Quantitative research is driven by research questions and hypotheses. For every hypothesis

there is an unstated null hypothesis. The null hypothesis does not need to be explicitly

stated because it is always the opposite of the hypothesis. In order to demonstrate that

a hypothesis is likely true researchers need to compare it to the opposite situation. The

research hypothesis will be about some kind of relationship between variables. The null

hypothesis is the assertion that the variables being tested are not related and the results

are the product of random chance events. Remember that null is kind of like no so a null hypothesis

means there is no relationship. For example, if a researcher asks the question

"Does having class for 12 hours in one day lead to nursing student burnout?"

The hypothesis would indicate the researcher's best guess of the results: "A 12 hour day

of classes causes nursing students to burn out."

Therefore the null hypothesis would be that "12 hours of class in one day has nothing

to do with student burnout." The only way of backing up a hypothesis is

to refute the null hypothesis. Instead of trying to prove the hypothesis that 12 hours

of class causes burnout the researcher must show that the null hypothesis is likely to

be wrong. This rule means assuming that there is not relationship until there is evidence

to the contrary. In every study there is a chance for error.

There are two major types of error in quantitative research -- type 1 and 2. Logically, since

they are defined as errors, both types of error focus on mistakes the researcher may

make. Sometimes talking about type 1 and type 2 errors can be mentally tricky because it

seems like you are talking in double and even triple negatives. It is because both type

1 and 2 errors are defined according to the researcher's decision regarding the null hypothesis,

which assumes no relationship among variables. Instead of remembering the entire definition

of each type of error just remember which type has to do with rejecting and which one

is about accepting the null hypothesis. A type I error occurs when the researcher

mistakenly rejects the null hypothesis. If the null hypothesis is rejected it means that

the researcher has found a relationship among variables. So a type I error happens when

there is no relationship but the researcher finds one.

A type II error is the opposite. A type II error occurs when the researcher mistakenly

accepts the null hypothesis. If the null hypothesis is accepted it means that the researcher has

not found a relationship among variables. So a type II error happens when there is a

relationship but the researcher does not find it.

To remember the difference between these errors think about a stubborn person. Remember that

your first instinct as a researcher may be to reject the null hypothesis because you

want your prediction of an existing relationship to be correct. If you decide that your hypothesis

is right when you are actually wrong a type I error has occurred.

A type II error happens when you decide your prediction is wrong when you are actually

right. It is much harder to convince a stubborn person that they are wrong, therefore it is

the number 2 possibility. One way to help you remember the meaning of

type 1 and 2 error is to find an example or analogy that helps you remember. As a nurse

you may identify most with the idea of thinking about medical tests. A lot of teachers use

the analogy of a court room when explaining type 1 and 2 errors. I thought students may

appreciate our example study analogy regarding class schedules.

It is impossible to know for sure when an error occurs, but researchers can control

the likelihood of making an error in statistical decision making. The likelihood of making

an error is related to statistical considerations that are used to determine the needed sample

size for a study. When determining a sample size researchers

need to consider the desired Power, expected Effect Size and the acceptable Significance

level. Power is the probability that the researcher

will make a correct decision to reject the null hypothesis when it is in reality false,

therefore, avoiding a type II error. It refers to the probability that your test will find

a statistically significant difference when such a difference actually exists. Another

way to think about it is the ability of a test to detect an effect if the effect really

exists. The more power a study has the lower the risk

of a type II error is. If power is low the risk of a type II error is high. Logically,

power increases as the sample size increases since you have collected more information,

which makes it easier to correctly reject a null hypothesis. Other factors including

the effect size and significance level also influence power.

Usually power is set at 0.8 or greater before a study begins, meaning that you should have

an 80% or greater chance of finding a statistically significant difference when there is one.

This value is used to calculate the needed sample size for a study before it begins.

A statistically significant result does not mean that there is an important or meaningful

difference in the influence of one variable on another variable. It simply indicates that

the researcher can be confident a difference exists. Particularly with large sample sizes

even very small differences can be statistically significant.

To determine how meaningful or important the difference is the effect size needs to be

calculated. Effect sizes are a standardized measure of how large the influence of one

variable is on another variable. Another way to look at it is the degree to which the null

hypothesis is false. Before a study begins an effect size (either

known or estimated) is used as a part of the power analysis in order to determine how large

of a sample is needed to achieve a power ≥0.80. There are a variety of ways to calculate the

effect size, which is measured in standard deviations. When interpreting or critiquing

research the guidelines developed by Cohen are often used. However, keep in mind that

these are simply guidelines. Effect sizes need to be interpreted within the context

of the research. Statistical significance is used to determine

how likely it is that the results of a study are due to chance. Before a study begins the

alpha level is set at a value that represents the error rate that a researcher is willing

to accept. Typically 0.05 is used, which means that if the null hypothesis is true it would

only be rejected in 5 out of 100 trials. If a smaller risk of rejecting the null hypothesis

is needed an alpha of 0.01 may also be used. You might wonder why we don't always use the

lowest possible alpha value. It is because type 1 and type 2 errors are inversely related.

If you decrease one you increase the other. People consider a type 1 error more serious

which is why 0.05 is the minimum conventionally accepted level in most disciplines.

Researchers need to accept a certain level of chance that they may be wrong. Before a

study begins a balancing act occurs where researchers decide what values are appropriate

for the study. Don't confuse alpha with p. Alpha levels are

determined before the study begins. P values are calculated from sample data after the

study has been completed. If the P value is less than the alpha value the null hypothesis

is rejected and the results are statistically significant.

Just because a study has statistically significant results does not mean the results will have

practical significance. Statistically significant differences can always be found even for very

small differences if the sample size is large enough. Practical significance relates to

how relevant the findings are to the question being asked.

For example, a study may find that there is a statistically significant difference in

the lifespan of people who take a new drug compared to the old one. The statistically

significant difference may be only a few hours and come with very severe side effects. Most

people would say that the improvement in lifespan is not practically significant.

Make sure you consider if the differences are big enough to have real meaning.

It may help to go back and consider the effect size, among other things, to help you answer

this question. A power analysis is most often used to calculate

what sample size is needed. However, it can be used to calculate sample size, effect size,

significance level or power if you have three of the four values.

Since alpha is usually 0.05 and power is usually 0.8 researchers need to pay the most attention

to the effect size in order to calculate the needed sample size.

Thank you for watching. If you are looking for more information you

are welcome to check out my eBook, website and related videos. Links are provided in

the description.