EViews: (2 of 2) Diagnostic Tests on OLS and ARDL Model (Estimation and Interpretation)

Steps on how to perform normality test, serial correlation test, heteroscedasticity test, linearity test and CUSUM stability test.

[@aiman6361]

Tq sahh for helping me

[@jennifero.sebego9796]

Very clear and helpful, thank you Sir.

[@jeny_ll2084]

Thank you sir for the video, it's very helpful 👍👍

[@maryeze6254]

Thank you so so much this is all I need all in one video thank you again

[@paduraruovidiu201]

Thank you for the content. Sorted my project out :)

[@nabilumar9420]

Thank you very much sir.
But my question here is how can someone correct the cusum analysis test when the lines graph is greater than 5% critical bound?.
Thank you sir🙏

[@oluwafemidada1278]

Thank you doc

[@theamaxingnature24_7]

Thanks a lot ☺️☺️

[@chinedufavournnadozie5519]

Please how do I locate the value of R² in ARDL using e-view 9
Thanks a lot for the good job sir 👍

[@hafizsalmanishrat7940]

Sir to correct the serial correlation u said take lag
Lag of a specified variable u have taken whose variable is lag taken how lag is taken these are combined results u.e durbin watson value belongs to whole data so all variables lag is taken to correct error

[@zoyashah7826]

Sir whether we should mention probability value of F-statistic or observed r square value of diagnostic tests in our research paper ??

[@krishnaiyer2556]

excellent

[@tundeomotehinse514]

Can I use this diagnostic test (ARDL) on the panel data analysis ?

[@SonomNayon97]

For ARDL model testing for multicollinearity is not necessary , right?

[@francischisenga1838]

please add dummy variables to the model and how do you test

[@tinaparate5400]

1) How can I conduct a CUSUM & CUSUM SQUARE test using 10% Significance level?
2) Are both essential? Or is it enough if my CUSUM Test is stable?

[@obezipacademy]

To conduct cusum test at 10 percent, right click where the 5% is displayed on the cusum result, then click on options, click where u see 5% Significance and edit to 10%, then click ok.
Both are actually good to test, though cusum of square is further used to ascertain structural break dates in a model. Some scholars just report cusum without really conducting cusum of squares

[@tinaparate5400]

@@obezipacademy Thank you so much for your answer! Though both 5% & 10% significance CUSUM & CUSUM SQUARES plots are coming as the same (same magnitudes, same plots). I thought with 10% the band-width should have become wider.

[@obezipacademy]

Discuss the outcome in terms of structural break that happened from the point of divergent

[@zoyashah7826]

Sir for estimating ARDL is there a requirement of any minimum no of observations.I have used 28 yrs data.is this enough.

[@obezipacademy]

28years is large enough

[@simbarashemuchakazi2321]

Do we have ARDL on eviews 8

[@zoyashah7826]

Sir whenever I m estimating my equation through Ardl its saying singular matrix..how to resolve this problem

[@obezipacademy]

Reduce the number of lags to either 2 or 3

[@zoyashah7826]

Sir I have estimated the equation with 2 lag..but there is one problem.my error correction model is coming negative as u said but all my explanatory variables are insignificant both in the long and short run.what to do regarding this??

[@obezipacademy]

You have to report and interpret the results the exact way you go it. Explain the possible economic reasons why it's insignificant

[@zoyashah7826]

Sir my error correction model is coming far below 0..its coming -2.what to do regarding this??

[@obezipacademy]

@@zoyashah7826 it simply means there is overconvergence. This means that the correction mechanism is oscillatory. in other words, the speed of adjustment fluctuates forward before settling to equilibrium. (note, the frequency of your data as well as outliers can play a role in this but by no means your ECM is wrong

[@degsewtube8701]

please attach the data here; so we can practice it with you.

[@jemmyrabaye2254]

what if there is serial correlation is there a video of how to correct it

[@obezipacademy]

ruclips.net/video/hM8lXyGu_mk/видео.html

[@angelicaantoine7688]

@@obezipacademy even if I'm using a first difference data I can use this technique to remove the serial correlation problem

[@obezipacademy]

Yes.

[@angelicaantoine7688]

@@obezipacademy thank you a lot

[@zoyashah7826]

Sir kindly share your email I'd.. I want to share the data with you so that you can tell me what wrong m doing in estimation

[@obezipacademy]

[e-mail]