represented by rating) serve as the single treatment given to all groups, is that correct?
I've been reading on the internet and been told that between subjects model normally have at least one control group, which my model apparently does not have? And between subjects models also normally have more than one treatments, every one of which are assigned to one group respectively.
@@user-sy9ev8hh5k Yes, research employing 2 x 2 between-subjects designs often has treatment and control groups. I.e., some participants are given a treatment (e.g., a medicine) and some are not. However, since your research doesn't have treatments, the terms "treatment" and "control" aren't really relevant (which is fine). Instead of different treatment conditions, you have young and old and friendly and grumpy.
@@DavidRobinsonPhD Thank you for your previous reply, David. I am fairly new to statistical analyses methods, and still am confused with some things. Currently, I am torn between conducting my experiment in a two way 2x2 between subjects anova, or a one way anova with four groups. As you know by now, my experiment primarily involves two characteristics whose roles I'd like to investigate: age (young, old) and traits (friendly, grumpy). Now I know, from your very helpful explanation, that I could conduct a 2x2 anova to find out. This analysis will give me info on main effects of both variables, one for each, as well as their interaction, correct? However, I'm only interested in the interaction between the two in terms of what I'm trying to investigate and not single effects of the respective variables. That is, I'm not interested in finding out whether age or traits alone predict fondess of children. In other words I do not care if fondess differ between young and old individuals, or between grumpy and friendly individuals. Rather, what I'm interested is to find out whether or not the fondness will differ between young grumpy individuals and old grumpy individuals, old grumpy and old friendly, etc. Only the interaction and not the main effects. Would it be better if I continue with the 2x2 anova just to find the interaction while ignoring the main effects, or should I resort to one way anova with multiple groups (mine produce six comparisons thereof) instead?
Hi what if you had different sample sizes for males and females e.g., 15 females and 20 males. Can you do this test then? thanks so much
Hi Niha,
It's understood that equal sample sizes are preferable, though this doesn't mean that the test is not permissible if they are not equal. Unfortunately, there's no rule about how much of a difference is acceptable.
Kind regards,
David.
Hat's off. i got it.
Hi David, thank you for the video. I am doing a three way 2x3x3 design with two between group variables and a within variable. The between group variables are 2 music genres (rock and classical, although I have a control condition where no one listens to music so would that make that 3?) and 3 tempos (very slow, slow and fast) and my within variable is different list lengths of words (4 word length lists, 9 word length lists and 15 word length lists). Am I right in thinking this needs a mixed ANOVA? I am doing a pre-registration report for part of a module at uni. We have also been asked to include planned comparisons. If I predicted something, what would be the appropriate tests to use for planned comparisons? Many thanks!
Hi Emma, thanks for your question. Yes, you could think of your genre variable as having three levels given that there's a no-music condition. As you have between- and within-subjects variables, a mixed ANOVA would be suitable. Re. the planned comparisons, t-tests would be an option.
@@DavidRobinsonPhD Thank you. I have since considered changing my design to a two-way 6x3 design (so I have 6 music genres and 3 word list lengths). If I were to administrate a mood questionnaire, how would that affect the design please? Would a mood questionnaire be classified as a within-subject variable as all my participants will do it?
@@emmawilliams3416 It depends on what you plan to do with the data from the mood questionnaire. For example, if you categorised participants as being high or low mood and then assessed whether how many words they remembered (I’m just guessing that this is what you’re measuring) differed between the high and low group, mood would be a between-participants variable. If you wanted to control for mood in an analysis that looks at whether the number of words remembered differs between genres, it could be viewed as a covariate. If you assess whether mood differs between music genres, it would be a dependent variable (the terms “between-subjects variable” and “within-subjects variable” are usually applied to independent variables).
Hi,
How would you report the results of Mauchley's Test of Sphericity in you analysis? The "p value" includes no numbers, just the same as mine, and I don't know how to report this value. But as I see in your analysis, we should get from that table that the sphericity assumption of ANOVA was met.
Hi Seray,
Thanks for your question and apologies for the delayed response.
In short, the test isn't relevant if the repeated measures variable only has to levels.
You can find out more here: www.researchgate.net/post/Mauchlys_test_of_sphericity-Why_no_results
Best wishes,
David.
@@DavidRobinsonPhD Thank you, David. Appreciate it!!!
I am having trouble running this analysis on my data. Maybe I set it up wrong? I have two groups and each group took 2 tests. There are 4 test in all. And I'm measuring the time differences. SO I only have 3 columns. First is the Group (1 or 2) and the second is Test (1,2,3, or 4) and the third are all the times. Is this correct?
Hi Alexandra, for within-group variables (which "test" seems to be in your experiment), you need to have a column for each of the levels of the variable. For example, if you had two groups (between-group variable) who each took two tests (within-group variable), you would have one column for group (and would use numbers [e.g., 1 and 2] to indicate which group someone belonged to) and would have two columns for each of the tests, in which the scores on the tests would be entered.
Hi, what if you had a third fruit (e.g. green) would this be possible? Thanks
Hi Megan,
Thanks for your question.
Yes, you would just need to add another column for this third fruit. The analysis would then be a mixed 2 (sex: male, female) x 3 (fruit colour: red, yellow, green) ANOVA. The process of running the analysis would essentially be the same.
Best,
David.
but if there is a signtificant interaction p should be less than .001 right? you saying it too. so it should be F(1,18)=31.78, p
Hi Andrej,
If a p-value is below .05, this is typically considered to indicate a significant result. If you see a p-value of .000 in SPSS, since values are only shown to three decimal places, all you know is that the real value is something below .001, hence this is reported as p < .001. Since .001 is less than .05, this indicates a significant result. In the video, the p-value in SPSS for the interaction is .000 (which would be reported as p < .001), which indicates that there is a significant interaction. Yes, the other values would be presented as you suggest in your comment.
Hope that helps,
David.
Where are you getting the numbers for males (M=17.1) and females (M=13.05)? Are these the means you are comparing?
Hi Keng,
Thanks for your question. Yes, that's right. You can see the table from which the means are derived towards the end of the video:
ua-cam.com/video/-APq5vUgi6U/v-deo.html
Best wishes,
David.
hi my box test of equality is showing as 0.008 how would i record my results. this is for my dissertation
Hi Jordan,
Some authors (e.g., Julia Pallant, in her “SPSS Survival Manual”) suggest that the Box test is very sensitive, so propose that a violation is evident only if the sig value is below .001. Therefore, you could argue that your value of .008 does not indicate a violation.
(I'm assuming .008 was in the Sig column.)
You could write something like this:
“Using an alpha of .001, as suggested by Pallant (2011), Box’s test indicated that the assumption of homogeneity of intercorrelations was not violated, p = .008.”
If you wrote something like this, you would need to include Pallant in your reference list.
Best,
David.
What about simple effect analyses? I know you mentioned post-hoc t-tests.
Hi,
Thanks for your question.
Yes, if the ANOVA is significant, you could use the Split File function ("Organize output by groups") in the Data menu to perform follow-up tests that focus on individual levels of one of the IVs (e.g., compare red and yellow fruit consumption in males or females only).
Best,
David.
@@DavidRobinsonPhD That's helpful, thank you David!
what if i have put e.g red and yellow on a single column? now i can't fill in both of the within subjects variables
Hi Luke,
If you have a between-subjects independent variable (sex in the video), use a single column to specify which condition each participant belongs to, then enter the scores for the dependent variable (number of fruits eaten in the video) in a separate column.
For within-subjects independent variables (fruit colour in the video), it's necessary to have one column per condition/level (red and yellow in the video). Therefore, rather than using, e.g., 0s and 1s within these columns to specify the condition/level (which is what you do for between-subjects variables), you enter the scores for the dependent variable.
Hope that helps, though let me know if you have any other questions.
Best,
David.
Is there a non-parametric equivalent of this test?
Hi Carol,
Thanks for your question.
For one-way ANOVAs, there are non-parametric equivalents (e.g., Friedman test, Kruskal-Wallis test), though not for mixed ANOVAs.
Best wishes,
David.
Just a quick one , are you implying that if the box test of equality matrices and the levine test is significant i.e p
In my case the levine test my p value is 0.068. but the box test of equality matrices the p value is 0.000.
Hi Joseph, if the results are significant (i.e., the assumptions are violated), you could still report the results of the ANOVA, though it would be necessary to say that they should be interpreted with caution. When doing this, researchers will often cite others who have claimed that ANOVAs are robust to certain violations of assumptions. Alternatively, non-parametric tests, which make fewer assumptions, could be considered instead of an ANOVA.
What if we have just this design but with two dependent variables? both of them ordinal. And we assume that participants who will score high to the first DV will also score high to the second one? Thank you!
Hi Maria,
Thanks for your question.
The type of analysis you need to run will depend on your hypotheses/research questions. However, if your study is very similar to the one in the video, though has two DVs, you might think about doing a MANOVA. The data in the video could be analysed with a MANOVA, which would tell you whether one sex eats significantly more fruits (regardless of colour) than the other, whether one sex eats significantly more red fruits (1st DV), and whether one sex eats significantly more yellow fruits (2nd DV). To run this, instead of going to Analyze, General Linear Model, Univariate, you would instead go to Analyze, General Linear Model, Multivariate.
Kind regards,
David.
Thank you very much for the response and for your time!
Actually i need to test that female participants (between group factor), who identify (fisrt dependent. accuracy scores) female faces (within factor) are score also higher in emotional intelligence(second dependent).
Do you think that the MANOVA still a good analysis for this? Or i have just to do seperate mixed factorial ANOVA, one for emotional intelligence and one for accuracy scores?
Kind Regards,
Maria
For ANOVAs and MANOVAs, you normally need to have categorial/nominal independent variables. E.g., sex is a categorical variable comprised of two levels: male and female. It sounds like you have two independent variables (related to participants and stimuli); what are the levels of these variables? E.g., is it that you have male and female participants (between-group factor) and male and female stimuli/faces (within-group factor)?
One thing to note is that it isn't really accurate to describe emotional intelligence (EI) as a DV as this is a characteristic of the participants that won't be affected by the IVs (conversely, their accuracy scores might be affected by the IVs). If you want to explore the relationship between accuracy and EI, a MANOVA wouldn't be suitable. However, you could consider an ANCOVA, which would tell you whether there's a significant relationship between accuracy and EI. To do this, follow the steps in the video, though enter EI into the Covariate(s) box in the window that opens after going to Analyze, General Linear Model, Univariate.
Thank you very much!!!
Thank you very much
You made my day
My pleasure!
@@DavidRobinsonPhD sorry where did you get males and females M
Got it 🤦🏼♀️😂
Great explanation!
If you don't mind me asking for some advice, I have a study design with 4 IVs (1 between and 3 within). I did a 4-way (2 x 2 x 2 x 3) analysis in SPSS, but was recommended to do a 2-way (2 x 3) analysis with my between-subjects and within-subjects variable. how would you recommend doing this analysis if I currently have 12 values for each of my participants?
Thanks Charlie! If you've been asked to ignore two of the within-subjects variables, can you just complete the analysis with the columns that relate to the one remaining within-subjects variable and with the one between-subjects variable?
@@DavidRobinsonPhD Thanks for your reply David! No, I'm unable to, because my study was set up to have 12 different values recorded since having 3 within-subjects made it 2 x 2 x 3. So if I try to follow the method in your video with one 3 level within-subjects variable, I'm unable to move the 12 variable (columns) values to the 3 levels (Within-Subjects Variables) in Define on the right.
Hence, I was wondering if you know of any other analysis or method I could use.
@@charliecopperfield8273 What if you created three new columns that contain means based on the existing columns (e.g., a column that contains means of the values in columns 1-4, a column that contains means of the values in columns 5-8, a column that contains means of the values in columns 9-12)? You could then use the three new columns when running the analysis. If you want to try that, this video shows how to do it: ua-cam.com/video/Ur2vpo3SofE/v-deo.html&ab_channel=BrunelASK
@@DavidRobinsonPhD Thanks for the suggestion. I did consider that, but it's difficult to take the means as the 3 within IVs are quite different from each other. I managed to find a workaround though!
Is there such a thing as a mixed model anova (or any model of the analysis for that matter) that has one or more independent variables with only one level? Suppose a 2 × 3 × 1 model with age (young, old), traits (friendly, grumpy, indifferent) and type of fondness (fondness of children). The first two independent variables are between subject factors. The dependent variable being observed is how much a person is fond of young children, presented as a continuous variable measured on a scale of 0 (not fond at all) to 100 (extremely fond). The dependent variable is really just numerical data specifying that one independent variable with only one level (fondness of children). Is this a correct model? Or is it better to represent that independent variable as just the dependent variable since they're both in this case refer to the same thing?
Hi, thanks for your question. If you want to examine whether fondness of children (DV) is impacted by two between-subjects IVs (age, traits), you could use a two-way between-subjects ANOVA. However, if you had two between-subjects IVs, a continuous DV, and another continuous variable, you might consider a two-way between participants ANCOVA, which would be the same as the ANOVA, though the additional continuous variable would be included as a covariate. Hope that helps!
Thank you so much . When we have 5 response variables and 1 independant variable with 4 levels, 4 replications in each levels, all measured at 6 time intervals... still mixed design anova could be done??? what else has to be done for the study to be effective??
Hi Gowri,
Thanks for your question.
If you have five response/dependent variables, you’ll likely need to run five tests (e.g., run five ANOVAs).
If, however, your response variables are all related to a single concept, a MANOVA might be appropriate. For instance, if you were interested in academic performance, you could assess English and maths ability (i.e., two response variables), both of which are measures of academic performance. In this case, a MANOVA would be suitable.
If your response variables don’t fit into a single category as in the above example, an ANOVA would be better.
Assuming that you’re interested in how your response variables vary across time, it sounds like a mixed ANOVA would be a good option.
However, if you were to calculate means scores for your response variables by adding together the values from the six intervals then dividing by six, your experiment would no longer have a repeated measures aspect, so a one-way independent ANOVA (i.e., not mixed) would work.
To summarise, a mixed ANOVA might be suitable for your purposes, though you would need to consider the above assumptions to make this decision.
Regarding other steps to take, if your ANOVA was significant, you would probably also complete post hoc tests, which would allow you to determine whether there were differences between all or only some of the levels of independent variable. E.g., there might be a difference between levels 1 and 2, though not between 2 and 3.
Hope that helps!
David.
Thank you very much for your detailed explanation, this study is based on measuring growth parameters( 5 response variables) of plants over time(6 intervals)in various treatments(4 levels). All measures the overall health of plants, and also measured at various time points. Each sample plantlets randomly allocated(all from same mother plant- cloned), but all varied marginally at their baseline values (response variables). Considering all these, which analysis should I go for?? I also have other growth parameters, some measured only at pre and post and others measured only as one final measure(post..). can I use one way anova for these (prepost difference and compare their means )???
Hi Gowri,
If all of your response variables measure the health of plants, a mixed MANOVA might be appropriate (i.e., a single statistical test).
An alternative would be to do mixed ANOVAs for each of your response variables (i.e., five statistical tests).
If you have a supervisor for this project, perhaps run these options by them, as they’ll have a better understanding of your specific field.
For the measures that were only taken before and after treatment, you might consider t-tests, which are suitable if your independent variable only has two levels; if the independent variable has more than two levels, consider using a one-way ANOVA.
All the best,
David.
Thank you very much for clarifying each and every one of my doubts!. I really appreciate your time and help!...
Hi. What if the Box's test and Levene's test are both significant? :(
Hi,
Thanks for your question.
You might find the below site useful. (The focus is on MANOVAs, but the comments will apply to ANOVAs also.)
www.researchgate.net/post/How_do_I_continue_with_my_analysis_if_the_Boxs_M_and_Levenes_tests_are_significant_in_MANOVA
All the best,
David.
Perfect explanation sir. Dear sir, your suggestions required. I have collected the data, and, I want to compare the groups of (male and female teachers) with pre and post conditions. What kind of analysis or test you suggest. Thanks in advance
Thanks Zeb! It sounds like you have a between-subjects independent variable (sex) and a within-subjects independent variable (time), in which case a mixed two-way ANOVA would be appropriate.
@@DavidRobinsonPhD Thank you so much sir, I will try the way you suggest. Thanks for your time and suggestions. Be happy
When I typed in 0 and 1 in the sex column on the data view page the numbers didnt automatically change to males and females. Help?
Hi, you just need to select "Value labels" in the View menu.
@@DavidRobinsonPhD I see, thanks!
an assumption is no outliers, how can i detect and remove outliers please? don't know if i should do studentized residuals or box plot or something else...
Hi Aliya, thanks for your question. Yes, box plots can be created in SPSS and used to identify outliers. I haven't yet made a video on how to do this, though if you search on UA-cam, e.g., "box plot outliers SPSS" you should be able to find a video that can show you what to do.
David Robinson, PhD great thank you 😊
In my research I have two treatment groups and in each group I have taken measurements of pressure at 3 intervals ie 0, 2, 4, 8 weeks. I want to compare significance between 2 groups. Which test to use? How to say which treatment is better than other?
Hi Fazeel,
Thanks for your question.
Given your design, a two-way mixed ANOVA would be appropriate. It would tell you whether there is an interaction effect, a main effect of group, and/or a main effect of time. A main effect of group would just mean there is a difference between the groups when looking at the data from all the time points together. Therefore, if you obtained a main effect of group, you might consider following this up with independent t-tests for the individual time points to get more specific insights.
Hope that helps!
David.
@@DavidRobinsonPhD I tried repeated measures two way anova. But does this method give us significance value between 2 groups as we get from 1 way anova?
Or should I run a 1 way ANOVA for each measurement comparing 2 treatments?
Thanks a lot for your help
@@fazeelzubair2787 "Repeated measures" is another term for "within groups", so a repeated measures ANOVA doesn't provide insights into between-group differences. However, even though the "within groups" option is selected in SPSS when you run the mixed ANOVA described in this video, the analysis is actually a combination of within- and between-group ANOVAS (hence mixed). Therefore, it does provide insights into between-group differences. Regarding, one-way ANOVAs, these can be within-groups (i.e., repeated measures) one-way ANOVAs (which look at differences within groups across time points) or between-groups one-way ANOVAs (which look at differences between groups at one time point). One-way ANOVAs are usually only used when the IV has at least three levels. Therefore, in your case it would only make sense to do within-groups one-way ANOVAs because there are at least three time points (though this wouldn't provide insights into between-groups differences). Conversely, the between-groups variable (treatment) only has two levels, so it wouldn't make sense to do a between-groups one-way ANOVA.
@@DavidRobinsonPhD thanks a lot for detailed explanation. I will use 1 way ANOVA at each time level to explain which treatment is better... this way I can also tell at what point of time did effect start
i love you
its not changing the numbers to letters for my nominal value :(
Do you mean that the numbers you entered into the values column in variable view don't appear as descriptions of the conditions when entered in data view? If so, select Value Labels in the View menu.
@@DavidRobinsonPhD thank you! i did this, i was wondering if you could offer some advice? im researching loneliness levels within autistic and non-autistic people, before and during covid. Diagnosis is the between and loneliness before and after was within. My within is not significant .247, but my between is 0.44. I dont understand what this means
@@apm7653 That would suggest that, when looking at the whole sample, loneliness didn't change significantly between the two time points (before and during). However, the results also suggest that there is a difference between the autistic and non-autistic individuals (when looking at the data from both before and during). If the interaction is significant, this would suggest that the relationship between time point and loneliness is influenced by whether or not the individual is autistic or non-autistic.
@@DavidRobinsonPhD sadly the loneliness_timescale * Diagnosis isn’t significant, however, autistic people have higher levels of loneliness. So I guess I confirmed at least one hypothesis. thank you
@@apm7653 No problem!
please help me :( im struggling to input mixed design data
pls :(
Hi,
Is there something in particular you're struggling with?
Remember that for within-group variables you need to use a separate row in variable view for each level/condition and that for between-group variables you only need one row, but need to use the values column to describe each of the levels/conditions.
Best,
David.
@@DavidRobinsonPhD i understand but i don't know how to input this. i have 50 participants in one condition and i test them after 5 mins. However i also test 25/50 of those participant after 1 week as well
i have 4 conditions= testing if they remember 2 words or 4 words after 5 minutes and 1 week delay
@@jamilaudz Are all of the participants tested on both 2 words and 4 words or are some tested on 2 words and others tested on 4 words?
@@DavidRobinsonPhD 50, 2 words and 50, 4 words
What if normality is violated and transforming doesn’t fix this? What is the non-parametric model?