1.3 UN ACERCAMIENTO A LAS PROPUESTAS CIENTÍFICAS DE LA CONSTRUCCIÓN DEL CONOCIMIENTO COTIDIANO
1.3.2 “LO SOCIAL” COMO PROTAGONISTA DE LA CONSTRUCCIÓN DEL CONOCIMIENTO COTIDIANO: EL MODELO DE LAS
Evaluating the quality of the survey and establishing trustworthy findings are achieved via testing the reliability and validity of the survey findings (Robson, 2011; Zohrabi, 2013). There is a relation between reliability and validity, however, if a questionnaire is valid, that does not mean it is always reliable (Neuman, 2003; Saunders et al., 2012). Figure 5.6 below illustrates the stages of a question’s test for its validity and reliability:
Figure 5.6: The stages for a question to be valid and reliable. (Saunders et al., 2012) Reliability is concerned with consistency of the findings obtained from given research (Collis and Hussey, 2009; Robson, 2011; Saunders et al., 2012; Zohrabi, 2013). In order to assess the reliability, three common approaches are outlined (Saunders et. al., 2012 cited Mitchell, 1996):
• Test re-test approach;
• Alternative form Approach;
• Internal consistency approach.
Test re-test and alternative form approaches have major limitations such as requiring the questionnaire to be completed twice by participants, which could create problems, or requiring two alternative forms of the same questions, which is difficult to ensure the questions are substantially equivalent (Saunders et al., 2012). That made internal consistency approach a useful form of reliability assessment. Hence, the internal consistency approach is adopted for this study.
Researcher is clear about the data required and designs a question
Respondent answers the question
Respondent decodes the question in the way the
researcher intended Researcher decodes the
answer in the way the respondent intended
The internal consistency approach measures the consistency of responses across all the questions from the questionnaire by calculating a reliability coefficient called Cronbach’s alpha (Saunders et. al., 2012). In this study, SPSS (Statistical Package for the Social Sciences) was used for computing Cronbach’s alpha to perform an internal consistency analysis for the responses to all the questions from the questionnaire. Appendix E shows the calculations of Cronbach’s alpha in SPSS.
Cronbach’s alpha values vary between 0 and 1, where values of 0.7 and above indicate that the questions combined in the scale are measuring the same thing (Saunders et. al., 2012). However, when the value of Cronbach’s alpha is less than 0.7, the reliability can be increased by removing an item or more from the questionnaire. The analysis was performed for all questions, and the values of Cronbach’s alpha were greater than 0.7. This indicates that the responses for all questions have internal consistency; therefore, the results from the analysis have internal consistency, and are thus reliable.
Furthermore, the interviews with specialists in pavement maintenance management indicate similar outcomes. Therefore, reliability is obtained for the questionnaire survey.
Items Included in the Questionnaire Original Cronbach’s Alpha Value Items for Deletion Cronbach’s Alpha Value if Item Deleted Remaining Service Life
0.740
_ 0.740
Road Condition Indicator (RCI) _ 0.740
Type of Deterioration _ 0.706
Observed Deterioration Rate _ 0.729
Traffic Diversion _ 0.726
Importance/Classification of Road _ 0.725 Annual Average Daily Traffic (ADDT) _ 0.718 Possible Conflict or Overlap with Other Road
Works _ 0.729
Risk of Failure _ 0.722
Safety Concern _ 0.725
Accident Rate (related to surface condition) _ 0.713
Scheme Cost _ 0.725
Available Funding _ 0.723
Whole Life-cycle Cost _ 0.726
Table 5.5: The Results of Reliability Analysis
Validity is concerned with whether the research measures or evaluates what the researcher intended to measure or evaluate (Badri, et al. 1995; Collis and Hussey, 2009; Saunders et al., 2012; Zohrabi, 2013). In order to assess the validity, three methods are outlined (Badri et al., 1995):
• Content validity;
• External validity or Predictive validity or Criterion-related validity;
• Construct validity.
Content validity depends on the judgements and evaluations of the researchers on whether the measurement instrument (the questionnaire) provides adequate coverage for all aspects of each item being measured. Judgement can be made through an
extensive review of literature and prior discussion with others, or by using a panel of individuals to assess the measures (Badri et al., 1995; Saunder et al,. 2012).
In this research, the determination of the measurement criteria in the survey was based on literature review. Therefore, a pilot survey was conducted to seek feedback from two postgraduate researchers from the University of Manchester and three specialists in pavement maintenance within local road authorities to assess the measurement items, and whether the items cover all the investigative questions. The items were refined and edited according to the participants’ feedback and evaluation, therefore, it can be said that the survey measures in this research have content validity.
External validity is concerned with whether the outcomes of a work of research are applicable in other settings (Zohrabi, 2013). The method seeks how representative the surveys are, and whether the selected participants are appropriate. In this study, profiles of participants are highly relevant to the context. Participants in the questionnaire survey involved specialists in pavement maintenance from different local road authorities. Therefore, it can be said that the representation capability of the questionnaire for the outcomes is externally valid.
Construct validity is concerned with whether a measure measures the theoretical construct that it was designed to measure (Badri, et al., 1995). In this study, interviews are conducted for triangulation purposes, and the outcomes of the interviews showed similar trend questionnaire results. Therefore, the questionnaire survey can be considered to have construct validity.