In the research paper: CREATIVE SPACE IN DESIGN EDUCATION:
A TYPOLOGY OF SPATIAL FUNCTIONS. 9 design students were included for this test which conclude with five different types of creative spaces. Why this sample? why not more or less?
I'm conducting a similar experiment with Design students and still can't decide the size of particeipants.
Theoretically, more sample leads to higher confidence.
So the question is what confidence interval you are targeting.
In some cases you can have a estimation of the sample size to reach your target. Read Estimation of a mean from here: https://en.wikipedia.org/wiki/Sample_size_determination
However, whatever sample size you pick, after your analysis you need to report on the "effect size" which will suffer from lack of sample.
You can also read on "statistical power" or "power analysis" to determine the minimum sample size.
Depends on the required confidence level. Can be as few as 5 for "indicative only" then in steps at 15 and 50 with T-tests etc to establish reliability of results obtained from that sample population.
The answer of Mr. Rob Cowdroy is well defined. Furthermore, you may go through the " An Easy Way to Maximize the Accuracy and Power of Your Research" at
http://psych.colorado.edu/~willcutt/pdfs/erceg_hurn_2008.pdf
Thank you all for the answers they were really helpful.
I'm designing an experiment to conduct with students in regards to my PhD research.
I want to know what is the acceptable number of participants I can relay on for a PhD study?
Specially more students means higher cost on me which I might not be able to cover or it would prevent me from conducting the experiment again after analysising the initial findings.
Any suggustions?
You can take a look at the related works that are published and follow their sample sizes. It is impossible to answer your question since we don't know anything about your experiment. you need to have a rough estimation of the effect size (i.e., the distance between the means of your null and alternative hypothesis).
Moreover, the paper that you mentioned initially does not perform any statistical test. You need to design you protocol first (e.g., deciding on you independent and dependent variables) and see how you are planning to quantify your variables.
For example,
I'm adapting Thoring et al.(2015) research methodology :
Conference Paper Workspaces for Design Education and Practice
They conduct the Cultural probs + workshop with 9 participants from different institutions and different design background and then in-depth discussion with all the 9 participants.
They are experts in this field and the participants themselves are experts in the design field so they would respond better (I suppose) than the students (my participants) as per their experience and background.
In the otherhand, in 2012, Thoring et al conducted a similar methodology with 18 design students from two different institutions.
Conference Paper Where do we learn to design? A case study about creative spaces
did you get what I mean?for example is 18 students would be acceptable for a PhD level?
or I would need either to recruit more students? or maybe redo the experiment more than one time?
They don't perform statistical analysis in none of the papers that you mentioned. They report "Although the limited number of participants does not allow for a statistical analysis of the results, some preliminary insights have been identified through the cultural probes that could partly be further clarified in the follow-up workshop".
So, according to them, if you want to perform statistical analysis under the same condition, you need more participants.
If you want to have similar preliminary insights, then you can conduct your experiment with similar sample size.
- Badges
- Science topic
- Similar topics
- Methodology
- Research Methodology
- Research Papers