Recent Changes

Tuesday, May 24

  1. page 4th edited ... You will be graded on content included in your wiki, as well as organization. [[include compo…
    ...
    You will be graded on content included in your wiki, as well as organization.
    [[include component="pageList" hideInternal="true" tag="4jackson" limit="100"]]Jake
    To be considered parametric, a data set is expected to follow three assumptions under Analysis of Variance (ANOVA) . These assumptions are observations are independent, sample data has a normal distribution, and scores in different groups are to have homogeneous variances.
    Homogeneity of variance is the assumption that the variance of one variable is stable at all levels. If two categories of variables are used in research, the variables for each column should follow this constant variance. Data sets are expected to be constant and with this it is expected that variance also remains constant (Field, 2007). When a data set does not meet this expected variance, the population is no longer normal and should be used as a parameter for research. This would produce incorrect or biased results as a result of skewed data.
    When parametric samples are in violation of homogeneity of variance, the researcher has options to deal with the situational conundrum. The first is to simply ignore the possibility of a violation. A great percentage of data sets used in behavioral sciences are in violation of homogeneity as researchers often do not even review for homogeneity. If the researcher chooses this approach, the Monte Carlo Test performed by Glass etc all can be cited for support as it found that many parametric tests are not seriously affected by violations of assumptions. There is no set amount of violation that is clearly safe and none that is proof of erroneous data so judgment calls on the issue are difficult.
    The researcher may choose to use a nonparametric test instead of disregarding the homogeneity violation. By not using parametric assumptions, there is no risk of a type II error. Non parametric tests are criticized for a number of reasons. Naturally, all assumptions about the population parameters are unattainable. Also, the reported research will lack precision and potency. Nonparametric research is taxing on those that collect research as there are no software programs for assistance. Finally, nonparametric research is not even guaranteed immunity from type II errors. These issues make it appear that nonparametric tests are an inferior method of research, but other sources have found them as reliable as popular parametric tests with some tests even being more reliable.
    Sometimes, data found in violation of homogeneity of variance can be still be trusted because of the strength or “robustness” of the test because a robust tests’ validity is not affected by poorly structured data. There are several options for robust tests depending on the homogeneity violation within the data set. Welch’s Test can be used when there is a degree of inequality of the variances of two data populations. Welch’s Test allows the research to amend the set parameters to produce valid data. When multiple issues abound, trimmed means and winsorized variances are recommended. The strength of adding robustness is that it can provide a specific and individualized fixture based on the unique violation. Contradictorily, robust methods are sometimes questioned because certain researchers have found nonparametric measures are more powerful than trimmed down means. Also, robust methods sometimes ignore outliers when the existence of them is critical to the validity of the data parameters.
    Regardless of what choice is made in regards to overcoming the homogeneity of variation violation, there are problems with all ANOVA research due to small sample sizes. Statistically significant differences do not always equate to practical differences. Particularly with large sample populations, a huge statistical difference may not equate to a huge difference in practice.

    (view changes)
    12:19 pm

Thursday, October 29

  1. page 4th edited ... You will be graded on content included in your wiki, as well as organization. [[include compo…
    ...
    You will be graded on content included in your wiki, as well as organization.
    [[include component="pageList" hideInternal="true" tag="4jackson" limit="100"]]Jake
    To be considered parametric, a data set is expected to follow three assumptions under Analysis of Variance (ANOVA) . These assumptions are observations are independent, sample data has a normal distribution, and scores in different groups are to have homogeneous variances.
    Homogeneity of variance is the assumption that the variance of one variable is stable at all levels. If two categories of variables are used in research, the variables for each column should follow this constant variance. Data sets are expected to be constant and with this it is expected that variance also remains constant (Field, 2007). When a data set does not meet this expected variance, the population is no longer normal and should be used as a parameter for research. This would produce incorrect or biased results as a result of skewed data.
    When parametric samples are in violation of homogeneity of variance, the researcher has options to deal with the situational conundrum. The first is to simply ignore the possibility of a violation. A great percentage of data sets used in behavioral sciences are in violation of homogeneity as researchers often do not even review for homogeneity. If the researcher chooses this approach, the Monte Carlo Test performed by Glass etc all can be cited for support as it found that many parametric tests are not seriously affected by violations of assumptions. There is no set amount of violation that is clearly safe and none that is proof of erroneous data so judgment calls on the issue are difficult.
    The researcher may choose to use a nonparametric test instead of disregarding the homogeneity violation. By not using parametric assumptions, there is no risk of a type II error. Non parametric tests are criticized for a number of reasons. Naturally, all assumptions about the population parameters are unattainable. Also, the reported research will lack precision and potency. Nonparametric research is taxing on those that collect research as there are no software programs for assistance. Finally, nonparametric research is not even guaranteed immunity from type II errors. These issues make it appear that nonparametric tests are an inferior method of research, but other sources have found them as reliable as popular parametric tests with some tests even being more reliable.
    Sometimes, data found in violation of homogeneity of variance can be still be trusted because of the strength or “robustness” of the test because a robust tests’ validity is not affected by poorly structured data. There are several options for robust tests depending on the homogeneity violation within the data set. Welch’s Test can be used when there is a degree of inequality of the variances of two data populations. Welch’s Test allows the research to amend the set parameters to produce valid data. When multiple issues abound, trimmed means and winsorized variances are recommended. The strength of adding robustness is that it can provide a specific and individualized fixture based on the unique violation. Contradictorily, robust methods are sometimes questioned because certain researchers have found nonparametric measures are more powerful than trimmed down means. Also, robust methods sometimes ignore outliers when the existence of them is critical to the validity of the data parameters.
    Regardless of what choice is made in regards to overcoming the homogeneity of variation violation, there are problems with all ANOVA research due to small sample sizes. Statistically significant differences do not always equate to practical differences. Particularly with large sample populations, a huge statistical difference may not equate to a huge difference in practice.

    (view changes)
    10:58 am

Tuesday, June 9

  1. page 6Basseybookyrailer edited {http://r75.cooltext.com/rendered/cooltext120952269732697.png} {Mark of The Dragonfly.ppt}
    {http://r75.cooltext.com/rendered/cooltext120952269732697.png}
    {Mark of The Dragonfly.ppt}

    (view changes)
    9:47 am
  2. 9:41 am
  3. file crossover.ppt uploaded
    9:28 am
  4. page Woodspowerpoint edited {Crossover.pptm}
    {Crossover.pptm}
    (view changes)
    9:28 am
  5. page Sesipowerpoint edited {Crossover.pptm}
    {Crossover.pptm}
    (view changes)
    9:23 am

More