Journal of Quantitative Criminology, Vol. 19, No. 2, June 2003 (ß 2003 )
Measurement and Other Errors in County-Level UCR
Data: A Reply to Lott and Whitley
Michael D. Maltz1,2 and Joseph Targonski1
Lott and Whitley note that our analyses of the errors in the county-level UCR data
used in More Guns, Less Crime (J. R. Lott, University of Chicago Press, Chicago,
1998, 2000) ignore the fact that all data have measurement error, that the largest
errors were in counties with low populations, and that population-weighted
regressions were used. We agree that this mitigates some of the effects of the errors,
but does not take them fully into account. We also note that this is but one of the
problems associated with the analysis. We therefore find no reason to alter our
original conclusion, ‘‘that in their current condition, county-level UCR crime
statistics cannot be used for evaluating the effects of changes in policy.’’
1. INTRODUCTION
In their critique of our paper, which described the effect of missing data
on the analyses in More Guns, Less Crime (Lott, 1998, 2000) that use
county-level UCR data, Lott and Whitley (2003) make a number of state-
ments. They include:
. we ignored the effect of population-weighted regression;
. all data sets have measurement error, and survey data have much more
missing data than are found in Lott’s (1998, 2000) analyses;
. we used the wrong states in our analyses;
. state-level analyses are similarly affected by missing data;
. state- and city-level analyses show the same pattern as the county-level
analysis.
1
Department of Criminal Justice, University of Illinois at Chicago, 1007 W. Harrison Street,
Chicago, Illinois 60607-7140.
2
To whom correspondence should be addressed. E-mail:
199
0748-4518/03/0600-0199/0 ß 2003 Plenum Publishing Corporation
, 200 Maltz and Targonski
Our original paper dealt primarily with the problems inherent in the
UCR, not with the specifics of the analyses in More Guns, Less Crime
(MGLC). This is the primary reason that we did not ‘‘directly test if it affects
the results’’ (Lott and Whitley, 2003).
In this paper, however, we respond to the Lott and Whitley with respect
to each of the above points. In Section 2 we discuss the effect of weighting
the observations by population. In Section 3 we examine the effect of
errors, including missing data. Section 4 discusses the choice of states we
used. In Section 5 we discuss state-level analyses. Section 6 contains our
conclusion.
2. POPULATION-WEIGHTED REGRESSION
Lott and Whitley are, of course, correct in their statement that a
regression weighted by population would diminish the effect of the small
counties that (generally) have less consistent reporting histories. As we
stated in our paper, Lott (2000, pp. 143, 155) criticized Black and Nagin’s
(1998) analysis for ignoring the smaller counties and focusing their analysis
on counties with 100,000 or more population, so we included all counties in
our paper. Their reanalysis of our data shows that when weighted by
population the problem is reduced. For example, we note from their Fig. 3
that, although there are 21 states that have over 10% of their observations
with extensive missing data (defined by 30% or more coverage gaps), only
15 states have over 10% of their population affected by this magnitude of
coverage gaps. This is still a substantial amount of error.
In both editions of MGLC the data were taken at face value, assumed
to be error-free. Now that the nature of the errors is more thoroughly
understood, Lott and Whitley suggest that they do not affect the results
substantially. They reanalyzed the MGLC data removing the 16 most error-
laden states; however they do not state how (or whether) they dealt with the
errors that remained in the data of the remaining 44 states. If they did not
adjust for the errors in those states, this step is inadequate.
3. ERRORS IN THE COUNTY-LEVEL DATA SET
The errors are not insignificant. Moreover, there are a number of dif-
ferent types of error that affect analyses, not just the ones to which Lott and
Whitley refer. They are correct in their assertion that ‘‘[v]irtually all data
have measurement error,’’ in referring to the problems with UCR missing
data that we documented (Maltz and Targonski, 2002). Actually, this is only
one type of error that affects their findings; instrumentation error and error
Voordelen van het kopen van samenvattingen bij Stuvia op een rij:
Verzekerd van kwaliteit door reviews
Stuvia-klanten hebben meer dan 700.000 samenvattingen beoordeeld. Zo weet je zeker dat je de beste documenten koopt!
Snel en makkelijk kopen
Je betaalt supersnel en eenmalig met iDeal, creditcard of Stuvia-tegoed voor de samenvatting. Zonder lidmaatschap.
Focus op de essentie
Samenvattingen worden geschreven voor en door anderen. Daarom zijn de samenvattingen altijd betrouwbaar en actueel. Zo kom je snel tot de kern!
Veelgestelde vragen
Wat krijg ik als ik dit document koop?
Je krijgt een PDF, die direct beschikbaar is na je aankoop. Het gekochte document is altijd, overal en oneindig toegankelijk via je profiel.
Tevredenheidsgarantie: hoe werkt dat?
Onze tevredenheidsgarantie zorgt ervoor dat je altijd een studiedocument vindt dat goed bij je past. Je vult een formulier in en onze klantenservice regelt de rest.
Van wie koop ik deze samenvatting?
Stuvia is een marktplaats, je koop dit document dus niet van ons, maar van verkoper nimrit. Stuvia faciliteert de betaling aan de verkoper.
Zit ik meteen vast aan een abonnement?
Nee, je koopt alleen deze samenvatting voor $8.49. Je zit daarna nergens aan vast.