CAPÍTULO I. PLANTEAMIENTO GENERAL DEL PROBLEMA DE INVESTIGACIÓN
2. FUNDAMENTACIÓN TEÓRICA DEL OBJETO DE ESTUDIO
2.2. LOS GOLPEOS EN BÁDMINTON
2.2.1. Clasificación de los golpeos en bádminton
Data analysis is conducted to make sense of the raw data in order that valid conclusions and recommendations can be made. The present study followed four steps in the data analysis process, namely, preliminary preparation, reliability and validity analysis, descriptive analysis and inferential analysis.
2.8.1 Preparing the data for analysis
The mass of raw data obtained from questionnaires is meaningless (Tustin, Ligthelm, Martins & van Wyk 2005:451). It first needs to be converted into a suitable form before it can be subjected to statistical analysis (Malhotra 2010:451). The way in which data is prepared and converted for statistical analysis will determine the quality of results obtained as well as the subsequent interpretation (Aaker, Kumar & Day 2007:432). The major data preparation techniques include data editing, coding and if necessary, statistically adjusting the data (Aaker et al 2007:432).
During the editing process, all questionnaires are reviewed to check whether they have been completed correctly and whether they are complete (Wiid &
Diggines 2009:229). The objective of editing is to increase the accuracy and precision of questionnaires (Malhotra 2010:453). All questions are checked or edited before the responses are captured to determine whether the
recorded data is acceptable and useful and, if so, to prepare the data for coding and capturing (Tustin et al 2005:454). Unsatisfactory responses are handled by either adjusting or discarding the questionnaire (Wiid & Diggines 2009:229). The current study used all the questionnaires which were returned, as no unsatisfactory responses were noted.
Coding refers to the process of assigning a code, usually a number, to the various responses to a particular question (Malhotra 2010:454). In surveys where mostly closed-ended questions are used, the items are pre-coded (McDaniel & Gates 2008:396). In the present study all questions from Sections A and B were pre-coded, as respondents had to select an option from the statements provided. The data was captured in Microsoft Excel and then imported into the statistical software package “Statistica”, for the statistical analysis.
2.8.2 Reliability and validity
“Reliability is the degree to which measures are free from random error, and therefore provide consistent data” (McDaniel & Gates 2008:247). Less error results in more reliability and therefore a measurement that is error-free is a correct measure (McDaniel & Gates 2008:247). Three methods are used to assess reliability, namely, test-retest, the use of equivalent forms and internal consistency (McDaniel & Gates 2008:247). The test-retest method repeats the measurement using the same instrument, and approximating the original conditions as closely as possible (Malhotra 2010:318). “Equivalent form reliability is determined by measuring the correlations on the two instruments
whereas internal consistency reliability assesses the ability to produce similar results when different samples are used to measure a phenomenon during the same period” (McDaniel & Gates 2008:248). The present study used internal consistency reliability.
A popular approach to measuring internal consistency is to use the coefficient alpha (Cronbach alpha). “The coefficient varies between 0 to 1, and a value of 0.6 or less generally indicates unsatisfactory internal consistency reliability” (Malhotra 2010:319). Cronbach alpha coefficients were calculated to assess the reliability of the research instrument. The resulting coefficients ranged from 0.76 to 0.89. The factor, price, was the only exception, yielding a Cronbach alpha coefficient of 0.60. Four items were omitted and deleted from this factor and the remaining items yielded a Cronbach alpha coefficient of 0.81.
A research instrument has validity when it measures what it is supposed to measure. Validity can be assessed from different perspectives, including content, criterion-related and construct validity (McDaniel & Gates 2008:250).
Content validity is sometimes referred to as face validity: it is concerned with whether the scale provides adequate coverage for the task at hand (McDaniel & Gates 2008:250). “Criterion-related validity reflects whether a scale performs as expected in relation to other variables selected as meaningful data” (Malhotra 2010:320). Construct validity is established if a measure behaves according to the theory behind it (McDaniel & Gates 2008:252). Construct validity includes convergent, discriminant and
nomological validity (Malhotra 2010:321). In the present study the discriminant validity of the questionnaire was assessed. Discriminant validity uncovers the extent to which a measure does not correlate with other constructs from which it should differ (Malhotra 2010:321).
Factor analysis was used in this study to assess the discriminant validity of the questionnaire. A factor analysis is a group of procedures primarily used to reduce and summarise data (Malhotra & Birks 2006:572). The factor analysis groups variables with similar characteristics (Tustin et al 2005:668.) The relationships among sets of interrelated variables are examined and represented by a few underlying factors (Malhotra & Birks 2006:572). In this study, exploratory factor analysis was performed, using Statistica V.10, to determine the underlying factor structure in the data. However the EFA did not yield the expected factor structure. The correlations between the items were fairly strong and no clear factor structure was found. Item–total correlations were subsequently determined to test the internal reliability of each anticipated factor. The results are shown in Table 2.2.
TABLE 2.2
ITEM – TOTAL CORRELATIONS AND CRONBACH ALPHAS
ITEM
ITEM TOTAL CORRELATION
DECISION MAKING AND PURCHASE BEHAVIOUR 1 I recognise the importance of buying environmentally-friendly
products. 0.64
2 I search for information on environmentally-friendly products. 0.64 3 I shop around for environmentally-friendly alternatives. 0.71 4 If possible I would like to buy environmentally-friendly products. 0.67 5 After I buy environmentally-friendly products I feel very happy with
myself. 0.69
Cronbach alpha 0.85
NORMS
6 Most of my friends think consumers should use products that are
safe for the environment. 0.63
7 Most of my friends expect consumers to recycle. 0.70 8 Most of my friends think consumers should be interested in the
environmental consequences of their purchases. 0.73 9 I feel a personal, moral obligation to do whatever I can to help
improve the environment. 0.67
10 I feel a personal, moral obligation to recycle household waste. 0.61 11 I have decided to only purchase environmentally – friendly
products in an attempt to contribute to protecting the environment. 0.60
Cronbach alpha 0.86
ENVIRONMENTAL CONCERN
12 I feel angry when I think of the ways industries are polluting the
environment. 0.63
13 I am concerned about environmental issues. 0.77
14 I contribute to slowing down pollution wherever I can. 0.79 15 Whenever possible, I purchase products which I know are not
harmful to the environment. 0.78
16 I have changed my buying behaviour because I am concerned
about the environment. 0.66
Cronbach alpha 0.89
ENVIRONMENTAL KNOWLEDGE
36 I know when a product is harmful to the environment. 0.42
37 I know what the causes of pollution are. 0.63
38 I know when I do something that can harm the environment. 0.67 39 I know what the causes of global warming are. 0.62 40 I will produce less carbon dioxide if I do not use my car. 0.52
41 I know what “carbon footprint” means. 0.40
Cronbach alpha 0.78
TABLE 2.2 (CONTINUED)
ITEM – TOTAL CORRELATIONS AND CRONBACH ALPHAS
ITEM
ITEM TOTAL CORRELATION
PRICE
17 Price is an important factor in my decision to buy or not to buy
environmentally-friendly products. 0.20
18 Although environmentally-friendly products cost more than my
regular products I would (still) purchase them. 0.39 19 If environmentally-friendly products cost the same as regular
products, I would purchase them. 0.39
20 I am willing to pay more for environmentally-friendly products
than for regular products. 0.44
21 I am willing to pay 15-20% more for environmentally-friendly
products. 0.47
22 Environmentally-friendly products are too expensive for me to
buy. 0.06
23 Consumers should pay more for products that can harm the
environment 0.28
Cronbach alpha 0.60
PRODUCT
24 When I have a choice between two equal products, I always
purchase the one which is least harmful to the environment. 0.52 25 I do not buy products in aerosol containers. 0.41 26 Wherever possible, I buy products packaged in reusable
containers. 0.72
27 Wherever possible I buy products in containers that are
recyclable e.g. glass, carton. 0.66
28 If environmentally–friendly products perform the same as my
regular products, I would purchase them. 0.51
Cronbach alpha 0.78
PROMOTION
32 I often see advertisements about environmentally-friendly
products. 0.59
33 Environmentally-friendly products are marketed to me in a way
which I find really engaging and relevant to my lifestyle. 0.71 34 I feel I can believe the claims made in advertisements
promoting environmentally-friendly products. 0.38
35 Most advertising makes it clear whether the product is
environmentally-friendly or not. 0.60
Cronbach alpha 0.77
PLACE
29 I am willing to spend considerable time and effort to buy
environmentally-friendly products. 0.54
30 I am willing to buy environmentally-friendly products if they are
readily available. 0.69
31 I am willing to buy environmentally-friendly products if they are
available at the supermarkets where I shop. 0.56
Cronbach alpha 0.76
All Cronbach alpha coefficients were above 0.70 (Table 2.2), and were regarded as acceptable (Christmann & Van Aelst 2006:1661) except for the factor price, which was the only exception, yielding a Cronbach alpha coefficient of 0.60. Low inter-item correlations (<0.30) were indicated for items 17, 19, 22, 23. As a result these four items were omitted from subsequent analysis. The remaining items (18, 20 and 21) correlated quite strongly and were therefore retained. Following the deletion of the four problem items, the Cronbach alpha for price improved to 0.81.
Because of the good internal reliability of the factors, all the anticipated person and marketing related factors were retained. Table 2.3 shows the mean scores and standard deviations for all these factors. Table 2.4 shows the correlations among the factors.
TABLE 2.3
MEAN SCORES AND STANDARD DEVIATIONS: ALL FACTORS
FACTOR MEAN STD. DEV
NORMS 3.03 0.66
CONCERN 3.32 0.65
KNOWLEDGE 3.27 0.53
PRICE 3.02 0.89
PRODUCT 3.02 0.66
PLACE 3.22 0.65
PROMOTION 2.80 0.71
As shown in Table 2.3, all the mean scores were fairly close to the middle of the five-point scale. Environmental concern attracted the highest mean score (M=3.32) and promotion the lowest (M=2.80).
TABLE 2.4
CORRELATIONS AMONG PERSON AND MARKETING RELATED FACTORS
All the factors were strongly and statistically significantly correlated at p<0.05 (See Table 2.4). Especially high correlations were noted among norms and environmental concern.
2.8.3 Descriptive analysis
Descriptive statistics summarise, organise and simplify data (Gravetter &
Wallnau 2005:5). These are techniques that organise or summarise raw scores into a form that is more manageable (Gravetter & Wallnau 2005:5).
Descriptive statistics are usually associated with frequency distribution and include measures of central tendency (mean, median and mode), measures
Factors
Norms Concern Knowledge Price Product Promotion Place
Norms 1.00
Concern 0.79 1.00
Knowledge 0.53 0.62 1.00
Price 0.61 0.55 0.36 1.00
Product 0.61 0.66 0.51 0.53 1.00
Promotion 0.48 0.36 0.35 0.38 0.38 1.00
Place 0.64 0.69 0.62 0.45 0.66 0.39 1.00
of dispersion (range, standard deviation and coefficient of variation), and measures of shape (skewness and kurtosis) (Aaker et al 2007:438). In this study, descriptive analyses relate to the respondents’ demographic characteristics and the mean scores for different factors.
2.8.4 Inferential analysis
After results have been described, the outcomes have to be interpreted. This is the role of inferential statistics (Gravetter & Wallnau 2005:8). Inferential statistics are used to make inferences about the population on the grounds of what has been observed (Tustin et al 2005:559). One of the most commonly used inferential procedures is hypothesis testing (Gravetter & Wallnau 2005:178). The current research had 10 hypotheses (section 1.4), which were tested.
It is important to use the most appropriate statistical technique to test the hypothesis (Tustin et al 2005:583). The statistical techniques employed in the current research included regression analysis, t-tests, ANOVA and Tukey’s alternate procedures. Regression analysis is a procedure used to analyse the relationships between metric dependent variables and one or more independent variables. Furthermore, it determines the mathematical equation relating to both the dependent and independent variables, and is used to predict the values of the dependent variables (Malhotra 2010:568). In the current research a single regression, t-tests and analyses of variance (ANOVA) were used.