Main Page: Difference between revisions

From formulasearchengine
Jump to navigation Jump to search
mNo edit summary
No edit summary
Line 1: Line 1:
In [[statistics]], the '''Jarque–Bera test''' is a [[goodness-of-fit]] test of whether sample data have the [[skewness]] and [[kurtosis]] matching a [[normal distribution]]. The test is named after [[Carlos Jarque]] and [[Anil K. Bera]]. The [[test statistic]] ''JB'' is defined as
'''S''' is an [[axiomatic set theory]] set out by [[George Boolos]] in his article, Boolos (1989). '''S''', a [[first order logic|first-order]] theory, is two-sorted because its [[ontology]] includes “stages” as well as [[set (mathematics)|sets]]. Boolos designed '''S''' to embody his understanding of the “iterative conception of set“ and the associated [[iterative hierarchy]]. '''S''' has the important property that all axioms of [[Zermelo set theory]] ''Z'', except the [[axiom of Extensionality]] and the [[axiom of Choice]], are theorems of '''S''' or a slight modification thereof.


: <math>
==Ontology==
    \mathit{JB} = \frac{n}{6} \left( S^2 + \frac14 (K-3)^2 \right)
Any grouping together of [[mathematical object|mathematical]], [[abstract object|abstract]], or concrete objects, however formed, is a ''collection'', a synonym for what other [[set theory|set theories]] refer to as a [[Class (set theory)|class]]. The things that make up a collection are called [[element (mathematics)|element]]s or members. A common instance of a collection is the [[domain of discourse]] of a [[first order theory]].
  </math>


where ''n'' is the number of observations (or degrees of freedom in general); ''S'' is the sample [[skewness]], and ''K'' is the sample [[kurtosis]]:
All sets are collections, but there are collections that are not sets. A synonym for collections that are not sets is [[proper class]]. An essential task of [[axiomatic set theory]] is to distinguish sets from proper classes, if only because mathematics is grounded in sets, with proper classes relegated to a purely descriptive role.


: <math>
The [[Von Neumann universe]] implements the “iterative conception of set” by stratifying the universe of sets into a series of “stages,” with the sets at a given stage being possible members of the sets formed at all higher stages. The notion of stage goes as follows. Each stage is assigned an [[ordinal number]]. The lowest stage, stage 0, consists of all entities having no members. We assume that the only entity at stage 0 is the [[empty set]], although this stage would include any [[urelements]] we would choose to admit. Stage ''n'', ''n''>0, consists of all possible sets formed from elements to be found in any stage whose number is less than ''n''. Every set formed at stage ''n'' can also be formed at every stage greater than ''n''.<ref>Boolos (1998:88).</ref>
    S = \frac{ \hat{\mu}_3 }{ \hat{\sigma}^3 }
        = \frac{\frac1n \sum_{i=1}^n (x_i-\bar{x})^3} {\left(\frac1n \sum_{i=1}^n (x_i-\bar{x})^2 \right)^{3/2}} ,
</math>
: <math>
K = \frac{ \hat{\mu}_4 }{ \hat{\sigma}^4 } 
        = \frac{\frac1n \sum_{i=1}^n (x_i-\bar{x})^4} {\left(\frac1n \sum_{i=1}^n (x_i-\bar{x})^2 \right)^2} ,
</math>


where <math>\hat{\mu}_3</math> and <math>\hat{\mu}_4</math> are the estimates of third and fourth [[central moment]]s, respectively, <math>\bar{x}</math> is the sample [[mean]], and
Hence the stages form a nested and [[well-ordered]] sequence, and would form a [[hierarchy (mathematics)|hierarchy]] if set membership were [[transitive relation|transitive]]. The iterative conception has gradually become more accepted, despite an imperfect understanding of its historical origins.
<math>\hat{\sigma}^2</math> is the estimate of the second central moment, the [[variance]].


If the data comes from a normal distribution, the ''JB'' statistic [[asymptotic analysis|asymptotically]] has a [[chi-squared distribution]] with two [[degrees of freedom (statistics)|degrees of freedom]], so the statistic can be used to [[statistical hypothesis testing|test]] the hypothesis that the data are from a [[normal distribution]]. The [[null hypothesis]] is a joint hypothesis of the skewness being zero and the [[excess kurtosis]] being zero. Samples from a normal distribution have an expected skewness of 0 and an expected excess kurtosis of 0 (which is the same as a kurtosis of 3). As the definition of ''JB'' shows, any deviation from this increases the JB statistic.
The iterative conception of set steers clear, in a well-motivated way, of the well-known [[paradox]]es of [[Russell's paradox|Russell]], [[Burali-Forti paradox|Burali-Forti]], and [[Cantor's paradox|Cantor]]. These paradoxes all result from the unrestricted use of the [[axiom of comprehension|principle of comprehension]] of [[naive set theory]]. Collections such as “the class of all sets” or “the class of all [[ordinal number|ordinals]]” include sets from all stages of the iterative hierarchy. Hence such collections cannot be formed at any given stage, and thus cannot be sets.


For small samples the chi-squared approximation is overly sensitive, often rejecting the null hypothesis when it is in fact true. Furthermore, the distribution of p-values departs from a uniform distribution and becomes a right-skewed uni-modal distribution, especially for small p-values. This leads to a large [[Type I error]] rate. The table below shows some p-values approximated by a chi-squared distribution that differ from their true alpha levels for small samples.
==Primitive notions==
This section follows Boolos (1998: 91). The variables ''x'' and ''y'' range over sets, while ''r'', ''s'', and ''t'' range over stages. There are three [[primitive notion|primitive]] two-place [[Predicate (mathematical logic)|predicates]]:
* Set-set: ''x''∈''y'' denotes, as usual, that set ''x'' is a member of set ''y'';
* Set-stage: ''Fxr'' denotes that set ''x'' “is formed at” stage ''r'';
* Stage-stage: ''r''<''s'' denotes that stage ''r'' “is earlier than” stage ''s''.


:{| class="wikitable"
The axioms below include a defined two-place set-stage predicate, ''Bxr'', which abbreviates:
|+Calculated p-value equivalents to true alpha levels at given sample sizes
:<math>\exist s[s<r \land Fxs].</math>
! True α level !! 20 !! 30 !! 50 !! 70 !! 100
''Bxr'' is read as “set ''x'' is formed before stage ''r''.
|-
! 0.1
| 0.307 || 0.252 || 0.201 || 0.183 || 0.1560
|-
! 0.05
| 0.1461 || 0.109 || 0.079 || 0.067 || 0.062
|-
! 0.025
| 0.051 || 0.0303 || 0.020 || 0.016 || 0.0168
|-
! 0.01
| 0.0064 || 0.0033 || 0.0015 || 0.0012 || 0.002
|}
(These values have been approximated by using [[Monte Carlo simulation]] in [[Matlab]])


In [[MATLAB]]'s implementation, the chi-squared approximation for the JB statistic's distribution is only used for large sample sizes (>&nbsp;2000). For smaller samples, it uses a table derived from [[Monte Carlo simulations]] in order to interpolate p-values.<ref name="MathWorks">{{cite web|url=http://www.mathworks.com/access/helpdesk/help/toolbox/stats/jbtest.html|title=Analysis of the JB-Test in MATLAB|publisher=MathWorks|accessdate=May 24, 2009}}</ref>
[[identity (mathematics)|Identity]], denoted by infix ‘=’, does not play the role in '''S''' it plays in other set theories, and Boolos does not make fully explicit whether the background [[first order logic|logic]] includes identity. '''S''' has no [[axiom of Extensionality]] and identity is absent from the other '''S''' axioms. Identity does appear in the axiom schema distinguishing '''S+''' from '''S''',<ref>Boolos (1998: 97).</ref> and in the derivation in '''S''' of the [[axiom of pairing|Pairing]], [[axiom of the empty set|Null set]], and [[axiom of infinity|Infinity]] axioms of ''[[Z]]''.<ref>Boolos (1998: 103–04).</ref>


==History==
==Axioms==
Considering normal sampling, and √''β''<sub>1</sub> and ''β''<sub>2</sub> contours, {{harvtxt|Bowman|Shenton|1975}} noticed that the statistic ''JB'' will be asymptotically ''χ''<sup>2</sup>(2)-distributed; however they also noted that “large sample sizes would doubtless be required for the ''χ''<sup>2</sup> approximation to hold”. Bowman and Shelton did not study the properties any further, preferring [[D’Agostino’s K-squared test]].
The symbolic axioms shown below are from Boolos (1998: 91), and govern how sets and stages behave and interact. The natural language versions of the axioms are intended to aid the intuition.
<!-- The symbolic axioms should not be changed without discussion on the talk page. Altering these axioms would constitute original research. However the English descriptions may be changed.-->


Around 1979, Anil Bera and [[Carlos Jarque]] while working on their dissertations on regression analysis, have applied the [[Lagrange multiplier principle]] to the [[Pearson distribution|Pearson family of distributions]] to test the normality of unobserved regression residuals and found that the ''JB'' test was asymptotically optimal (although the sample size needed to “reach” the asymptotic level was quite large). In 1980 the authors published a paper ({{harvnb|Jarque|Bera|1980}}), which treated a more advanced case of simultaneously testing the normality, [[homoscedasticity]] and absence of [[autocorrelation]] in the residuals from the [[linear regression model]]. The ''JB'' test was mentioned there as a simpler case. A complete paper about the JB Test was published in the ''International Statistical Review'' in 1987 dealing with both testing the normality of observations and the normality of unobserved regression residuals, and giving finite sample significance points.
The axioms come in two groups of three. The first group consists of axioms pertaining solely to stages and the stage-stage relation ‘<’.


==Jarque–Bera test in regression analysis==
'''Tra''': <math> \forall r \forall s \forall t[r<s \land s<t \rightarrow r<t] \,.</math>
According to Robert Hall, David Lilien, et al. (1995) when using this test along with multiple regression analysis the right estimate is:


: <math>
“Earlier than” is transitive.
    \mathit{JB} = \frac{n-k}{6} \left( S^2 + \frac14 (K-3)^2 \right)
  </math>


where ''n'' is the number of observations and ''k'' is the number of regressors when examining residuals to an equation.
'''Net''': <math> \forall s \forall t \exist r[t<r \land s<r] \,.</math>
 
A consequence of ''Net'' is that every stage is earlier than some stage.
 
'''Inf''': <math> \exist r \exist u [u<r \land \forall t[t<r \rightarrow \exist s[t<s \land s<r]]] \,.</math>
 
The sole purpose of ''Inf'' is to enable deriving in '''S''' the [[axiom of infinity]] of other set theories.
 
The second and final group of axioms involve both sets and stages, and the predicates other than '<':
 
'''All''': <math>  \forall x \exist r Fxr \,.</math>
 
Every set is formed at some stage in the hierarchy.
 
'''When''': <math> \forall r \forall x [Fxr \leftrightarrow [\forall y (y \in x \rightarrow Byr) \and \lnot Bxr] ] \,.</math>
 
A set is formed at some stage [[iff]] its members are formed at earlier stages.
 
Let ''A''(''y'') be a formula of '''S''' where ''y'' is free but ''x'' is not. Then the following axiom schema holds:
 
'''Spec''': <math> \exist r \forall y[A(y) \rightarrow Byr] \rightarrow \exist x \forall y[y \in x \leftrightarrow A(y)] \,.</math>
 
If there exists a stage ''r'' such that all sets satisfying ''A''(''y'') are formed at a stage earlier than ''r'', then there exists a set ''x'' whose members are just those sets satisfying ''A''(''y''). The role of ''Spec'' in '''S''' is analogous to that of the [[axiom schema of specification]] of [[Z]].
 
==Discussion==
Boolos’s name for [[Zermelo set theory]] minus extensionality was ''Z-''. Boolos derived in '''S''' all axioms of ''Z-'' except the [[axiom of choice]].<ref>Boolos (1998: 95–96; 103–04).</ref> The purpose of this exercise was to show how most of conventional set theory can be derived from the iterative conception of set, assumed embodied in '''S'''. [[Extensionality]] does not follow from the iterative conception, and so is not a theorem of '''S'''. However, '''S''' + Extensionality is free of contradiction if '''S''' is free of contradiction.
 
Boolos then altered ''Spec'' to obtain a variant of '''S''' he called '''S+''', such that the [[axiom schema of replacement]] is derivable in '''S+'''&nbsp;+&nbsp;Extensionality. Hence '''S+''' + Extensionality has the power of [[Zermelo–Fraenkel set theory|ZF]]. Boolos also argued that the [[axiom of choice]] does not follow from the iterative conception, but did not address whether Choice could be added to '''S''' in some way.<ref>Boolos (1998: 97).</ref> Hence '''S+''' + Extensionality cannot prove those theorems of the industry-standard set theory [[ZFC]] whose proofs require Choice.
 
'''Inf''' guarantees the existence of stages ω, and of ω&nbsp;+&nbsp;''n'' for finite ''n'', but not of stage ω&nbsp;+&nbsp;ω. Nevertheless, '''S''' yields enough of [[transfinite numbers|Cantor's paradise]] to ground almost all of contemporary mathematics.<ref>”…the overwhelming majority of 20th century mathematics is straightforwardly representable by sets of fairly low infinite ranks, certainly less than ω&nbsp;+&nbsp;20.” (Potter 2004: 220). The exceptions to Potter's statement presumably include [[category theory]], which requires the weakly [[inaccessible cardinal]]s afforded by [[Tarski–Grothendieck set theory]], and the higher reaches of set theory itself.</ref>
 
Boolos compares '''S''' at some length to a variant of the system of [[Frege]]’s ''Grundgesetze'', in which [[Hume's principle]], taken as an axiom, replaces Frege’s Basic Law V, an [[Comprehension_axiom#Unrestricted_comprehension|unrestricted comprehension axiom]] which made Frege's system inconsistent; see [[Russell's paradox]].


==References==
==References==
{{reflist}}
* [[George Boolos]] (1989) “Iteration Again,” ''Philosophical Topics'' 17: 5–21. Reprinted in his (1998) ''Logic, Logic, and Logic''. Harvard Univ. Press: 88–104.
<HR>
* [[Michael Potter]] (2004) ''Set Theory and Its Philosophy''. Oxford Univ. Press.
{{refbegin}}
* {{cite journal
  | first1 = K.O. | last1 = Bowman
  | first2 = L.R. | last2 = Shenton
  | title = Omnibus contours for departures from normality based on √''b''<sub>1</sub> and ''b''<sub>2</sub>
  | year = 1975
  | journal = Biometrika
  | volume = 62 | issue = 2
  | pages = 243–250
  | jstor = 2335355
  | ref = CITEREFBowmanShenton1975
  }}
* {{cite journal
  | first1 = Carlos M. | last1 = Jarque | authorlink1 = Carlos Jarque
  | first2 = Anil K. | last2 = Bera
  | title = Efficient tests for normality, homoscedasticity and serial independence of regression residuals
  | year = 1980
  | journal = Economics Letters
  | volume = 6 | issue = 3
  | pages = 255–259
  | doi = 10.1016/0165-1765(80)90024-5
  | ref = CITEREFJarqueBera1980
  }}
* {{cite journal
  | first1 = Carlos M. | last1 = Jarque | authorlink1 = Carlos Jarque
  | first2 = Anil K. | last2 = Bera
  | title = Efficient tests for normality, homoscedasticity and serial independence of regression residuals: Monte Carlo evidence
  | year = 1981
  | journal = Economics Letters
  | volume = 7 | issue = 4
  | pages = 313–318
  | doi = 10.1016/0165-1765(81)90035-5
  | ref = CITEREFJarqueBera1981
  }}
* {{cite journal
  | first1 = Carlos M. | last1 = Jarque | authorlink1 = Carlos Jarque
  | first2 = Anil K. | last2 = Bera
  | title = A test for normality of observations and regression residuals
  | year = 1987
  | journal = International Statistical Review
  | volume = 55 | issue = 2
  | pages = 163–172
  | jstor = 1403192
  | ref = CITEREFJarqueBera1987
  }}
* {{cite book
  | first = | last = Judge
  | coauthors = et al.
  | title = Introduction and the theory and practice of econometrics
  | year = 1988
  | edition = 3rd
  | pages = 890–892
  }}
* {{cite book
  | first1 = Robert E. | last1 = Hall
  | first2 = David M. | last2 = Lilien
  | coauthors = et al.
  | title = EViews User Guide
  | year = 1995
  | pages = 141
  }}
{{refend}}
 
== Implementations ==
* [http://www.alglib.net/statistics/hypothesistesting/jarqueberatest.php ALGLIB] includes implementation of the Jarque–Bera test in C++, C#, Delphi, Visual Basic, etc.
* [[gretl]] includes an implementation of the Jarque–Bera test
* [[R (programming language)|R]] includes implementations of the Jarque–Bera test: ''jarque.bera.test'' in package ''tseries'', for example, and ''jarque.test'' in package ''moments''.
* [[Matlab|MATLAB]] includes implementation of the Jarque–Bera test, the function "jbtest".
* [[Python (programming language)|Python]] [[statsmodels]] includes implementation of the Jarque–Bera test, "statsmodels.stats.stattools.py".


{{Statistics}}
==Footnotes==
{{Reflist}}


{{DEFAULTSORT:Jarque-Bera test}}
[[Category:Set theory]]
[[Category:Normality tests]]
[[Category:Systems of set theory]]
[[Category:Z notation]]

Revision as of 11:37, 18 August 2014

S is an axiomatic set theory set out by George Boolos in his article, Boolos (1989). S, a first-order theory, is two-sorted because its ontology includes “stages” as well as sets. Boolos designed S to embody his understanding of the “iterative conception of set“ and the associated iterative hierarchy. S has the important property that all axioms of Zermelo set theory Z, except the axiom of Extensionality and the axiom of Choice, are theorems of S or a slight modification thereof.

Ontology

Any grouping together of mathematical, abstract, or concrete objects, however formed, is a collection, a synonym for what other set theories refer to as a class. The things that make up a collection are called elements or members. A common instance of a collection is the domain of discourse of a first order theory.

All sets are collections, but there are collections that are not sets. A synonym for collections that are not sets is proper class. An essential task of axiomatic set theory is to distinguish sets from proper classes, if only because mathematics is grounded in sets, with proper classes relegated to a purely descriptive role.

The Von Neumann universe implements the “iterative conception of set” by stratifying the universe of sets into a series of “stages,” with the sets at a given stage being possible members of the sets formed at all higher stages. The notion of stage goes as follows. Each stage is assigned an ordinal number. The lowest stage, stage 0, consists of all entities having no members. We assume that the only entity at stage 0 is the empty set, although this stage would include any urelements we would choose to admit. Stage n, n>0, consists of all possible sets formed from elements to be found in any stage whose number is less than n. Every set formed at stage n can also be formed at every stage greater than n.[1]

Hence the stages form a nested and well-ordered sequence, and would form a hierarchy if set membership were transitive. The iterative conception has gradually become more accepted, despite an imperfect understanding of its historical origins.

The iterative conception of set steers clear, in a well-motivated way, of the well-known paradoxes of Russell, Burali-Forti, and Cantor. These paradoxes all result from the unrestricted use of the principle of comprehension of naive set theory. Collections such as “the class of all sets” or “the class of all ordinals” include sets from all stages of the iterative hierarchy. Hence such collections cannot be formed at any given stage, and thus cannot be sets.

Primitive notions

This section follows Boolos (1998: 91). The variables x and y range over sets, while r, s, and t range over stages. There are three primitive two-place predicates:

  • Set-set: xy denotes, as usual, that set x is a member of set y;
  • Set-stage: Fxr denotes that set x “is formed at” stage r;
  • Stage-stage: r<s denotes that stage r “is earlier than” stage s.

The axioms below include a defined two-place set-stage predicate, Bxr, which abbreviates:

s[s<rFxs].

Bxr is read as “set x is formed before stage r.”

Identity, denoted by infix ‘=’, does not play the role in S it plays in other set theories, and Boolos does not make fully explicit whether the background logic includes identity. S has no axiom of Extensionality and identity is absent from the other S axioms. Identity does appear in the axiom schema distinguishing S+ from S,[2] and in the derivation in S of the Pairing, Null set, and Infinity axioms of Z.[3]

Axioms

The symbolic axioms shown below are from Boolos (1998: 91), and govern how sets and stages behave and interact. The natural language versions of the axioms are intended to aid the intuition.

The axioms come in two groups of three. The first group consists of axioms pertaining solely to stages and the stage-stage relation ‘<’.

Tra: rst[r<ss<tr<t].

“Earlier than” is transitive.

Net: str[t<rs<r].

A consequence of Net is that every stage is earlier than some stage.

Inf: ru[u<rt[t<rs[t<ss<r]]].

The sole purpose of Inf is to enable deriving in S the axiom of infinity of other set theories.

The second and final group of axioms involve both sets and stages, and the predicates other than '<':

All: xrFxr.

Every set is formed at some stage in the hierarchy.

When: rx[Fxr[y(yxByr)¬Bxr]].

A set is formed at some stage iff its members are formed at earlier stages.

Let A(y) be a formula of S where y is free but x is not. Then the following axiom schema holds:

Spec: ry[A(y)Byr]xy[yxA(y)].

If there exists a stage r such that all sets satisfying A(y) are formed at a stage earlier than r, then there exists a set x whose members are just those sets satisfying A(y). The role of Spec in S is analogous to that of the axiom schema of specification of Z.

Discussion

Boolos’s name for Zermelo set theory minus extensionality was Z-. Boolos derived in S all axioms of Z- except the axiom of choice.[4] The purpose of this exercise was to show how most of conventional set theory can be derived from the iterative conception of set, assumed embodied in S. Extensionality does not follow from the iterative conception, and so is not a theorem of S. However, S + Extensionality is free of contradiction if S is free of contradiction.

Boolos then altered Spec to obtain a variant of S he called S+, such that the axiom schema of replacement is derivable in S+ + Extensionality. Hence S+ + Extensionality has the power of ZF. Boolos also argued that the axiom of choice does not follow from the iterative conception, but did not address whether Choice could be added to S in some way.[5] Hence S+ + Extensionality cannot prove those theorems of the industry-standard set theory ZFC whose proofs require Choice.

Inf guarantees the existence of stages ω, and of ω + n for finite n, but not of stage ω + ω. Nevertheless, S yields enough of Cantor's paradise to ground almost all of contemporary mathematics.[6]

Boolos compares S at some length to a variant of the system of Frege’s Grundgesetze, in which Hume's principle, taken as an axiom, replaces Frege’s Basic Law V, an unrestricted comprehension axiom which made Frege's system inconsistent; see Russell's paradox.

References

  • George Boolos (1989) “Iteration Again,” Philosophical Topics 17: 5–21. Reprinted in his (1998) Logic, Logic, and Logic. Harvard Univ. Press: 88–104.
  • Michael Potter (2004) Set Theory and Its Philosophy. Oxford Univ. Press.

Footnotes

43 year old Petroleum Engineer Harry from Deep River, usually spends time with hobbies and interests like renting movies, property developers in singapore new condominium and vehicle racing. Constantly enjoys going to destinations like Camino Real de Tierra Adentro.

  1. Boolos (1998:88).
  2. Boolos (1998: 97).
  3. Boolos (1998: 103–04).
  4. Boolos (1998: 95–96; 103–04).
  5. Boolos (1998: 97).
  6. ”…the overwhelming majority of 20th century mathematics is straightforwardly representable by sets of fairly low infinite ranks, certainly less than ω + 20.” (Potter 2004: 220). The exceptions to Potter's statement presumably include category theory, which requires the weakly inaccessible cardinals afforded by Tarski–Grothendieck set theory, and the higher reaches of set theory itself.