“Should I do a factor analysis?”

The following set of questions was sent to me by email from a colleague in a university in the Gulf:

I am conducting a research where I am trying to compare expectations of the teachers and expectations of the students about English studies under five different categories. The questions  are designed differently, but mostly on a likert scale with four options (Strongly Agree to Strongly Disagree). The survey was distributed through Qualtrics which calculated the median for me. I contacted a statistician and he suggested Factor Analysis for statistical significance. […]Here are the questions that I wanted to ask:

(1) Is Factor Analysis a perfect test for this study?
(2) How can I spread data on an SPSS sheet? Screen shots will be very helpful in this regard.
(3) Running the test-a step by step process (screen shots).

As I have written often in this blog, the kinds of statistical procedures one should use depend on research questions. Factor analysis can used in order to discover ‘hidden’ psychological constructs (latent variables) in a data set, or to confirm whether such constructs exist. However, the research objective is “to compare expectations”, and it does not involve finding hidden constructs, as far as I can tell. It is difficult for me to see how Factor Analysis would be useful to answer this question.

It would be possible to use ‘confirmatory’ factor analysis if you want to make sure that all the items in each scale (the observed variables) are measuring the same thing (latent variable). However, the interpretation of factor analysis output is not always straightforward. In fact it has been suggested that, inexpertly done, factor analysis is no more useful than a set of tarot cards. Besides, there are simpler methods to confirm the internal consistency of each scale (e.g., estimate Cronbach’s α).

Note that even if you do perform a confirmatory factor analysis that will not answer your research question, which involves comparing two populations. One way to make this comparison, after consolidating the data, is a cross-tabulation, followed up by an x-square test (or similar) to check for statistical significance.

A good resource for learning how to use SPSS is Andy Field’s Discovering Statistics Using SPSS. Chapter 3 describes the SPSS environment, and Chapter 17 explains how to do Factor analysis. Doing Quantitative Research in Education with SPSS by Daniel Muijs is a similarly useful resource. Chapter 11 deals with Factor Analysis.

If you are interested in buying any of the books above, you may want to consider following the links from this post to Amazon.com to make your purchase. Amazon will provide me with a small fee (I think it’s around 4%) on any purchases made from links in my blog. Any revenue from these sales will go towards funding this blog. 

Featured image: Programmable calculator by Weedwhacker128 [CC BY-SA 3.0 or GFDL], via Wikimedia Commons

Recently read: On King’s London, bogus science, author mills, and predatory publishers

There are a number of misconceptions about academic and scientific worlds, and the round-up of stories from this week’s academic journalism seeks to address some of them. The first one is that the people running the universities know what they are doing (hmmm!). The second story questions the veracity of the science that is reported to the public. The third one, focussing on academic fraud, challenges assumptions of honesty in scientific publishing; and the last story invites us to reflect on the costs of disseminating academic knowledge.

King’s College London no longer a college?

A bizarre story that made the rounds this week concerned the plan, by King’s College London (KCL), to rebrand itself as King’s London. Continue reading Recently read: On King’s London, bogus science, author mills, and predatory publishers

Teaching English to Young Learners: some insights from the literature

In the space of the last couple of years there has been a surge of interest in Teaching English to Young Learners (TEYL). For instance, in IATEFL 2014 there was a debate about ELT in primary schools (Primary ELT does more harm than good), and many major journals in the field have run special issues on topics such as Age in Second Language Learning (Applied Linguistics), and Special Issue on Teaching English to Young Learners (ELT Journal). My own interest in TEYL stems from my professional involvement in piloting various early ELT projects in primary schools in Greece, and – on a more personal level – from observing my daughter’s engagement with Primary ELT. This is as good a place as any to put a disclaimer: So far, I have been unimpressed by the ways in which TEYL is implemented in Greece. But what I want to do in this post is look into the question of whether TEYL projects in other settings have been more successful, and how the sub-par implementation of TEYL in Greece might be improved.

To that end, in this post, I summarise information from the 2014 special issue of Studies in Second Language Learning and Teaching (Vol. 4 No. 3), which came out this October and focussed on Age and Language Acquisition. There are seven articles in the issue, of which four focus on instructed second language learning. These are presented below, and in the section that immediately follows I synthesise their findings. This is followed by some comments, from the articles, about language education policy, which seems to be driven by political rather than pedagogical motivations. I conclude by relating all of the above to the language education in Greece. Continue reading Teaching English to Young Learners: some insights from the literature

Αχιλλέας Κωστούλας Ιστοσελίδα και Ιστολόγιο


Get every new post delivered to your Inbox.

Join 2,241 other followers