Statistics for Research - L11 - What are Outliers and How to Solve the Issue using SPSS?

[@sohailaftab4209]

Jazak Allah.

[@researchwithfawad]

Thanks for watching.

[@ahmedabdulahimahamud3272]

Perfect explanation, thank you very much

[@researchwithfawad]

Glad it was helpful!

[@hajarzakzouk7458]

Excellent explanation .. if I don’t have any values greater than +3 or -3 that means there is no outliers so why tukey post hoc test didn’t appear in ANOVA ?? Help me if you please

[@researchwithfawad]

You need to check it in Post Hoc Analysis.

[@mahmoudfaouzichaoubi5855]

Thank you = شُكراً

[@researchwithfawad]

Thanks. I am glad you liked it.

[@mansoralzahrani2926]

Should I consider checking outliers with Likert data? And how? Please provide a reference to refer to. Thanks.

[@researchwithfawad]

Yes with Likert Scale data. For reference Based on the recommendation of Fein et al. (2022), outliers were assessed using the Standardized (z) scores. A total of ......….. responses were outside the acceptable z score (+/-3.3) (Fein et al., 2022). The responses outside the acceptable threshold were removed from further analysis. Fein, E. C., Gilmour, J., Machin, T., & Hendry, L. (2022). Statistics for Research Students: An Open Access Resource with Self-Tests and Illustrative Examples. University of Southern Queensland.

[@mansoralzahrani2926]

@@researchwithfawad Thanks! Do I have to do it even if I will use SmartPLS for SEM-PLS? and why?

[@researchwithfawad]

Yes it is process of cleaning the data.

[@sumayyaalsefri717]

Hello, Can I use just minimum and maximum values to detect outliers as you did in the first way? without Z score? can you give me a reference to support this?

[@researchwithfawad]

Yes, you can use the box plot. Any SPSS book can be used as a reference that discusses outliers using Boxplot.

[@chantellewong1952]

Thank you very much for your tutorial. May I know if +/-3.3 the magic number? I have seen some said 2 some said 3. May I know if you have a reference to it? Thanks a bunch!

[@researchwithfawad]

Thanks for watching. Here is the reference for outliers. Based on the recommendation of Fein et al. (2022), outliers were assessed using the Standardized (z) scores. A total of ......….. responses were outside the acceptable z score (+/-3.3) (Fein et al., 2022). The responses outside the acceptable threshold were removed from further analysis. Fein, E. C., Gilmour, J., Machin, T., & Hendry, L. (2022). Statistics for Research Students: An Open Access Resource with Self-Tests and Illustrative Examples. University of Southern Queensland.

[@chantellewong1952]

@ResearchWithFawad Thank you very much Sir!

[@mahwishayanjawed4239]

Asslamalaikum Sir Is there any video for identifying multivariate outliers using Mahalanobis distance?

[@researchwithfawad]

Not yet.

[@alittlefishhh]

jazakallahu khair, but may i ask you a question. i checked my data, which is total scores of a scale used on my psych research. i followed your guide, 1. checked the boxplot, but there is no star/circle outlier, 2. but when i check the frequencies, 22.5% of my data is on the highest quartile 3. checked the histogram, but my data is not showing bell-curve you normally found on normal distribution data 4. checked for the z-score (standardised score), but all of the value is in range of +/- 3.3 (highest is +1.32, there are 25 data out of 109 total that has the EXACT same value; while the lowest is -1.9, there are only 2 data out of 109 that has the exact same value) do you think i should transform my data? please guide me :(

[@alittlefishhh]

and to add more context, when i analyze my second variable. the data is normal: 1. the histogram is kinda bell-curved in the middle 2. the highest and lowest z-score is +2, and -2 (its all different, unlike the first variable above) 3. and the significance value is >0.5 4. no outliers on boxplot

[@researchwithfawad]

Thanks for watching. Even if one of the techniques shows no outliers, that can be enough for further analysis.

[@rhafi2201]

Assalamualaikum brother, if i have 495 samples, would trimming 100 outliers is okay? Thankyou very much

[@researchwithfawad]

WAS. It is slightly on a higher side. But still if the remaining sample meets the minimum sample size criteria, you are good to go.

[@rhafi2201]

@@researchwithfawad thankyou very much for the reply brother. May i ask again if it's okay. How much is too much regarding trimming the outliers and how much is enough.

[@kamilaammar2310]

many thanks, may i have a reference to justify keeping outliers based on z score

[@researchwithfawad]

Pleasure. I am glad you liked it. Here is the reference Statistics for Research Students: An Open Access Resource with Self-Tests and Illustrative Examples by ERICH C. FEIN; JOHN GILMOUR; TANYA MACHIN; AND LIAM HENDRY, UNIVERSITY OF SOUTHERN QUEENSLAND TOOWOOMBA

[@Grace-bz3px]

Why looking the outlier from min and max value? and why value 16 is an outlier even its within the range ?

[@researchwithfawad]

Thanks for watching. Please refer to Based on the recommendation of Fein et al. (2022), outliers were assessed using the Standardized (z) scores. A total of ......….. responses were outside the acceptable z score (+/-3.3) (Fein et al., 2022). The responses outside the acceptable threshold were removed from further analysis. Fein, E. C., Gilmour, J., Machin, T., & Hendry, L. (2022). Statistics for Research Students: An Open Access Resource with Self-Tests and Illustrative Examples. University of Southern Queensland.

[@Grace-bz3px]

@@researchwithfawad does it mean its the IQR X1.5 = Outlier value?

[@MsSaddal]

Sir I used standardized z value..I got some values in minus sign.. should I remove above -3??

[@researchwithfawad]

Remove over +/-3.33

[@MsSaddal]

@@researchwithfawad thankyou so much for your instant reply

[@buzztutorials]

Nice👍❤

[@researchwithfawad]

Thanks for the visit

[@maazsatti5427]

Multivariate outliers?

[@researchwithfawad]

Hopefully soon I can do a separate video on it