Specification Error

Assalammualaikum! Gud nite semua..eh! jgn tido lagi yew..malam ni stay up lagi ..hhuhuhu..nak wat macam mane..tak boleh nak mengeluh ye, sebab ni adalah lumrah hidup student..ecece..so, malam ni aku nak share perkara ilmiah dengan korang..sebab ape?  ni tajuk aku kene present ok..so, sambil aku siapkan report tak salah rasanya aku nak share ilmu ni dengan rakan2..heee..tapi aku mintak maaf yew..ni mostly untuk student mathematics or statistics...tapi tak salah kan nak belajar sebab ni penting bagi menjalankan research! so..jom buke buku sambil pakai spek mata (budak skema!)..kikiki..what is a specification error?

DEFINITION

A model is considered to be correctly specified if it includes the correct set of independents variable. Specification error means the correct set of independence variable do not include in the model but the model contains incorrect set of independence variables means that invalid variables .The invalid variables means variable does not help explain the underlying phenomena. Specification error also can define a mistake when deciding the correct model to use for predicting the outcome variable.


EXAMPLE SPECIFICATION ERROR 

1)      Relevant variables are omitted from model

The omitted variables occur when unnecessary independent variables is omit  from the equation model.Usually the researcher forget to include all the relevant independent variables when he first specify an equation. So, your important variable may be left out of equation model.This also can known as underfitting problem.

2)      Irrelevant variables are include in the model

The included variable occurs when the unnecesarry independent variables is include in the equation model.Variables that we add nothing to the ability to predict the outcome of the variable.So, your addition of a variables to an equation where it does not belong to equation and can cause a bias.This also can known as overfitting problem.

3)      The wrong functional form

The model does not account for some important non-linearity also one of the cause specification error.We use the wrong functional form of model to generate the outcome of output.

4)      Measurement error

This problem occurs when the magnitude of the variable of interest cannot be accurately measured. We simply do not have accurate measuring instrument to get a accurate measure. This also can occurs from the respondent when they sometimes misunderstand the questionnaires that researcher ask. Measurement error can occurs by a recoding or coding error include the wrong decimal place.

HOW TO DETECT THE PROBLEM 

To detect the specification errors like omitted, adding variables and incorrect functional form, we must do a test.Mostly researcher recommend use the R2 Adjusted to compare model.We have many test to detect the specification error such as examination error (Durbin Watson), Ramsey RESET test, Likelihood Ratio Test, Wald Test, Lagrange Multiplier Test , Hausman Test.Usually, the reseacher prefer use examination error (Durbin Watson), Ramsey RESET test and also R2 adjusted.

   REMEDIAL TO OVERCOME SPECIFICATION ERROR 

The specification can cause a bias and inconsistent outcome of the variables, so, we must take remedial to overcome this problem. For  relevant variables are omitted from model and Irrelevant variables are include in the model, we can use the stepwise regression, forward and backward regression.This procedures can avoid the omitted relevant variables from model and included irrelevant variables in the model.

For the wrong functional form, we must transform the data into logarithms. But we also can proceed with that model as long as the model is still linear in parameter and OLS is still appropriate. Measurement error can overcome by using a good measuring instrument. Researcher also prefer used test-retest reliability which is a quantitative approach to determining whether an instrument measure the same thing at time one that it measure at time two. But when the measurement error is small and do not effect the significant of the outcome variables, we do nothing.

yes..i'm a statistics student!

ni just point je aku share..kan share banyak2..ada je kan yang tutup blog ni..hehehe..maklum la ilmiah memang tak banyak kan orang minat ..lain la citer2 melayu..huhuhu..tapi aku tak minat pon citer melayu ..jalan citer yang sama je..so, dari buang mase tengok citer tu, baik baca post aku nie..dapat jugak ilmu kan? =) walaupon sikit..

mutiara kata dari puan google..hiihi

p/s: aku ingat kata2 Dr Zainuddin Awang "harta kalau di bagi dan dikeluarkan..akan abis jugak lama2 nanti..tapi ilmu kalau dibagi dan dikeluarkan..tetap akan bertambah! "