My research involves assessing banking customers' perceptions on the use of social media by banks. So what would be a reasonable sample in this case since I can not readily quantify how many customers there are in the different banks in my country.
Sample size is the small fraction of the population which is considered a vital element to reduce the sampling error. As you are assessing the customers' perception, your population is unknown. Roscoe (1975) provides the ‘rule of thumb’ for determining sample size; as it is declared that sample size larger than 30 and smaller than 500 are appropriate for most of the studies. Whereas, some statistical experts suggest a data range between 5-10 times the number of items used in the scale (Hair, Black, Babin and Anderson, 2010). Thus, it is always advisable that the number of respondents or sample size should be 10 times of the number of items in the instrument.
very nice justification, also there is another way clarified it by Haire et al., 2011 you should multiply every item in the instrument of the study 5 times, and it better 10 tomes but the best way if you multiply every question 20 times. moreover, you can use G-power to calculate the sample size and it is very good way because this program will take into account the number of variables the you will use it, and the way of the analysis that you will use it as will.
Also, Bollen (1989) who stated that empirical ratio of estimates parameter has to have at least 10 observations for every variable, this is similar to Roscoe (1975) because Roscoe (1975) suggested the same way
Estadisticamente hablando una muestra es representativa cuando cubre el 30% o mas de la poblaciòn total, pero cuando no se sabe el tamaño de la poblaciòn total, se puede recurrir a planteamiento emiricos que otros investigadores han considerado, por ejemplo la experiencia, porejempli cosiderar al menos 10 observaciones para cada variable inestigada
The best way to test for an optimal sample size is to test whether by increasing the sample size you are getting any appreciable improvement in the fit of your model. Because statistical results mostly use the normal distribution assumption which is symmetric.
S.K.Mallick
for S.K.Mallick, N. Hamburger, S.Mallick
if you are to use advanced multivariate models like Structural Equations Modeling, minimum sample size should be 250( Hair et.al.,)
where the constructs are more than six, sample size should be more than 500
best regards
For your study, anything above 400 would be a safe sample. This is based on a 5% margin of error.
Please see the attached document for more information.
1- Estimating Sample Size based on proportion
2- Estimating Sample Size based on mean
http://documents.routledge-interactive.s3.amazonaws.com/9780415628129/Chapter%209%20-%20Determining%20sample%20size%20final_edited.pdf
Monika Dhochak My instrument has 26 items, Does it safe to have a sample size of 130-260? Thanks!
https://archive.org/details/fundamentalresea0000rosc/page/157/mode/1up?q=500
- Badges
- Science topic
- Similar topics
- Publishing