Cronbachs Alpha Reliability Analysis Psychology Essay

Scale measurement is an important element in survey. The survey in this study involved 3 sections, which are section A, section B and section C. Nominal scale was used in survey section A, and ordinal scale was used in survey section B and C.

Nominal scale is scale that labels the items rather than scales it (Howell, 2009). It includes male or female and yes or no. Normally it brings no meaning. Ordinal scale is simple rank model. The objects, individuals or events are categorized. It is known as ranked data as the ranking is ordered from highest to lowest or smallest to biggest (Jackson, 2011).

Section A was created to gather the socio-demographic characteristics of respondent such as gender, ages, location of study in the UTAR and education level.

Figure 3.1

Select your Gender:

Male

Female

There were five dimensions in Section B which contained 20 statements to measure the independent variables. The sequence of these dimensions was social influence, environmental concern, environmental attitude and self-image. In the Section C are contained 5 statements to measure the dependent variable which was green purchasing behavior.

In this research, 5-point likert scale is used (5=strongly agree, 4=agree, 3=neutral, 2=disagree and 1= strongly disagree). A 5-point Likert scale is good enough as increase in scale may confuse respondents (Hair, Bush & Ortinau, 2003). Likert scale can achieve high degree of validity and reliability (Sarantakos, 1993; cited by Kronberg, 2011).

Figure 3.2

3.7 Data Processing

After all data had been collected from a sample of the target respondents, data processing is the next step to ensure the data in the standard of quality and complete.

3.7.1 Checking

The first step of data processing is to check on questionnaire before it has been distributed. Hence it can reduce the unqualified data such as typing error, inconsistent questions and incomplete content. Therefore, the mistakes can be detected and corrected in advance. In addition, the questionnaire has been passed to our supervisor to double check. Collected data needs to be checked whether the feedback from target respondents are valid for the research. Pilot test has been used to test the reliability and accuracy of the questionnaires. The objective of data checking is to ensure the questionnaires are in quality and completeness. In this stage the grammar error corrections will be done.

3.7.2 Editing

Editing is a process of examining the collected raw data to detect errors and omissions in the information returned by the respondents of the research, and to correct the data when possible.

3.7.3 Coding

The next process is data coding which involves assigning a number to the participants’ response so that the data can be entered into a database. The code includes an indication of the column position (field) and data record. For example, the five point likert scale of variable, “strongly disagree” codes as 1 and “strongly agree” codes as 5. This coding is simplifying storage of data with digit codes and easier for categorizing when used SPSS software.

3.7.4 Transcribing

Transcribing data is a process that transferring coded data from questionnaires or coding sheet into disks or computers for analyzing purpose. The data will pre-check to ensure the data is error free. When the data is transferred in SPSS software, it can be used for the research and generate the accuracy analysis from the questionnaire.

3.8 Data Analysis

The raw data collected in research will be further analyzed by statistical method. After the questionnaires were returned to the researcher, the data were recorded and entered into a Microsoft Excel spreadsheet, which was uploaded into Predictive Analysis Software (PASW, previously known as The Statistical Package for Social Sciences, SPSS) for more detailed statistical analysis. PASW is a good first statistical package for people who want to perform quantitative research in social science because it is easy to use (Cheah, 2009). In this study, the analysis utilized were descriptive statistics which included frequency distribution, measure of central tendency and measure of dispersion, Cronbach’s Alpha Reliability Analysis, Pearson correlation, and multiple regression analysis.

3.8.1 Descriptive Analysis

Descriptive analysis is used in the research to transform the raw data to a way that is meaningful (Zikmund, 2002). It may include graph, bar, pie chart, or any number that use to describe that raw data. It helps the researchers to summarize the study variables (Parasuraman, Grewal & Krishnan, 2004). So that researchers can get to know the results and make interpretation based on descriptive statistics.

3.8.1.1 Reliability Test

Reliability refers to dependency or consistency, which indicates that the same things is repeated or recurs under the identical conditions. The reliability of a measure shows the degree to which the measure is without bias (error free) and hence offers consistent measurement across time and across different items in the instrument (Sekaran, 2000). A reliable measure would show the stability and consistency with which the instrument measures the concept and help to access the ‘goodness’ of a measure. According to Sekaran (2000), a measure with reliability less than 0.6 is considered poor, it should be at least 0.70 or above.

The most widely used method to measure reliability is Cronbach’s alpha. Cronbach’s alpha is used for multipoint-scaled items (items in the scale are at least internal in nature). Generally, Cronbach’s alpha will increase when the correlations between the items increase. Alpha value can take values between negative infinity and 1, although only positive values make sense (Cheah, 2009). The value of Cronbach’s alpha should be at least 0.6 to be accepted, and the ideal value is 0.7 or above.

3.8.2 Descriptive statistics

Descriptive studies are quite frequently undertaken in education institutions to learn about and describe the characteristics of a group of students, such examples as the age and years of education in an education institution. There are three types of measures used in descriptive statistics: frequency distribution and measures of central of tendency and measures of dispersion or variability. Researcher was able to find the frequencies, percentages and determine the mean and standard deviation for the variables in the questionnaires by using descriptive statistics.

To understand the use of measurement terms, frequency distribution plays a critical role. According to Leech et al. (2005), frequency distribution is a tally or count of the number of times each score on a single variable occurs. The use of frequency distribution is to show the number of responses to each value of a variable. Normally, variable name, frequency counts for each value of the variable and cumulative percentages for each value related to a variable are shown by a frequency distribution. The distribution is said to be approximately normally distributed when there are small numbers of scores for the low and high values and most scores are for the middle values.

The mean is the arithmetic average of a set of data. Typically, the data shows some degree of central tendency with the most responses distributed close to the mean value (Hair, Money, Samouel and Page, 2007). The mean is said to be a “robust” measure of central tendency as in most instances, it is not sensitive to data values being added or deleted (Hair et al., 2007). According to Joseph et al. (2007), standard deviation describes the spread or variability of the sample distribution values from the mean, and is perhaps the most valuable index of dispersion.

3.8.3 Cronbach’s alpha reliability analysis

As mentioned earlier, the most commonly type of measurement of internal consistency reliability used is Cronbach’s coefficient alpha. When there are several Likert-type items that are summed, alpha is used to make a composite score or summated scale. According to Leech et al. (2005), alpha is based on the mean or average correlation of each item in the scale with every other item.

3.8.2 Inferential Analysis

Inferential analysis refers to data analysis which is used to test specific hypothesis (Parasuraman et al., 2004). According to Greer & Kolbe (2003), inferential statistics include drawing conclusions from information obtained in the data.

3.8.4 Pearson correlation analysis

Sekaran (2003) stated that when the researcher interested in defines the important variables associated with the problem, the study is called correlation study. For example, correlation studies can be undertaken to find out whether any relationship between gender and income level exists? If yes, positive or negative relationship? The index ranges in value from -1 to +1, with zero indicating absolutely no relation between two variables. This coefficient indicates the degree that low or high scores on one variable tend to go to with low or high scores on another variable. Pearson correlation was used to determine the relationship between the independent variables (social influence, environmental concern, environmental attitude and concern for self-image in environmental protection) and dependent variable (green purchasing behavior).

3.8.5 Multiple regression analysis

Multiple regression is a widely used statistical technique in sociology. The result of multiple regression can generate two things (Neuman, 2009). First, a measure called R-squared (R2) in the result can tell how well a set of variables explain a dependent variable. In other word, it shows the accuracy of predicting the dependent variable based on the information about the independent variables. For example, an R2 of .50 means that knowing the independent variables improve the accuracy of predicting the dependent variable by 50 percent (Neuman, 2009).

The second thing generated by multiple regression is the measurement of the direction and size of the effect of each independent variable on a dependent variable (Neuman, 2009). For instance, the way how five independent or controlling variables simultaneously affect a dependent variable, with all the variables controlling the effects of one another can be seen by researcher.

According to Neuman (2009), a standardized regression coefficient is used to measure the effect on the dependent variable. It is similar to a correlation coefficient (Neuman, 2009). A high standardized regression coefficient indicates a strong relationship between an independent variable and dependent variable.

3.9 Chapter Summary

Research methodology is important as it can help researcher to systematically resolve the research problem. In this chapter, research framework and hypotheses are clarified. The other sections of this chapter include data sources, sampling design, data collection and data analysis.