Lab 4 Flashcards
the purpose of the Mean
- replication in experimental design design generates robust data sets.
–> multiple measurements for the same manipulation/ treatment generated. - allows for interpretation regarding variability and potentially the relationship between the IV and DV.
- Trying to communicate or present replications separately is ineffective.
- the mean [average] value of these replicated measures can effectively graph and provide clear communication of the experimental results.
what does Standard deviation determine?
What can be made using standard deviation?
where does variation occur?
- determines what the average degree of variation from the mean is for the replicated measures
- Once mean values have been graphed as data points, specific standard deviation value for each set of replicated measures, can be applied to the related data point as an error bar.
- Within a set of replicated measures for the same manipulation/ treatment, there is variation associated with either protocol performance, equipment, or biological uniqueness.
what does an Error bar represent?
- the standard deviation error bar visually represents the variation of the replicated measures around the mean value.
what a The Scientific Hypothesis?
what do the statements represent?
- is a statement which defines the independent and dependent variables, and the potential relationship between them.
- the statement represents the observations, knowledge gathering and formulated speculation of what will happen when the independent variable is manipulated, and the dependent variable measured.
what is a The Null Hypothesis?
what is its condition?
what does the condition indicate?
- cannot use the data to prove the scientific hypothesis as true. Instead, develop the scientific hypothesis to be falsifiable.
- there needs to be a condition in which the manipulation of the independent variable does not result in measured change from the dependent variable
- thus, an indication that there is no relationship between the variables – this condition defines the null hypothesis.
what is Inferential Statistics?
what is required for this?
- Statistics is the statistical approach to drawing conclusions or making inferences about what the data means.
- In order to make these conclusions/ inferences it is the null hypothesis that must be tested.
- To do this, the data is analyzed through a series of mathematical equations that result in a test statistic and, a p- value.
what is a p- value?
What does it mean if the p value is higher than the alpha/significance level?
What does it mean if its lower?
- represents the probability of the observed experimental outcome.
- test statistic use data when the null hypothesis is considered true.
- if the p- value of the observed experimental outcome is higher than the defined significance/alpha level [the accepted probability of making an erroneous conclusion], then the data is more likely to represent the condition of the null hypothesis.
- If the p- value of the observed experimental outcome is lower than the alpha/significance level, then the data is less likely to represent the condition of the null hypothesis and therefore, the scientific hypothesis is supported.
what are 95% Confidence Intervals?
what do they represent?
what do their ranges suggest?
- 95% CI error bars can be used to estimate the p- value by the degree of overlap of the error bars when comparing means.
- confidence intervals represent the range in which the ‘true means’ can be found.
- Therefore, if there is a degree of overlap between these ranges, then there is an increasing likelihood that the two means being compared are from the ‘same population’.
- This supports the conclusion that the means are not different, which is the expected experimental outcome when the condition of the null hypothesis is true.
Incorporating Inferential Statistics to Experimental Design and Analysis
Build background knowledge
—>
Define a testable scientific hypothesis
—>
Define null hypothesis
—>
Choose appropriate hypothesis test based on design and set the desired level of significance
—>
Set desired level of significance [alpha]
—>
Perform experiment, collect data, simplify with descriptive stats
—>
Perform the hypothesis test, obtain a p-value
—>
Compare p-value to alpha to conclude support or argument towards the null hypothesis
2 ways it can go from here
Null hypothesis is supported
—>
p-value is above the pre-set alpha = data is more likely to be observed under the conditions of the null hypothesis
—>
The null hypothesis is supported and the scientific hypothesis may not be worth further pursuit.
Null hypothesis is not supported
—>
p-value is below the pre-set alpha = data is less likely to be observed under the conditions of the null hypothesis
—>
The null hypothesis is argued against, and the scientific hypothesis may be worth further pursuit.
Incorporating Inferential Statistics to Experimental Design and Analysis
And
Questions to consider:
— Is it possible the p-value was not accurate due to a large range of equipment or technique error?
— Is it possible the p-value was not accurate because there were too few biological replicates?
— Even if the data appears to support the scientific/research hypothesis, is it well supported by the field literature?
— Has the experimental design made considerations for the larger system?
— What are the limitations of the experimental [model] system that was used?
HORIZONTAL DATA ENTRY
Using HORIZONTAL data entry is effective when you have either a single row of measured values below the various -axis labels [independent variable manipulations] or many rows of replicate measures as you would in a large data set.
VERTICAL DATA ENTRY
Using VERTICAL data entry is effective for organizing replicates and allowing you to calculate the MEAN at the bottom to have ready for graphing.
Ideally, the most efficient way to organize data, is to set it up for descriptive analvsis [calculating MEAN] and also for graphing at the same time; typically a table format where the replicates run vertically and the treatments run horizontally.
Certain graph types will require you to enter the data only one way. If you enter the data in one of the formats and it does not generate the appropriate graph, then try using the other entry format.
