Before administration of the survey, my colleagues who were the interviewers were trained in conducting surveys and other relevant issues. They were provided with instruction sheets, maps, and a list of targeted streets with the addresses and explanation of relative acronyms and codes. Quality control was enabled through continued
consultation and briefing between interviewers and main researcher by phone during data collection and after a day’s work. The instrument, recording sheets etc. were also
crosschecked by the researcher and clarification sought if needed. Additionally, during administration, one researcher was designated the recorder at each survey site, and recorded on the specified sheet, the premises visited, need for call back, and completed interviews. In Paterson, the Spanish version of the questionnaire was delivered by a
Spanish-speaking researcher to this ethnic group. All respondents were provided with sheets with the Likert attitude scales for their responses as they were read the questions.
To maximize data collection return, householders unavailable on the interview days including the days of ‘call backs’, were sent questionnaires by mail. (The instrument is designed so that it can function for interview purposes and self-administration). A cover letter seeking support and explaining the survey, and with instructions for the questionnaire completion was sent to each “absent” householder in all three
municipalities. This was not successful. Only one questionnaire from Hawthorne and two from Paterson were returned.
. As stated before, permission was sought from the relevant public authorities and business owners to interview eligible persons in identified public places close to the sites. To ensure the respondent was eligible for interview, a map of the delineated
neighborhood was shown to the individual to ascertain place of residence is in the
targeted area. All aforementioned other criteria for eligibility were enforced too. This was a rewarding strategy.
3.6 Quality Control
During data entry, quality control was assured by proof reading the database. Each interview schedule in the database was crosschecked with the hard copy to spot
discrepancies in data entry and coding. Corrections were made as necessary. This activity was done solely by the main researcher therefore avoiding inter-coder mistakes.
Secondly, during exploratory analysis/screening of the data, careful attention was given for mistakes in data coding and entry and rectified as necessary.
3.7 Analysis
SPSS statistical software was used to do both descriptive and inferential analysis of the instrument. Because, some statements (in measuring a variable) were written both in the negative and positive, then reverse coding had to be done before executing the factor analysis process. Seeing that each item statement for the measured scales ‘Access to the
Decision Making Process’ and ‘Public Acceptance’ have to be summed in the process of
obtaining an overall mean score for the individual, it is critical that each item statement is measuring the same latent factor and is highly correlated with other statements of the factor. A latent variable can be defined as “an underlying characteristic that cannot be
observed or measured directly; it is hypothesized to exist so as to explain [ manifest]
variables such as behavior, that can be observed.”(Warner, 2008: 754 citing Vogt,
1999:154-155) The results should show if the item statement is a suitable candidate for inclusion in measuring this underlying factor. The factor can then be named based on the type of information supplied by the inter-related variables. Therefore to determine the structure of the data, Principal Component Analysis (PCA) and Factor Analysis (FA) with Varimax rotation was done. PCA provides information about the variance that the
retained factors explain. This can provide insight into the number of factors that can be retained for the measured scale provided by Eigen values. In other words, the number of underlying factors present in the measured scale. An Eigen value of one (1) or over is highly desirable and a value of .9 is acceptable. However, as said before, retained factors can be used in analysis based on theoretical and conceptual issues and it is desirable that retained factors should explain a range of about 40% -50% of variance (Warner, 2008).
The number of components or factors retained in the model by no means exhausts the number of variables that could be used to measure the pertinent underlying constructs. For the purposes of the study, they were considered adequate based on theoretical and conceptual issues. This decision was supported when, during an interview, without any prompting, a highly educated respondent said the questionnaire was “good” and took into consideration his, and the community’s issues. Secondly, in the pre-tests, respondents said it captured pertinent community issues concerning the redevelopment. In this regard, the instrument ensures face validity by measuring what it is supposed to measure.
According to Warner (2008: 864), “face validity is sometimes desirable, when it is helpful
for test takers to be able to see the relevance of the measurements to their concerns, as in
some evaluation research studies where participants need to feel that their concerns are
being taken into account.” Additionally, factor analysis enables construct validity. The FA
and PCA tests yielded only one (1) factor/component for the dependent variable ‘Public
Acceptance’. For the independent variable Access to the Decision making process’ two
(2) latent factors were identified. One I named ‘Influence Criteria’ and the other ‘Normative Criteria’. Normative means, how things ought to be; that is, how the community participation exercise ought to be regarding standards that ought to be followed. Applicable here are implied issues of fairness and justice terms of procedural democracy. The variable ‘Influence Criteria’ considers the individual’s perception of internal control in the process. Beierle and Konisky (2000:590) states, Process attributes
are those over which agencies and participants have considerable control when
participants”.
During exploratory analysis of the data to determine the method of inferential statistics to be used in analysis, the missing scores for three respondents for the independent variable were substituted with the mean score for the respective
municipality. Histograms were used to check for the distribution shapes, box plots to observe outliers, and assessment for equality/similarity of group variances tests for violations were done. Cross tabulations were also done to check for violated
consistencies. Based on the results, and because of some violations of the data, the non- parametric analysis for hypothesis testing was done. These are Chi-square for testing relationships, Kruskal – Wallis (H) tests for differences in means and Spearman’s Rho (the non- parametric equivalent of Pearson’s r).