Interpreting binary logistic regression output (SPSS demo, 2018)

[@MadMax-pu1kj]

I appreciate the video but simply circling a number and telling me what the number is, does not interpret the data by any definition. Yes, I see it is 72.5%.. but is that good? bad? does it tell me something? If it is less than x, should I trust it? Should I ignore it? Is there a threshold? As I listen to the Classification Table portion, I think you keep saying the wrong thing, then correcting yourself, then confusing your terms again. Yes, we got it the 72/100 =72% but what does that mean? (Interpretation). I think the Table predicts what might happen and then also shows what DID happen, but I wouldn't know that from the explanation. I am not being disrespectful, it just feels like you are making it up as you go along. I suggest a script or, at least an outline, with the word "Interpretation" or "Analysis" in bold.

[@mikecrowson2462]

For much more up-to-date videos (with ADDITIONAL RESOURCES) on binary logistic regression, please go to the following KZbin links. 2023 videos Part 1: kzbin.info/www/bejne/nmLcnZ2Nodt6iJI Part 2: kzbin.info/www/bejne/mIeUqHx4nKtof6M 2021 video: kzbin.info/www/bejne/rJLFaoGbr92jeMU 2019 video: kzbin.info/www/bejne/maG6hIaAfdqHaNU New video on multinomial logistic regression: kzbin.info/www/bejne/i3q9eYGKiNV4epo

[@shaafiyahkhairunnisa9127]

Can u explain about the meaning of number in Coefficient B? such as in the sex variable, it is written if the result of Coefficient B is 1.025, can u explain what the meaning of 1.025? Thank you

[@mikecrowson2462]

Hello. It has been a really long time since I put this video together. With a binary predictor, the regression slope (B) is the predicted difference in log odds (Y=1) between the two gender groups. I can't recall how I coded sex, but I'm assuming it's 0=male, 1=female. In that case, if B=1.025, then females are predicted to score 1.025 points higher with respect to the log odds (alternatively referred to as logits) than males. This probably won't mean very much since the regression slopes in logistic regression represent a transform of the relationship between your predictor variables and the predicted probability Y=1. If you exponentiate the regression slope, exp(1.025) = 2.787, this is the odds ratio and it will tell you that the odds(Y=1) for females is 2.787 times that of the odds for males. I spend a lot of time going through these details in much more recent videos (and supplemental info you can download) here (kzbin.info/www/bejne/nmLcnZ2Nodt6iJI) and here (kzbin.info/www/bejne/mIeUqHx4nKtof6M). I hope this helps!

[@mikecrowson2462]

DEAR VIEWERS: PLEASE CONSIDER VISITING MY NEWEST (UPDATED) VIDEOS ON BINARY LOGISTIC REGRESSION AT kzbin.info/www/bejne/maG6hIaAfdqHaNU AND kzbin.info/www/bejne/hmG5fIuofdd5i8U . THESE VIDEOS INCLUDE LINKS TO POWERPOINT PRESENTATIONS AND SPSS DATA FOR YOU TO FOLLOW ALONG!

[@abhiswetabhattacharjee2595]

This saved my thesis! Thank you!

[@mikecrowson2462]

Hi there. I'm glad it was helpful! By the way, there is another presentation I have up as well that was made more recently (kzbin.info/www/bejne/maG6hIaAfdqHaNU). You can also download a powerpoint under the video description and it provides more description of results. Cheers!

[@sakshamarora1599]

Very informative - Can you also comment about how should one interpret Wald's statistic?

[@micah1754]

It would be very helpful to see how you would write these results up APA style. I'm so lost with that side of it

[@adandomisrael8698]

Thank you Dr Mike for this straightforward and insightful tutorial. A quick question, if we change the code labelling for gender, are we still gonna get the results as it concerns the target gender? I mean, if we coded male for 1 and female for 0? OR 1 male, and 2 for female OR 2 for male and 1 for female? In what ways will it affect the result of the analysis and interpretation?

[@ehioghaeosaze6893]

Please any response to this?

[@adandomisrael8698]

@ehioghaeosaze6893 there has been no response yet, I guess I was late to the party. But I believe that assigning the codes as factors and not numeric does the job. The result will always remain the same.

[@ehioghaeosaze6893]

@@adandomisrael8698 Thank you very much Chief. Factors as in strings please? I've got a problem where the output doesn't include all the categories in a categorical independent variable. For instance, occupation: Teacher, student, trader. The output lists just 1 and 2, doesn't add all 3 categories. Any idea how to interpret this or go around please?

[@stuartquan4735]

In my logistic regression analysis, I have a categorical independent variable which I believe shows a quadratic trend. SPSS says that this would be tested using polynomial contrasts and that the 2nd degree of freedom will show the quadratic effect assuming there is one. The output shows rows with variable, variable (1), variable (2), variable (3)...variable (8). Am I to assume that variable is the overall effect, variable (1)=linear effect and variable (2)=quadratic effect? I can't seem to find this answer anywhere on the web.

[@OskarBienko]

Share the data so we can manipulate it and then help you.

[@jakubkarwacki1214]

Great video, actually the first one that I found to be insightful enough. Thank you!

[@pupitski]

Thank you for the straight forward explanation, these videos are getting me through my advanced stats, much appreciated :)

[@pikapikacheww_]

you're my savior !!

[@mikecrowson2462]

I'm so glad you found this useful! By the way, I have a more recent video on binary logistic regression at kzbin.info/www/bejne/rJLFaoGbr92jeMU Cheers!

[@毕五二]

you are saving lives! Thank you !!!

[@cristiankebede5521]

powerpoint ?

[@v_ronyca3244]

Do you need or want there to be significance at Step O? If so, why does it matter?

[@mikecrowson2462]

Block 0 is just information related to the null/intercept-only model. There's nothing to report from it. Cheers!

[@v_ronyca3244]

@@mikecrowson2462 Thank you!

[@sarahh1007]

thank you so much!!

[@LissRAWR]

thanks a lot! :)

[@mikecrowson2462]

Glad I could be of help. By the way, I have a much better video on logistic regression (with corresponding Powerpoints you can download under the video description) here: kzbin.info/www/bejne/maG6hIaAfdqHaNU . In fact, I have a much more thorough Powerpoint on interpretation that I finished this week. You can download here: drive.google.com/file/d/1JP4oSflFxf0ZhIWC6Brthv_V5HMOQhE1/ I hope you find it useful! Best wishes

[@CHEFRidi]

Thank you!

Thank you!

[@pranavsatish1291]

great video, thank you

[@mikecrowson2462]

Thank you for visiting! By the way, I have a more recent video on binary logistic regression that I hope you visit at kzbin.info/www/bejne/rJLFaoGbr92jeMU . Best wishes to you!