More From This User
Related Docuements
Regression Algorithm
this is a presentation by cornel university on regression algorithm.I feel th...
From smarttag99
Logit Regression Techniques in the Study of Religion
Robert Wuthnow; Larry Blackwood
Review of Religious Research
, Vol. 19, No. 1. (Autumn, 1977), pp. 83-90.
Stable URL:http://links.jstor.org/sici?sici=0034-673X%28197723%2919%3A1%3C83%3ALRTITS%3E2.0.CO%3B2-Z
Review of Religious Research
is currently published by Religious Research Association, Inc..Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available athttp://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtainedprior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content inthe JSTOR archive only for your personal, non-commercial use.Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained athttp://www.jstor.org/journals/rra.html.Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printedpage of such transmission.The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community takeadvantage of advances in technology. For more information regarding JSTOR, please contact support@jstor.org.http://www.jstor.orgThu Jul 5 00:02:14 2007
RESEARCH
AND
COMMENTARY
8 3
(1967) and Benson (1960) present similar stages of Hadden, Jeffrey
K.
religious institutional development; Herberg (1960), 1970 The Gathering Storm in the Churches. Gar-Hudson (1965), Clcbsch (1968) and Taylor and Jones den City, N.Y.: Doubleday. (1964) describe the specific case of American Prot-Herberg, Will estantism.1960 Protestant-Catholic-Jew. New York: Anchor. Hudson. Winthro~S.1965 'Religion 'in America. New York:: Scribner's.Aaker, David A. and John G. MyersKlausner, Samuel
%.
1975 Advertising Management. Englewood Cliffs, 1964 "Methods of data collection in studies ofN.J.: Prenctice-Hall. religion." Journal for the Scientific Study ofBenson, Purnell Hardy Religion 3:193-203.1960 Relieion in Contemoorarv Culture. New York:
.
Knudten. Richard D.Harper. 1967The Sociology of Religion, New York: Mere-Berger, Peter L.dith.1967 The Sacred Canopy. Garden City, N.Y.: Leymore, Varda LangholzDoubleday. 1975 Hidden Myth, Structure and Symbolism inClebsch, William A.Advertising. London: Heineman.1968 From Sacred to Profane America. NewMead, George
H.
York: Harper.1934 Mind, Self, and Society. Chicago: UniversityColfax, J. David and Susan Frankel Sternbergof Chicago Press.1972 "The Perceotion of racial stereotypes: BlacksOJDea. Thomas F.in mass circulation magazine adve~tisernents."1966' The Sociology of Religion. Englewood Cliffs,Pnblic Opinion Quarterly 36:8-18.N. J.: Prentice-Hall.Cooley, Charles IIortonSandace. C. H. and Frvburcrer. Vernon1922 Human Nature and the Social Order. New1967 '~dvertisi~g6eor; and Practice. Homewood,York: Schocken.Ill.: Invln.Cox, HarveySchneider, Louis1966 The Secular City. New York: Macmillan.1970 Sociological Approach to Religion. New York:Eister, Allen
W.
Wiley.1965 "Empirical rerearch on religion and society:Schneider, Louis and Sanford M. Dornbuschr~ef survey of some fruitful lines of in-1957 "Inspirational religious literature: From latenttuirby:' Review of Religious Research 6:to manifest functions of religion." American125-130.Journal of Sociology 62: 476-481.Elkin, Frederick Scott, Walter Dill1969 'LAdvertisiug themes and quiet revolutions:1921The Psychology of Advertising in Theory andDilemmas in French Canada." AmericanPractice. New York: Boston, Small and May-Journal of Sociology 75:112-122.nard.Shiner, LarryEmbree, Robert A.1967 "The concept of secularization in empirical1973 "The religious association scale: A preliminaryresearch." Tournal for the Scientific Study ofvalidation study." Journal for the ScientificStudy of Religion 12:223-226.Religion 6: -207-220.Taylor, Lee and Arthur R. JonesFlint, John T.1964 Rural Life and Urbanized Societv. New York:1968 "Historical role analysis in the study of secu-Oxford University Press.larization: The laity/clergy ratio in Norway1962 "Measuring religion: Two methods com-1800-1950." Journal for the ScientiJic Studypared." Review of Religious Research
3:
of Religion 17:272-279.159-165.Glaser, Baarney G. and Anselm
L.
StraussWeiss, E. B., F. C. Kendall and Carroll B. Larrabee1967The Discovery of Grounded Theosy. Chicago:1938 The Handbook of Advertising. New York:Aldine. McCraw-Hill.
LOGIT REGRESSION TECHNIQUES IN THE STUDY OF RELIGION
Robert Wuthnow
Department of SociologyPrinceton University
andLarry Blackwood
Department of SociologyUniversity of Arizona
Abstract
The techrzique of logit regression has only recently been introduced
to
the social.sciences from biology and has yet failed to receive much empirical use.
It
is a relativelysimple technique that is especially appropriate to the kinds of data and the problemstypically addressed in the study of religion. The present paper describes and illustrateshow to use the technique and discusses some of its potential applicafions.
All
too often, new statistical procedures of us who may suspect them of being usefulseem mysterious and forbidding. Even those often throw up our hands in despair and fall
REVIEW
OF
RELIGIOUS
RESEARCH
back on the familiar methods we learned ingraduate school. Unless we are fo~tunateenough to know a patient statistician withidle time on his hands, the eEort it takes towade through the statistical literature our-selves-together with the risk that even thenwe might not be told how to perform thenecessary computations-makes it simply tootime-consuming for most of us to take ad-vantage of the latest statistical developments.Logit regression is one such method.Though it has been used extensively in biol-ogy and economics, there ha9 been only onediscussion of it in sociological journals (Theil,
1970)
and scarcely anyone has tried to ac-tually use it. This is unfortunate because legitregression is well-suited to the Binds of datawith which we typically work, especially thoseof us who study religion, and is capable ofsolving some of the analytic problems wetypically confront. The purpose of this paperis to give a brief, non-technical "cook-book"rendition of this method, including an illus-tration and a sample con~puter rogram write-
up;
in other words, to provide a bridge be-tween the statistician and the practitioner.To begin, imagine a typical three-variabletable showing, let us say, the relations amongeducation, occupation (blue collar versuswhite coliar), and belief in God. Suppose wewanted to know whether education is sig-nificantly related to belief in God, controllingfor occupation. What statistics might we use'?ProbabIy we would do a chi-square test tosee if the relation was significant and, sincethe data probably doesn't need the stringentassumptions for regression or con-elations,we would probably use gamma or some com-parable measure of association to see howstrong the relation is. But we would have todo this for the relation between educationand belief in God within each occupationalcategory separately; thus, we would have nosingle measure of the significance or thestrength of the relation between educationand belief in God controlling for occupa-tion. Furthermore, what if we suspected aninteraction effect between education and oc-cupation? We would have no way to test forthe significance of such an effect. Finally,suppose we also wanted to know whether thenet relation between belief in God and edu-cation waslinear, curvilinear, or some othernoniinear form. How could we tell in anyprecise way?Logit regression would answer all thesequestions. Wherever tile effects of one ormore nenconlinuous (calegorized) indepen-dent variables on a
dichoton~ized
dependentvariable sre to be examined,
it
providestests of the significance, strength, direction,and shape of the relations among such vari-ables, including estimates of po3s;ble inter-action effects.
In
this respect, it
is
similar tothe new "log-linear" methods that Leo Good-man (e.g.
1972)
has advanced, but it gives
a
simpler measnre of strength than Goodman'smetbocls and
it
allows the linearity or non-linearity of relations to be cxaniined, whereasGoodman's methods do not. Unlike regularregression and correlation procedures, more-over, this method does not require us to have
normally-distributed
interval or ratio varia-bles.
1SESCRLl'TION
OF
THE TECHNIQUEAssuming that one has a dichotomizeddependent variable and one or more cate-gorized independent variables, the steps forconducting a logit regression analysis are asfollows:
1.
Compute a standard contingency tableshowing the relations between
a
dichotomousdependent variable and one or more discreteindependent variables. The cells of the tableshould contain raw frequencies rather thanpercentages.
2.
Arrange the table such that it has twocolumns, A and
B,
which correspond to thetwo categories of the dependent variable, andn rows, where n equals the number of cate-gories specified by the independent variableor variables (e.g.
4
if there is one indepen-dent variable having
4
categories;
8
if thereare two independent variables, one having
4
categories, the other having
2;
etc.).
3.
Calculate the "logit" value for eachrow of the table. The logit is the natural logof the adds on one category of the dichoto-mized dependent variable relative to theother. Calculate as follows: let
Ai
be thenumber in one column of the table and
Bi
its corresponding number in the other col-umn. Then the formula for the values of
of 00
Wuthnow and Blackwood - Logit Regression Techniques in the Study of Religion
444 Reads
Info and Rating
| Category: | Uncategorized. |
| Rating: | |
| Upload Date: | 09/30/2009 |
| Copyright: | Attribution Non-commercial |
| Tags: |
Leave a Comment