DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu



ŠUMARSKI LIST 3-4/2010 str. 48     <-- 48 -->        PDF

Š. Pezdevšek Malovrh, L. Zadnik Stirn, J. Krč: INFLUENCE OF OWNERSHIPAND PROPERTYSTRUCTURE ... Šumarski list br. 3–4, CXXXIV (2010), 139-149


size and fragmentation of the forest property, form of
ownership and number of co-owners.These variables can


2


either be continuous, discrete or attributive.The . test
was used in order to find out whether there is a relationship
between pairs of variables. Statistically significant relationship
between two variables was defined where p


2


value was less than 0.05.The . tests were performed by
using theSPSS for Windows 16.0computer package.


Willingness of private forest owners for cooperation
regarding the age of forest owners and the ownership
and property structure was performed by the nominal logistic
regression method (Backhaus,2004,Grimm
et. al, 2002,Košmelj,2001a,Košmelj,2001b, Ho-
s mer, Lemeshow, 2000, Albright et. al., 2000,
Norman,2000), by means of theBackward stepwise
algorithm (Field,2009) of theSPSS for Windows 16.0
software package.


The nominal logistic regression is a generalised version
of logistic regression. Logistic regression belongs
to the generalised linear models, which are used for the
prediction of binary dependant variables (“yes”/“no”).
In our case, however, the dependant variable has three
values – associated, willing to cooperate and unwilling
to cooperate.This kind of statistical model is referred
to as nominal logistic regression model. (Košmelj,
Vadnal,2003).The last category of the independent
variable was chosen to represent the reference cate


gory.The estimation of the parameters in logistic re


gression is based on the method of maximum likelihood.
TheWald testwas used to establish the statistical
significance of the correlation between the dependant
variable and the independent variables.The uncharacteristicWald
test enables the exclusion of insignificant
variables from the model, thus ridding the model of unnecessary,
disturbing variables.


The logistic regression method was chosen due to the
fact that it involves fewer statistical requirements than
alternative methods, such as discriminant analysis, pro-
bit analysis, etc. Unlike the aforementioned ana lyses,
the nominal logistic regression is not based on the assumption
of linear correlation between the independent
and the dependant variable nor the assumption of homoscedasticity.
However, logistic regression has an inconvenience,
namely, the multicollinearity (Poje, 2003).
Multicollinearity refers to a situation in which independent
variables in a regression model are a linear combination
of other independent variables. In logistic
regression, to avoid multicollinearity, none of the independent
variables of a multiple model may therefore represent
a linear combination of other independent
variables (Jesenko, 2007).The independent variables
which are a linear combination of other independent variables
are thus to be excluded from the model.


RESULTS AND DISCUSSION – Rezultati i rasprava


Basic information about sample- private forest owner


Osnovni podaci o anketiranima šumoposjednicima


The sample represents 322 forest owners, 75.8 % of (68.9 %) live in rural area, a hamlet or a small village of
whom are male and 24.2 % are female.The average age up to 500 inhabitants with neither shop nor post office,
of the respondents is 54 years and the average level of who generally belong to a village local community
education is completed primary education(31.7 %) or (82.9 %). More than half of the respondents (58.9 %)
vo cational school (32.6 %). The majority of respondents consider their economic status to be average.


General data relating to property conditions – Opći podaci o strukturi šumoposjeda


The mean area of forest ho lding of the respondents
is 16.7 ha.The size of the smallest forest property is 0.1
ha, while the biggest forest property extends up to 150
ha.The respondents were classified depending on the
size of their forest property into categories referred to
as stratums.The stratums and the shares of the respondents
per stratum are given inTable 1.


Table 1 shows that approximately the same number
of forest owners from all stratumsparticipated inthe sur-


Table 1.Size of forest property (stratum)


Tablica 1.Veličina posjeda (grupe)


Stratum (ha)– Grupe (ha)
Up to 0.99
Do 0,99
1 to 4.99
1 do 4,99
5 to 9.99
5 do 9,99
10 to 29.99
10 do 29,99
Over 30
Više od 30
n1990649158
%5.928.019.928.318.0


vey. The exception is the first stratum (up to 0.99 ha)
with the share of only 5.9 %. Such small number of the
respondents from the first stratum is due to several causes:
some owners refused to take part in the survey or
were not even aware that they own a forest, the address
of some owners could not be found or it was impossible
for us to contact them or they are deceased.The refusal
of the owners of forest properties smaller than 1 ha to
participate in the survey shows their inactiveness, lack of


interest or negative attitude towards
foresters and even the forest itself.


The respondents, generally know
how many separate plots of forest
they possess. Only 2.5 % of the respondents
could not answer this
question and were therefore excluded
from the analyses relating to the