Springen naar inhoud

Transforming skewed variables for Normality

  • Dit onderwerp is gesloten Dit onderwerp is gesloten




  • >25 berichten
  • 86 berichten
  • Ervaren gebruiker

Geplaatst op 17 juni 2012 - 18:34

Dear Everyone!
I am having a hard time transforming some negatively skewed variables for my Msc Thesis. I have tried several transformation (such as square root, log10, inverse, reflect and square root, reflect and Log10 and reflect and inverse), but none of these seem to working. I also tried different values for K (=largest possible value of scale +1), but again no appropriate sig for the kolmogorov-smirnov statistic.
I keep on getting significance values around 0.013, (with 0.017 being the 'best' until now).

For the remainder of the data analysis i planned to use correlations and multiple regression, which apparently are not suitable for non-normal distributed variables.

Does anyone know how to fix this? What are transformations are there which I can still try?
Also, for correlations and multiple regression I am wondering whether both, my dependent variable and independent variable have to be normally distributed or only my DV? (I am having problems with both)

I have attached the histogram, such that the distribution is visible as well as the tests for normality.

I really really appreciate your help, and if its in Dutch, it doesnt matter! any answer is highly welcome,

thanks a lot!!!
Screen shot 2012-06-17 at 7.31.15 PM.png Screen shot 2012-06-17 at 7.31.18 PM.png Screen shot 2012-06-17 at 7.31.28 PM.png

Dit forum kan gratis blijven vanwege banners als deze. Door te registeren zal de onderstaande banner overigens verdwijnen.




  • >5k berichten
  • 10217 berichten
  • Moderator

Geplaatst op 18 juni 2012 - 09:50

Dit is een Nederlandstalig (wetenschaps)forum. Bijgevolg verwachten we ook dat de berichten op dit forum in het Nederlands worden opgesteld. Dit topic wordt daarom gesloten.

Since this is a Dutch forum, we expect the messages to be in Dutch. Furthermore, you have proven in the past to be able to express yourself in Dutch. For these reasons, this topic has been closed.
Zoek je graag naar het meest interessante wetenschapsnieuws? Wij zoeken nog een vrijwilliger voor ons nieuwspostteam.

0 gebruiker(s) lezen dit onderwerp

0 leden, 0 bezoekers, 0 anonieme gebruikers

Ook adverteren op onze website? Lees hier meer!

Gesponsorde vacatures