Scribd Logo
Upload

Welcome to Scribd!

  • 34 views

    BioEpi Lab Module 9

    Uploaded by

    Noriel Lacsina
    Power analysis is conducted before data collection to determine the minimum sample size needed to detect an effect with adequate statistical power. It helps avoid type II errors and accounts for factors like significance level, effect size, and statistical test used. Software like G*Power can compute power and required sample size given these parameters. Power analysis is important for study design and balancing statistical accuracy with practical constraints.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    BioEpi Lab Module 9

    Uploaded by

    Noriel Lacsina
    34 views2 pages
    Power analysis is conducted before data collection to determine the minimum sample size needed to detect an effect with adequate statistical power. It helps avoid type II errors and accounts for factors like significance level, effect size, and statistical test used. Software like G*Power can compute power and required sample size given these parameters. Power analysis is important for study design and balancing statistical accuracy with practical constraints.

    Original Title

    BioEpi-Lab-Module-9

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Power analysis is conducted before data collection to determine the minimum sample size needed to detect an effect with adequate statistical power. It helps avoid type II errors and accounts for factors like significance level, effect size, and statistical test used. Software like G*Power can compute power and required sample size given these parameters. Power analysis is important for study design and balancing statistical accuracy with practical constraints.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    34 views2 pages

    BioEpi Lab Module 9

    Uploaded by

    Noriel Lacsina
    Power analysis is conducted before data collection to determine the minimum sample size needed to detect an effect with adequate statistical power. It helps avoid type II errors and accounts for factors like significance level, effect size, and statistical test used. Software like G*Power can compute power and required sample size given these parameters. Power analysis is important for study design and balancing statistical accuracy with practical constraints.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    You are on page 1of 2

    Biostatistics and Epidemiology (LAB)

    Module 9  Probability that the researcher will make a


    POWER ANALYSIS correct decision to reject the null hypothesis
    (Ho) when it is in reality, False.
    POWER ANALYSIS  Probability of avoiding Type II error.
     Is directly related for the test for hypothesis  This is what we actually solve for the Power
    (t-test and ANOVA) and is usually Analysis.
    conducted before the data collection.
     REMEMBER: We are always trying to void
    o Hypothesis is usually done after the
    Type II error
    collection or after the treatment of the
     In computing for statistical power, we are
    data.
    encountering 2 types of error: Type I and II
     A critically important aspect of study design
    is determining the appropriate sample size ERRORS IN ESTIMATING SAMPLE SIZE
    to answer the research question. TYPE I ERROR TYPE II ERROR
     Suggested to be Done before data  Refers to the  Probability we do
    collection situation where not reject Ho when
    *Note: not always we incorrectly it is False.
     Main Purpose: Determine smallest sample reject Ho when in  “False Negative
    size that is suitable to detect the effect of a fact, True. Result”
    given test at the desired level of  “False Positive
    significance. Result”
     The smallest sample size that could be best
    effect for what you are trying in your study. The power of your study is the probability that you
     The power of your study is the probability will find a significant difference or a relationship if a
    that you will find a significant difference or difference/relationship truly exist in the population.
    relationship if a difference or relationship
    truly exists in the population. Even if your null hypothesis is indeed false, if your
    study is underpowered, you will not find significant
    WHY DETERMINE SAMPLE SIZE? results. In other words, you will have a False
    (Title of the study) A study is proposed to evaluate Negative result which is also known as Type II
    a new screening test for down syndrome. error.
     For evaluation, pregnant women will be asked
    to provide a blood sample and undergo Some factors that will affect the power of your
    Amniocentesis (Gold standard in comparing study includes sample size, significance level,
    results). effect size, and the type of statistical analysis you
    *Amniocentesis is the method of obtaining plan to conduct.
    samples through the use of the umbilical cord.
     Blood Sample Test: $250.00 The best way to increase your study’s power is to
     Amniocentesis: $900.00 increase your sample size. The (↑) higher the
     $1150/respondent sample size = the (↑) higher the statistical
    power. However, risk in increasing your sample
    Is it still beneficial? Knowing you just want to size beyond what your power analysis indicates will
    evaluate something and then you will need to increase the risk of heading a False Positive
    spend $1150/respondent? What if you have 30 Result. So, (↑) higher sample size could cause a
    respondents? It is very expensive. False Positive Result. In this case, our False
    Positive Result is the Type I error.
    How do we try to minimize the financial constraint
    of the study? That is through the use of Power Luckily, if you plan ahead, you can achieve
    Analysis. You will try to determine the desired/ sufficient power for your study. So, it is suggested
    suggested sample size that is applicable foe the that you try to compute for the sample size first
    study given a level of significance. And through before you actually proceed with collecting data.
    this process, you could defend the financial Always plan your study ahead.
    constraints of your study. Because financial
    constraints alone might substantially limit the Increase in sample size would benefit the statistical
    number of women who can be enrolled in the power because it would increase the power of your
    research. Thus, it is important to consider both study. However, it will give you a risk for the Type I
    statistical and tactical issues when interpreting error (False Positive).
    the statistical analysis.
    Given is the diagram of the possible decisions that
    When you are conducting a research study, even you will make in power analysis, as well as the
    though your research has a very big potential or corresponding types of error you might encounter.
    has very big margin in the future with regards to
    advancement technology, you should also look
    into the practicality of your study.

    STATISTICAL POWER

    1
    In this case, you will reject your null hypothesis, You would just input all of the values and then you
    where in fact, your null hypothesis is false, and that will get the desired power/sample size that you
    is a correct decision. need that has adequate power to prove if there is a
    What if you reject your null hypothesis, however, significance in your study.
    your null hypothesis is type II, that is what you call POWER ANALYSIS
    your Type I error, which is a false positive. If you  Used to calculate the minimum sample size
    are falsely accepting your null hypothesis, that is needed to produce a reasonable level of
    Type I error (aka False Positive). You rejected a accuracy.
    true hypothesis.  2 types:
    Priori Analysis Post Hoc Analysis
    What if you accept the null hypothesis, where in  Part of research  Done after a study
    fact, your null hypothesis is true? So, that is a learning process  Determine the
    correct decision. On the other hand, what if you  Done before the achieved level of
    accept the null hypothesis, where in fact, the null research study power given the
    hypothesis is false? That is what we call Type II  Determine the sample size that
    error. required sample was actually used
    size that has in the study
    Other term for Type I error is alpha and for Type II adequate power to  Assist in explaining
    is beta. prove the any potential non-
    significance of the significant results
    Statistical Power is described as: study.
     Most common type

    *Power analysis is not only done before the study.


     As power increases, there is a lower risk of It can be done before (Priori Analysis) but it can
    Type II error (false negative). However, the also be done after (Post Hoc Analysis).
    higher the sample size = the higher the
    statistical power = the higher the chance for Both types of analysis are trying to explain/
    Type I Error. determine if there is actual power in your study. In
    Priori, it will give you power through the use of
     Ideally, value should be at least 0.80 to
    giving you a sample size that you will use for the
    detect a reasonable departure from the null
    study. On the other hand, Post Hoc will show you
    hypothesis. (0.80 is the suggested value of
    power after the study through the use of sample
    the statistical power).
    size you used.
    G POWER SOFTWARE
    DETERMINANTS OF POWER
     An example of the software you could use
    The power of any test of statistical significance will
    for computing the statistical power.
    be affected by four main parameters:
     Automatic generator for statistical power. 1. Effect size
    2. Sample size (N)
    3. Alpha significance criterion, α
    4. Statistical power, (or the chosen or implied
    beta, β)

    You might also like