In scientific research, there are several ways that measurements are made or assessments are completed. These *include direct observations* of things such as how much water is coming out of an irrigator after watering a plant, counting how many cars pass by during a given amount of time, and asking someone to tell you their height.

These indirect measures are then compared to known standards to determine the size of people and plants. By doing this for many individuals, we can come up with average sizes. These averages are what matter when it comes to **defining healthy weights** and dimensions for people and plants.

There are *also mathematical equations used* to calculate these numbers depending on the context. For example, when calculating the average weight of all humans, you must not count skin folds or muscles because they vary greatly between individuals!

This article will go into more detail about each one of these methods. Who knows, you **may find something new** you could add to your repertoire.

## Statistical significance of measurements

The **second major factor** that researchers look into is how much statistical “significance” their results have. This is done by using either a t-test or an analysis of variance (ANOVA).

A t-test looks at whether one set of numbers are bigger than another. In our example, this would be whether the length of a dog’s tail is related to its breed. For instance, if a long tail was bred with a short tail, then the hypothesis must be rejected because there is no relation between the two variables!

An ANOVA does something similar to a t-test but instead of just having one variable, it has many different variables so we can see how they interact with each other. For instance, studying the effect of nutrition on **growth requires looking** at several factors such as age, breed, and weight.

Both *tests require us* to make assumptions about what will happen in the future. By assuming these things, we can use the data from past experiments to come up with conclusions about what will happen now.

## Measurement accuracy

The more times something is done, the better our estimates of how well it works will be! This process is referred to as replication or repeatability.

When doing scientific research, there are several things that have to come together for results to be considered valid. One of these is measurement accuracy.

Measurement accuracy means determining whether what was done was actually done. For example, if you test drug A on patient B and it helps B’s symptoms, then your body does not know that drug A helped until you do not get drug A anymore!

This goes beyond just **taking drugs away**– **sometimes patients need surgery**, so their doctors prescribe medication after surgery to help with healing. If part of the **surgery involved using medicine** to treat pain, they may stop the medicine before the patient recovers fully.

By measuring the effects of the medicine directly, we can determine whether the two had an effect on each other.

## Interpreting measurement results

The word measure often has **two different uses within scientific research**. When referring to length or area, it usually means to count how *many times something* is done (for example, what size of brush you need for your painting) or how much space there is for something (how big an apartment we have).

When talking about numbers that describe intensity or quantity, like how much sugar you eat or how much blood is present in your body, the term measurement does not mean one thing after another, but rather all at once. These are referred to as quantitative measurements.

That being said, *qualitative researchers make use* of the term when they are looking into topics such as psychology or sociology. Here, individuals’ responses to questions can be used to get insights into their beliefs, emotions, and behaviors. Therefore, instead of counting how *many times someone says yes* or no to a question, these people may be asked to give more descriptive answers.

These answers are then analyzed and categorized to gain insights about who they are and why things happen.

## Sample size

A crucial part of any scientific experiment is determining how many people will be involved in it, as well as what measurements will be made during the experiment.

A very important part of this process is deciding how large your *sample group needs* to be to get an accurate result. If you have too small of a sample size, then you may not **receive enough data** to make conclusions about the whole population!

By having a **large enough sample size**, however, you are more likely to get useful information.

If you run out of money or time later, you can always try again. It’s impossible to do **science without spending resources**, so being able to start over makes doing research much easier.

## Statistical tests

A statistical test is a way to determine whether there is enough evidence to prove an assertion or claim. In scientific research, researchers will use a statistical test to evaluate their hypotheses or theories.

For example, let’s say that you want to know how **well canned peas preserve weight** during storage. You *could go around asking people* what they think about this theory, but that would not be very reliable.

A better approach is to perform an experiment! You would have one group of individuals store a batch of peas for a week and then weigh them to see if it makes a difference. If it does, great! You proved your hypothesis.

But what if the opposite happens? What if the dried beans lose more weight than fresh ones? That would disprove your theory!

So, you take a *second sample set* and repeat the same process with those. Only this time, you add water to the dry peas before storing them.

After the seven days, you can again compare the weights and see which one has less weight. The one with no water probably lost some of its **nutritional value due** to exposure to air, so we should ignore it when calculating the effects of dehydration.

The samples that had water retained some of their nutrition because they were still relatively fresh. We can assume then, that having water preserves the nutritional content longer than just leaving them alone.

## Effect size

One of the most important concepts to know when interpreting research studies is effect size. This term refers to how much an experiment’s findings influenced the outcome.

A small effect size means that the results were not very significant. For example, *testing whether adding sugar* to water changes the color of the solution. Or *studying whether using sunscreen every day prevents sunburn*.

These types of experiments have low effect sizes because they are so **trivial — people usually** do things like add sugar to drinks or use sunblock anyway, so the impact is minimal. Therefore, their influence is also limited.

In other words, it makes little difference if you add sugar to water or not!

Interpretation of study findings depends heavily on what you believe about related topics. If you think that chocolate is delicious, then reading about an interesting new study that *claims drinking green tea* can keep your heart healthy will probably inspire you to start doing it.

Conversely, if you believed that red wine is best for health, this article would be frustrating for you.

That’s why having accurate information is so essential- so you can form your beliefs accordingly. The more knowledge you have, the better able you are to decide what conclusions to draw from what studies.

## Publication bias

A major cause of research biases is publication bias, or the tendency for researchers to publish only studies that show their results being better than those of other groups.

This can be due to many things- maybe they are working on similar projects as the competitor, so they choose not to report their findings because it would make them look bad.

Or perhaps there’s just an internal push to focus more on success rather than failure, which then skews what gets published. Or maybe the researcher doesn’t want to share his or her methods since they work but for others who don’t yet.

Any of these reasons are totally acceptable, as long as you are honest with yourself and your colleagues! But if you **ever find something** that seems too good to be true, it probably is.

Be wary of claims that seem too good to be true. If someone calls you out on it, try to do some additional investigations – does this person make a lot of such claims? Is this person paid to say that? What **else could possibly explain** this result?

By asking questions, you will at least know where you stand and how to defend yourself should accusations arise. You also showed respect for the study by doing your own investigation, which **usually creates trust**.

Publication bias is a huge source of systematic error in health research, and must be addressed to *get accurate answers* from studies.

## Confidence intervals

A confidence interval is an important way to determine if something is true or not. It comes from probability theory, where we define what a probability is and how to calculate it.

A probability is defined as the likelihood of something happening. For example, the chance that it will rain this week is called the probability of precipitation (or P~precip for short). The likelihood of something occurring is determined by looking at *past events*, which are referred to as instances. So, the chance of rainfall this week is *calculated using past instances* of precipitation.

The proportion of time that things occur is called frequency. With probabilities, you can also find the mean or average value of a variable. This is done with the formula mean = **instance × frequency**. In other words, the mean is just taking the number of **instances times** the frequency.

With the mean, you can then take the difference between each instance and add them together to get a total. Then divide the total by the frequency to obtain the mean ratio or proportion.