Quantitative Data Analysis (Using SPSS) Albard Khan, M.Ed Saturday, 7 November 2015
Quantitative versus qualitative research 1. Quantitative research is explaining phenomena by collecting numerical data that are analysed using mathematically-based methods (in particular: statistics) (Allaga & Gunderson, 2002). 2. Qualitative research relies heavily on the views of participants collects data consisting largely of words (or text) from participants, and analyse these words for themes (Creswell, 2008, p. 46).
Quantitative versus qualitative(cont.) 1. Quantitative research is concerned with numerical data, or numbers. 2. Qualitative research is interested with words or texts. 3. SPSS is exclusively a quantitative data analysis tool. 4. What about qualitative research such as interviews, case studies, ethnographic research, and discourse analysis; what software is to be used? 5. For qualitative research, use other softwares such as nvivo.
Pair Discussion With the person next to you, discuss if the following research is quantitative, qualitative, or mixed? The following excerpts were taken from educational research reports. They are either taken from the abstract or the full report of the research.
1 A set of policy documents was collected and analysed against literature on international schools and education decentralization. Kustulasari, 2009, The international school standard project in Indonesia: A document analysis
2 In this paper, we use data from Indonesia to examine the effectiveness of public versus private schools. We use labour market earnings as our measure of effectiveness. Beare, 2000, The effectiveness of private versus public schools: The case of Indonesia
3 A survey questionnaire was designed to elicit perceptions about the impact of prior learning, competence in language and communication, quality of student-staff relations and cultural interactions on student learning. Ramburuth & Tani (2009), The impact of culture on learning: exploring student perceptions. Multicultural Education & Technology Journal, Vol. 3 Iss: 3, pp.182-195
4 Data were collected during one academic semester through in-depth interviews, a focus group interview, classroom observations, and collection of relevant documents. Tatar, 2005, Classroom participation by international students
5 In stage 1, 1209 questionnaire responses were received from Y1 and Y2 students across these institutions in the identified subjects. In stage 2 we interviewed the students at key decision making moments. Reay, 2008, The socio-cultural and learning experiences of working class students in higher education
6 The project interviewed 162 first-year students at the University of the Arts, London, a university with a high proportion of international students. Sovic, 2009, Hi-bye friends and the herd instinct: international and home students
7 A representative sample of undergraduate and postgraduate international students at a large Australian university (n = 979, 64% females) completed a mail-back survey examining their perceptions of social connectedness. Rosenthal, 2009, Social connectedness among international students
What QUANT research can answer: 1. When we want quantitative answer, e.g. how many Tarbiyah students came from SMA? How many from MA? 2. When we want to study numerical change, e.g. is academic achievement going up or down this year? 3. When we want to explain phenomena with many factors, e.g. what factors predict students English performance? 4. When we want to test hypotheses, e.g. is that true that a romantic person tends to have higher GPA?
What QUANT research CANNOT answer: 1. When we want to explore a problem in-depth. 2. When we want to develop theories and hypotheses. We have to study the literature or conduct qualitative study. 3. If the issues are particularly complex. 4. When we want to reveal the meaning of particular events and circumstances.
Do you agree or do you not agree? What you cannot measure doesn t exist. (Anonymous)
Research Design: Experimental, Quasi-experimental and Nonexperimental
Experimental Design Definition A test under controlled conditions that is made to demonstrate a known truth or examine the validity of a hypothesis. (Muijs, 2004, p. 13). Experiment group vs control group As in natural science experiments, control groups receive no treatment while experiment groups do. Purpose To establish causality between variables. Usually involves pre-tests experiment post-tests.
Quasi-experimental Design Definition This is almost the same as the experimental design in that extraneous factors are controlled, but it differs from the experimental design as allocation of group is not randomized. Experiment group vs comparison group As in natural science experiments, comparison groups receive no treatment while experiment groups do. Purpose To establish causality between variables. Usually involves pre-tests experiment post-tests.
Non-experimental Design 1. Survey research 2. Observational research 3. Dataset analysis Today, our focus is on survey research, as it is the most popular and easiest method to conduct compared to experimental and quasiexperimental design. In non-experimental design, we re interested in relationship or correlations among variables, not causality.
The online questionnaire
Let s open the questionnaire 1. How many sections are there? Three sections: Demographic information (Name, GPA, Gender) Romantic Beliefs Scale Theories of Intelligence Scale 2. Romantic Beliefs Scale is unidimensional (measuring only one domain/construct: romanticism). 3. Theories of Intelligence Scale is also unidimensional, measuring whether someone has incremental or entity theory of intelligence.
Research Questions
Discuss in small groups By looking at the questionnaire, figure out what questions can be answered using the questionnaire/instrument!
Descriptive and Inferential Statistics A. DESCRIPTIVE STATISTICS 1. How many respondents are male and female? 2. What is the mean of participants GPAs? 3. Is there a difference between male and female GPAs? 4. In general, how romantic are the respondents? 5. In particular, is there a difference between male and female romantic belief levels? 6. In general, what is the type of respondents theories of intelligence? 7. In particular, is there a difference between male and female theories of intelligence?
Descriptive and Inferential Statistics (cont.) B. INFERRENTIAL STATISTICS 1. Is the difference between male and female GPAs significant? 2. Is the difference between male and female romantic belief levels significant? 3. Is there a correlation between romantic beliefs and GPA? If there is, how strong, in what direction and how significant? 4. Is there correlation between romantic beliefs and intelligence beliefs? If there is, how strong, in what direction and how significant? 5. Is there a correlation between intelligence beliefs and GPAs? If there is, how strong, in what direction and how significant?
Coding and Recoding
1. Only numbers can be analysed. You have to assign numbers to the data. 2. That s why, we need to create a codebook. 3. MS Excel is the first place to go before SPSS. 4. In Theories of Intelligence Beliefs Scale, one half are quite the opposite the other half. (We ll do recoding on SPSS). 5. Types of Data: Nominal, Ordinal, Scale (separate slide, maybe later). 6. Let s start creating a codebook. 7. Now let s copy the data into SPSS.
Research questions answered
Let s answer these questions (1) A. DESCRIPTIVE STATISTICS 1. How many respondents are male and female? 2. What is the mean of participants GPAs? 3. Is there a difference between male and female GPAs? 4. In general, how romantic are the respondents? 5. In particular, is there a difference between male and female romantic belief levels? 6. In general, what is the type of respondents theories of intelligence? 7. In particular, is there a difference between male and female theories of intelligence?
Let s answer these questions (2) B. INFERRENTIAL STATISTICS 1. Is the difference between male and female GPAs significant? 2. Is the difference between male and female romantic belief levels significant? 3. Is there a correlation between romantic beliefs and GPA? If there is, how strong, in what direction and how significant? 4. Is there correlation between romantic beliefs and intelligence beliefs? If there is, how strong, in what direction and how significant? 5. Is there a correlation between intelligence beliefs and GPAs? If there is, how strong, in what direction and how significant?
Further topics for your personal studies... Parametric and non-parametric tests Factor analysis Reliability analysis
Thank You