(As a background info: this is the methods section of the example I showed in class - working with the research question as to whether membership in church or religious organizations impact activism and membership in other organizations).
This paper focuses onBrazil Brazil Latin America .
The first index is the variable activism and is constructed using four questions that were put forth to measure the degree of social activism of the interviewed individuals. The original set of questions identifies five types of activism – signing a petition, joining in boycotts, attending lawful demonstrations, joining unofficial strikes and occupying buildings and factories – but the last item is not relevant for this study and will be omitted. Interviewees were asked to respond to each one by stating whether they have done it (1), might do it (2) or would never do it (3). The WVS is designed such that the validity of each measurement item is maximized. However, a brief discussion of validity is necessary. Validity refers to whether an instrument measures what it is supposed to measure (Frankfort-Nachmias and Nachmias, 2000).
The independent variables used in
order to explain the variation in activism
and membership and the descriptions
of these variables are given below.
3.4.1.
Membership in Church or a Religious Organization
Table 3. Frequency
Distribution of Membership in Church or Religious Organization
3.5.5. Employment Status
The employment status is coded as follows: 1- employed; 2- retired; 3- housewife; 4- student; 5 unemployed. The intervening effect of the employment status may be of relevance for the purposes of this study. It is possible to conjecture that the effect of church or a religious organization membership on activism or membership is dependent on the employment status of the person. Those who are employed will have more opportunity to join demonstrations, strikes and boycotts and therefore their activism level may be higher in comparison to those who are retired or unemployed.
This paper focuses on
3.1.
Sampling and Data
The WVS was initiated in 1981 to study
the values and attitudes of people belonging to different economic,
educational, and cultural backgrounds. This cross-national survey explores
whether mass belief systems are changing throughout the world. The surveys are
administered using stratified, multistage random sampling and covers adult
citizens aged 18 and older. The samples are selected in two stages, first of
which is a random selection of sampling locations. This is done in order to
ensure all types of locations are represented. Among these sampling locations a
random selection of individuals is drawn. Multi-stage random sampling first
selects the household randomly from these locations, and then chooses the
individuals in the household by using either a grid to select the person to be
interviewed from a listing of all those in the household or in other countries
through quota sampling with quotas assigned on the basis of sex, age and occupation,
using census data as a guide to the distribution of each group in the
population. The interviews are structured and are carried out face-to-face at
home. The advantages of face-to-face interviews are high response rates,
collection of supplementary information and control of the interview situation
(Frankfort-Nachmias and Nachmias, 2000). The interviews were carefully
worded/translated in order to facilitate the correct understanding of
questions. The surveys from most low-income countries undersample the
illiterate portion of the public and oversample the urban areas and the more
educated strata (Inglehart, 1994). Since the present data set is also from a
developing country it is weighted by using the weight variable (V236 in the WVS
Codebook) to correct for these errors.
The WVS data for Brazil have been gathered by the Instituto Gallup de Opiniao Publica in Sao Paulo
3.2.
The Dependent Variable: Activism
As it was stated earlier, this study will
test whether membership in a church or a religious organization affects the level
of mobilization of Brazilians. For the purposes of this paper mobilization
means participation in non-violent acts organized by civil society groups and
is tapped using two indices.[1]
The variables that constitute the indices and how they are constructed will be
outlined below. Their characteristics in terms of frequency distributions and
descriptive statistics will also be provided. Subsequently their validity and
reliability will be examined.
It is necessary to examine different
types of validity that are relevant for the measurement process of this paper.
For a measurement to be content valid, it should cover all attributes of the
concept to be measured. One sub-category of construct validity is face validity
and it refers to the subjective evaluation of the researcher – through
comparing the survey questions with already existing surveys, for example, or
through consulting specialists in the area. The WVS has been administered in different
countries over three waves and there was ample possibility to check the face
validity of the measures. The second category is sampling validity and it
refers to whether the “full domain of content that is relevant to the
particular measurement situation (Carmines and Zeller, 1979)” is specified. It
can be argued that the variables making up our index (activism) could have been more valid had there been a question
about joining organizations of self-help and participation in neighborhood
associations. Some of the variables – such as joining a boycott or signing a
petition – may be more related to the Western European (where the surveys
originated) context rather than the Latin American one. Hence there is the
possibility of a mismatch of repertoires of action and the range of questions
concerning activism for the case of Brazil Brazil
The index adds the values for the answers
of each individual for the four types of activism, i.e. the four items. The new
variable ranges between 4 and 12, where an individual who scores 4 has done all
types of activism at least once whereas someone who scores 12 would never do
any of them. This new scale was tested for its internal consistency or
reliability[2] using
the reliability analysis incorporated into the SPSS. The value of alpha – which
signifies the proportion of variation in the subjects that is explained by the
new scale – generated by the program (0.70) indicates that the index is
reliable.[3]
Subsequently, the index was re-coded in order to obtain a new scale variable
where the 4-7 range becomes 1 and means that the person is active; 8-10 range
becomes 2 and means that the person is potentially active; and 10-12 range
becomes 3 and means that the person is not active. The characteristics of the activism variable are shown in Graph 1.
Graph. 1
Frequency Distribution of Activism
3.3
The Dependent Variable: Membership
The second index, membership, is constructed using the questions that tap the level
of membership in voluntary organizations other than the church. Types of
voluntary organizations used in order to calculate the index are labor unions
(which may not always be voluntary), political parties, environmental
organizations, professional associations, charitable organizations and any
other voluntary organization.[4]
The question that tapped these variables asked the respondents whether they are
active members (1), inactive members (2) or not a member (3) of these
organizations.
The concerns related to validity for the
items of the previous index are not relevant here due to the existence of the
residual item – any other voluntary organization. The index is constructed by
adding the items. The new variable, which ranges between 6 and 18, was re-coded
into a nominal variable where some membership, namely 6-16, was coded as 1 and
no significant membership, namely 17-18, was coded as 2. However, the
reliability analysis of membership
index reports an alpha of 0.45, which indicates that the index is unreliable.
Therefore the results obtained using this variable should be interpreted accordingly.
The frequency distribution of the membership
variable is given in Graph 2.
Graph 2.
Frequency Distribution of Membership
3.4
Independent Variables
The independent variables used in
order to explain the variation in activism
and membership and the descriptions
of these variables are given below.
3.4.1.
Membership in Church or a Religious Organization
This variable is the main independent variable
and is measured using the question that asks the respondents whether they are
active, inactive members of or do not belong to a church or a religious
organization.[5] The
code for active member is 1 whereas that of a non-member is 3. This variable is
moderately correlated with attendance in religious services (see Table 2) and
hence the measure we will be using for this study as an independent variable
has predictive validity. Namely, the church and religious organization
membership is an accurate measurement instrument and taps the relevant
information.
However, the exploration of the characteristics
of this variable demonstrates that only 12.8 % of the sample is actively member
of a church or religious organization (see Table 3). This skewed distribution
will not create a problem in terms of the size of the sub-category of active or
inactive membership during the analysis phase since Nactive=228 and
Ninactive=168 are still large samples (as a rule of thumb N>30 is
considered large).
Table 3. Frequency
Distribution of Membership in Church or Religious Organization
3.5. Control
Variables
General descriptive
characteristics of the control variables – gender, age, education and size of
town – are provided in Table 4.
Table
4. Descriptive Statistics
3.5.1. Gender
The gender of the respondent is coded 1 in
the case of a male interviewee and 2 in the case of a female. The gender is
also a very important aspect for the study of mobilization and activism in Latin America . However, due to the concerns articulated
in the previous sub-section the effect of gender may not be as significant as
those reported by the microanalyses conducted in Latin
America .
3.5.2. Age
The age variable is important to include as a
control variable because there can be significant differences among age
groups in terms of activism and membership in voluntary organizations.
This variable was measured using a question that asks how old the person is.
However, in the Brazilian case it was coded within intervals as follows:
18-21, 22-27, 28-35, 36-45 and 46+. The frequency distribution of this
ordinal variable is given in Graph 3.
3.5.3. Education
Education was tapped using the following
question: What is the highest
educational level that you have attained? For the case of Brazil
Education is an important variable because
research on social movements and mobilization in Europe
tends to find higher levels of education being correlated with higher levels
of participation. However, in the case of Brazil Latin America
and their typical members being women and not very educated. However, since
the type of activism captured by the questions of the WVS is closer to the
European types it is still important to analyze whether education is a factor
which affects the level of activism and membership in voluntary
organizations.
3.5.4. Community Size
Community size, or the size of town, where
the individual lives can affect the level of mobilization in various ways.
The categories used for coding the Brazilian sample and the frequency
distribution are given in Table 5. As it was mentioned before, large cities
are oversampled in the survey but the use of weights in the analysis will
prevent this from biasing the analysis.
Table 5. Coding and Frequency of Size of Town
(N=1782)
3.5.5. Employment Status
The employment status is coded as follows: 1- employed; 2- retired; 3- housewife; 4- student; 5 unemployed. The intervening effect of the employment status may be of relevance for the purposes of this study. It is possible to conjecture that the effect of church or a religious organization membership on activism or membership is dependent on the employment status of the person. Those who are employed will have more opportunity to join demonstrations, strikes and boycotts and therefore their activism level may be higher in comparison to those who are retired or unemployed.
[1] An index is a composite variable constructed using several
variables. The variables that constitute an index are called items
(Frankfort-Nachmias and Nachmias, 2000).
[2] Reliability refers to the extent to which a measuring procedure
yields the same result on repeated trials. In order to measure the internal
consistency or reliability a measure called the Cronbach’s alpha is used
(Carmines and Zeller, 1979).
[3] Frankfort-Nachmias and Nachmias (2000) state that in order to claim
a scale is reliable the alpha value should be greater than or equal to 0.70
whereas the limit set by Cramer (1998) is 0.80.
[4] This variable excludes all sorts of religious or church
organizations.
[5] This variable will sometimes be referred to as church membership
for the sake of simplicity.
