If i intend to carry out a cross sectional comparative study (analytical study design) to assess the association of cardiac diseases among young stroke patients, where do I begin? Do I enroll stroke patients in a medical ward and then select old and younger stroke patients based on an age demarcation and then proceed with the analysis? In that case, how do I calculate my sample size? Do I use the formula:
n = P(1-P)Z2/(error)2 for descriptive, hospital-based cross-sectional study or,
do i need two proportions for two different groups (P1, P2) and use this one instead:
n=(Z1-a+Z1-b)2 (P1Q1+P2Q2)/ (P1-P2)2
[a=alpha=probability for type I error, b=beta=probability for type II error]
If the latter equation is to be used, then how do I find out P1 and P2 from previous studies to construct a sample for this study? What would P1 and P2 stand for this study? P1=young stroke patients? P2=old stroke patients??
Sample Size Rule
Sekaran (2013) wrote:
"Roscoe (1975) proposes the following rules of thumb for determining sample size:
1. Sample sizes larger than 30 and less than 500 are appropriate for most research.
2. Where samples are to be broken into sub-samples;(male/females, juniors/seniors, etc.), a minimum sample size of 30 for each category is necessary.
3. In multivariate research (including multiple regression analyses),the sample size should be several times (preferably 10 times or more) as large as the number of variables in the study.
4. For simple experimental research with tight experimental controls (matched pairs, etc.), successful research is possible with samples as small as 10 to 20 in size."
Reference
Sekaran, U., 2003. Research methods for business: A skill building approach. John Wiley & Sons.
Sample Size Rule
Sekaran (2013) wrote:
"Roscoe (1975) proposes the following rules of thumb for determining sample size:
1. Sample sizes larger than 30 and less than 500 are appropriate for most research.
2. Where samples are to be broken into sub-samples;(male/females, juniors/seniors, etc.), a minimum sample size of 30 for each category is necessary.
3. In multivariate research (including multiple regression analyses),the sample size should be several times (preferably 10 times or more) as large as the number of variables in the study.
4. For simple experimental research with tight experimental controls (matched pairs, etc.), successful research is possible with samples as small as 10 to 20 in size."
Reference
Sekaran, U., 2003. Research methods for business: A skill building approach. John Wiley & Sons.
Hello Ishtiaque,
I recommend software such as the G*Power program (freely available: https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/gpower.html) to help, rather than rely on textbook guidelines.
The key elements that govern what sort of sample size is needed are:
1. The types of scores/data you'll be using (e.g., is it a measurement or are these counts/frequencies, or rank values?)
2. Is it one outcome variable or multiple outcome variables that are under consideration (univariate or multivariate)?
3. Number of groups for the comparison (sounds like two from your description).
4. Type I and Type II risk levels you consider acceptable (your response suggests you've already given this some thought).
5. The minimum magnitude of effect/difference you would like your study to be capable of detecting, if it truly exists in the population. This is the target effect size, and plays a substantial role in the determination of appropriate sample size. Identifying a defensible ES may be the most challenging part of this process. Extant literature or decision-makers' consensus may help inform this determination if no other guidance is available.
Given your query, it seems to me that multiple regression or multivariate regression might be plausible methods for your analysis. In the absence of your specific research questions, though, that's just a guess at this moment.
I know you're aware of the limitations of cross-sectional studies, but you may not be in a position to implement a research design that would offer more protection against so-called "rival hypotheses."
Good luck with your work!
I agree with David Morse recommendation here. G-Power is a freely available program and relatively straight forward. I have my graduate students use the same program, which can be used for a variety of different research designs and statistical analyses to estimate appropriate sample size.
In terms of extracting proportions typically this is done from self-collected pilot data or from previously published literature.
Thank you all for the valuable suggestions.. I have not used G*Power yet, however, will make sure to check it out.
I had to mention the sample size formula in the protocol for my comparative cross sectional study. Later on, i ended up using the 2nd equation in my above-mentioned problem statement.. With a P1=.61 and P2=.39, power=90%, CI= 95% i ended up with a sample size of 103.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size