Sign In | Subscribe
Start learning today, and be successful in your academic & professional career. Start Today!
Loading video...
This is a quick preview of the lesson. For full access, please Log In or Sign up.
For more information, please see full course syllabus of Statistics
  • Discussion

  • Download Lecture Slides

  • Table of Contents

  • Transcription

  • Related Books

Bookmark and Share

Start Learning Now

Our free lessons will get you started (Adobe Flash® required).
Get immediate access to our entire library.

Sign up for Educator.com

Membership Overview

  • Unlimited access to our entire library of courses.
  • Search and jump to exactly what you want to learn.
  • *Ask questions and get answers from the community and our teachers!
  • Practice questions with step-by-step solutions.
  • Download lesson files for programming and software training practice.
  • Track your course viewing progress.
  • Download lecture slides for taking notes.
  • Learn at your own pace... anytime, anywhere!

Research Design

Lecture Slides are screen-captured images of important points in the lecture. Students can download and print out these lecture slide images to do practice problems as well as take notes while watching the lecture.

  • Intro 0:00
  • Roadmap 0:06
    • Roadmap
  • Descriptive vs. Inferential Statistics 0:51
    • Descriptive Statistics: Data Exploration
    • Inferential Statistics
  • Variables and Relationships 1:44
    • Variables
    • Relationships
  • Not Every Type of Study is an Experiment… 4:16
    • Category I - Descriptive Study
    • Category II - Correlational Study
    • Category III - Experimental, Quasi-experimental, Non-experimental
  • Category III 7:42
    • Experimental, Quasi-experimental, and Non-experimental
  • Why CAN'T the Other Strategies Determine Causation? 10:18
    • Third-variable Problem
    • Directionality Problem
  • What Makes Experiments Special? 17:54
    • Manipulation
    • Control (and Comparison)
  • Methods of Control 26:38
    • Holding Constant
    • Matching
    • Random Assignment
  • Experiment Terminology 34:09
    • 'true' Experiment vs. Study
    • Independent Variable (IV)
    • Dependent Variable (DV)
    • Factors
    • Treatment Conditions
    • Levels
    • Confounds or Extraneous Variables
  • Blind 38:38
    • Blind Experiments
    • Double-blind Experiments
  • How Categories Relate to Statistics 41:35
    • Category I - Descriptive Study
    • Category II - Correlational Study
    • Category III - Experimental, Quasi-experimental, Non-experimental
  • Example 1: Research Design 43:50
  • Example 2: Research Design 47:37
  • Example 3: Research Design 50:12
  • Example 4: Research Design 52:00

Transcription: Research Design

Welcome to www.educator.com.0000

Today we are going to be talking about research designs, also known as research strategies 0002

Here is the roadmap for today.0007

First we have been talking about descriptive versus inferential statistics. 0010

You now that now were talking about different ways of doing inferential statistics so we are just going to review this again.0015

Then we are going to talk about variables in relationships.0021

There are different kinds of variables and different kinds of relationships you need to know about.0025

We are going to introduce three different categories of research strategies or designs 0029

and then we are talking about why cannot other strategies besides the experimental strategy determine causation.0035

We are going to talk about what is so special about experiments. 0042

Finally, we are going to end with some experimental terminology. 0046

Now that we have been doing a lot of descriptive statistics so far looking at visualizing data, summarizing it, 0050

and we could do that for both one and two variables.0059

Then we talked about why inferential statistics is different.0062

Number 1, we are interested in generalization and because of that we covered some sampling methods that allow our samples to generalize to the population. 0066

We also introduced the idea that we would like to have comparison in order to examine causation.0077

We would like to know whether one variable causes another not just the relationship between them in terms of a regression line 0085

but we like to know does one cause the other one.0093

In order to do that, we are going to have to learn about experiments. 0098

But before we learn about experiments, it is going to be helpful to learn about other kinds of research strategies in addition to experiments, 0103

just so we could see why experiments are so special and different.0112

but in order to understand those different kinds of research strategies we are going to need to review variables and the relationship that they can have.0117

One thing you need to know about variables is that we have covered 2 broad kinds of variables so far.0125

Categorical, categories as the variable levels instead of actual numerical values or continuous 0132

where the values of the variables are numbers that are meaningful in some way.0144

An example of categorical variables are whether kids have pets or not, versus continuous which might be something like a score on a biology test.0149

Then you can have maybe 1 – 100 and each of those numbers are meaningful.0163

In terms of relationships, there is a bunch of different relationships we actually covered.0169

All of those relationships so far have been descriptive relationships.0175

We have covered relationship such as linear relationship, curvilinear, we have also looked 0180

at relationships such as negative correlation and positive correlation. 0188

But all of those relationships are descriptive.0199

They are just describing what the relationship is like.0204

They are not actually telling you whether one variable causes the other variable.0208

Those are causal relationships.0213

They are not necessarily telling you that one variable does not cause the other one either.0215

This is a new kind of relationship that we are going to be looking at.0219

Only one kind of study, what kind of research design can actually look at causal relationships.0224

And that is going to be through the experimental research strategy or design.0234

This is the only one.0245

That one is pretty special. 0248

So far we have only covered these other kinds of relationships. 0251

Before we go on to the nitty-gritty of an experiment, it helps to know what are the non-experiment. 0255

Not every type of research study is an experiment just because you collect data and some variables does not mean you have an experiment.0266

That is a very meaning to start using a mere context. 0276

A lot of people especially in the media or popular press they only use it in a more loose way before they call it an experiment.0280

Usually we use that for a very specific kind of research strategy.0288

First, category 1 are the kind of studies that are only descriptive.0293

These are descriptive studies.0302

A variable can either be categorical or continuous but they do not look at any relationships.0304

No relationships.0314

They do not even describe relationship between variables. 0316

They only look at variables in isolation. 0321

They are really only interested in the distribution of peoples heights, I.Q, or the distribution of a bacteria in the stomach.0323

They are only interested in describing one variable or a couple of different variables but not looking at the relationships between them. 0337

That is very important. 0345

That is the descriptive study.0347

That is category one. Category two moves on a bit. 0349

Here we are interested only in continuous variable. 0350

Only continuous variables can be used in category 2, no categorical ones.0357

Here the only relationship we can look at is the descriptive. 0362

Here because we have multiple continuous variables multiple, we often look at these on a scatter plot.0367

Those are called correlational studies and that is category two.0381

Category 3 has three different kinds of research strategies in their.0393

One is experimental, the other is quarter experimental, and the third kind is non experimental.0399

In these, you need to have at least one categorical variable. 0406

You can have also continuous ones and other categorical variables but you need to have at least one categorical variable.0416

Here you can look at both descriptive relationships, both for one of these the experimental one, you could also look at causality.0426

You cannot look at causality with cause experimental, not quite, not all the way, nor non experimental.0438

Even though you can look at causality within a category 3 research strategies it is really only experimental research strategy that can look at causality. 0449

In category 3, let us break down these three different types.0465

Experimental research strategy you will know that you need at least one categorical and that is actually the case for all of these at least one categorical.0469

That is what these three have in common but for experimental, you can look at both the descriptive relationship and causal relationships.0493

You can look at both of those kinds of things.0512

For quasi experimental, we could look at descriptive and we can come up with causal explanation, but not quite.0515

And that is what called quasi experimental.0521

It gets close to get pretty close as closer you can be without actually being experimental and for non experimental you can only look at descriptive.0523

These are the three different kinds and quasi experimental ones are also often called accidental experiments 0549

or one very common example is using pretest and post tests where you are looking at how something in the middle is causing you, 0562

but you also have compound where because your testing people two times or your measuring something twice.0577

It might just be an artifact of measuring something twice. 0586

For example, if you wanted to know how good a math software is, you might have a pretest 0590

and post test but it might just get better just because they did the same test twice.0596

That is why quasi experimental sometimes has issues.0601

Non experimental variables are going to be quite different from experimental, categorical variable.0606

We are going to talk about next what the difference is.0614

Okay, actually before we do that we are going to talk about why cannot the other strategies determine causation.0617

It is because it seems exclusive for experimental research strategy.0626

What is wrong with the other one?0631

All of the other strategies have 2 fundamental problems that they cannot escape.0633

Here is the big issue. 0641

We have some variable one and we want to know whether they will want causes changes in variable 2.0643

For instance, let us say we want to know if having a radio in developing countries will lead to having women having fewer children.0656

If your household has a radio does it mean you have fewer children in your house or greater use of birth control?0671

That might be variable 1 radio and available 2 might be something like use of birth control.0679

We want to know whether one variable causes the other variable.0691

Because if we knew about causation and we want to reduce or increase the use of birth control somewhere then we could adjust radios in households.0699

We could give people free radios or something, and then by doing that change the use of birth control.0721

It would be nice to know whether this causal relationship actually exists or not.0727

Now, one thing you can do is to ask people whether they have radios and then ask people how much they use birth control. 0730

Whenever you do studies like that and that would largely be a non experimental design because this is a categorical variable radio, no radio.0737

Maybe use of birth control might be how many times do you use birth control in a month and maybe that is continuous.0747

Here one issue is you do not know whether one causes the other because we have what is called a third variable problem.0755

A third variable problem can be described like this.0766

There might be another variable that actually impacts both your variable 1 and 2.0770

This is what we call a extraneous variable.0776

One that is sort of outside your scope of study.0779

You are not looking at it in your study. 0783

This extraneous variable might actually impact both variable 1 and variable 2.0787

You might see variable 1 and 2 having a relationship together like when one goes up, the other goes up 0794

or the other one goes down and the other one goes down.0802

You might see that relationship.0804

That relationship might actually be caused by a third variable that is affecting both of those.0806

For instance, perhaps how much money your household earns affects whether you have a radio but affects the use of birth control.0814

Or the level of education that a woman has might affect whether they own a radio and whether she uses a birth control.0823

Those things might be caused by something else.0831

They look like they are moving together but they are actually not causing each other.0834

One sort of picture that I like to draw from my students, sort of a mental picture is this idea of let us say there are 2 boats.0839

There is a little blue boat and there is little red boat.0851

And you whenever you see the blue boat sort of rise up in sea level, the red boat rises up in sea level.0866

When the blue boat goes down in sea level, the red boat goes down in sea level but actually both of those are caused by the water.0875

It is not that the blue boat is actually causing the red boat to go up or down but it is really the water 0888

that is causing both the blue boat and red boat to go up and down.0893

In this case, this might be variable 1, variable 2, and the water is the extraneous variable.0897

It might really be this extraneous variable that you are not studying for some reason that is really causing the both of these things.0906

That is always the problem in a non experimental correlational designs.0915

They always have this problem of the third variable that might be lurking back there.0925

The third variable is often also called a confound.0931

Something that is moving along with variable 2, the variable of interest, but it is sneaking in there.0935

We are not measuring it, so we do not know about it.0946

Even when you solve this third variable problem, sometimes you still have another problem that remains and this is called the directionality problem. 0951

We do not actually know whether variable 1 impacts variable 2 or is it that variable 2 impacts variable 1. 0961

We do not know how to distinguish between those and that is called the problem of directionality.0974

Even if two variables have a relationship to each other either a positive correlation or negative correlation, we do not know whether X causes Y or Y causes X.0979

That is hard to distinguish in a correlational design or non experimental design. 0991

Let us say we take a correlational picture.0999

For instance, here, let us say that variable 1 is a woman's educational attainment and variable 2 let us say is use of birth control.1003

Well, we do not know whether a woman's education actually impacts the use of birth control 1028

or perhaps it is that using birth control allows women to stay in school longer.1034

It could be either one of those directions that we do not know from doing a correlational study.1041

Let us say we ask women how many years of schooling do you have and we ask women how often in a month do you use birth control?1046

That would be a correlational design, but one issue that we will have is we do not know whether X causes Y or Y causes x.1055

On top of that right now we also do not know whether there is a third variable lurking that might be more explanatory. 1064

Experiments have found a way to solve both of these problems based on the third variable problem and the directionality problem.1076

That is why experiments are special.1085

Here is how experiments have solve the problem.1087

They are special because they use manipulation and what we call control.1090

Manipulation and control these two solve the third variable and directionality problem.1097

Here is what we mean by manipulation.1105

Manipulation means that in an experiment a categorical variable, the variation is created by experimenter or researcher team.1107

It is not that we just ask women if they own a radio, it is that the experiments has actually go in there and create that variation.1133

They might take a whole bunch of people, none of whom have radios, they are all the same at first 1141

but then the experimenters create variation by randomly assigning half of them to get a free radio.1146

In that way, the experimenters are the one who is creating.1155

They are the ones walking around in there and that is manipulation.1160

To the categorical variable gets manipulated.1165

Whenever you think manipulation I want you to think create it, like rolling up your sleeves and get in there making them a certain way.1169

That is one thing that experiments do differently.1178

How does these address the third variable problem?1181

Well, it addresses the third variable problem because the experimenters created some condition right. 1188

They created the radio condition.1196

Then once the experimenters introduced this then you can look for changes in the second variable. 1198

Now the experimenters did not do anything to that second variable.1213

If that second variable changes after the man made change, then you know that it is caused by that man-made change because all of the things stay the same.1216

Their level of education and their socioeconomic status. 1227

Everything else remain the same. 1231

The only thing that changed was the introduction of that radio.1233

They addressed the third variable problem.1238

What about directionality?1243

The way that they solve directionality is like this.1247

They gave them the radio.1253

The experimenters gave the participants the radio.1255

Is the radio caused by birth control?1258

No, because the experimenters caused it.1263

We already know how variable 1 was caused.1266

It was caused artificially by the experimenters.1270

Because we have ruled out that Y causes X or variable 2 causes variable 1.1275

We know there is only one direction that could possibly be the case because we ruled out the other one.1282

This one, we rule out variable 1 causes variable 2.1290

We basically ruled this arrow out because we know what caused variable 1, the experimenters. 1297

That is one of the important components of an experiment that they had manipulation of your categorical variable. 1308

The second component is what we call control.1320

Now often people mistake the word control to mean manipulation, but I need you to separate that in your mind.1323

When you think of the word control I want you to think isolation.1331

What you need to do in order to have control is isolate the variable of interest.1340

Everything else has to be the same, except for this one thing that is different for instance, having a radio.1347

The problem with other studies is that they do not have control.1356

The women who have a lot of education, they might also be wealthy or live in a different neighborhood or already been using birth control.1361

Those women are different in many ways and variables. 1372

What we want to do in control is isolate the variable of interest and how do we do that?1377

We make sure we have 2 equivalent groups first.1390

Only difference is manipulated variable.1404

What we want to do is start off with two groups of people that look exactly the same for all intensive purposes.1416

We have two groups.1421

They are all the same.1424

They are women, there are some rich women here and some poor women here.1425

There are some educated women here. 1431

There are some uneducated women there.1433

Also this other group has the same mix of women.1436

Once you have two equivalent groups then one group get something different.1443

And that is the manipulated condition or the treatment condition.1452

So then, one gets the manipulated condition.1458

The other that does not get anything special is called the control group.1464

It is because we want to compare these two groups because they only have one thing that is different.1469

There is only one variable that is different then we know that any other changes that come from that point on is caused by that one difference.1476

Because for everything else they are the same.1486

That is what we call control.1490

Remember you have to stop reading your mind.1492

Remember control, we do not mean less when we mean control.1495

That is manipulation.1499

When we mean control think of this picture, separating out, isolating the variable of interest, 1501

making two equivalent groups, having only one thing the different.1507

When you have two equivalent groups and only one thing is different, that one stands out.1514

That is how we isolate the variable.1517

How is this solved the third variable problem?1521

While here is the thing, I had 2 equivalent groups and you have ruled out the other variables.1527

Having the two groups, 2 equivalent groups that rules out the third variable because on all these other variables like how rich or poor, ethnicity, whatever.1534

On all these other things, these women are roughly equivalent. 1551

How do we solve directionality?1556

Actually control by itself does not actually address directionality.1562

It has to be control plus manipulation in order to get directionality.1567

Directionality is taken care of by manipulation.1572

These 2 things are what make experiment special. 1580

The manipulation of a categorical variable and the control or what we call isolation of that variable.1584

Having a control group that is identical in every other way except for that one manipulated difference.1591

There are a couple of different methods of control that one could use.1602

The question is how do we create these equivalent groups but obviously a lot of times you never have exactly the same people.1606

You might have an Asian women and in one condition an Asian women in another condition.1617

They are not exactly the same.1623

How do we create 2 roughly equivalent groups or one thing we can do is hold the other variables constant.1627

Let us say where interested in two groups of women and we are worried about education.1637

Maybe we will have all college educated women.1644

They all have a university degree.1649

Everybody has a university degree and one group gets radios.1659

That might be one way we do. 1663

We just have these two groups be the same on some extraneous variable or perhaps you are really interested in whether they are married or not.1665

We might only have a single woman in our sample.1676

Both groups are made up of single women, or perhaps we are interested in age.1680

Maybe we want them all to be a similar age.1687

Holding constants means we are holding some suspected third variable constant.1691

The problem with holding constant, although it is a really good method of control is that it is just really hard to do for every single variable 1707

and there are tons of variables.1718

It is really hard to hold more than one or two variables constant.1719

Or else you might have had to find a group of women or educated the same class, the same gender, age, same locale, race.1724

After a while like it gets smaller and smaller and smaller and then you will only have 3 people that fit that description.1738

Often you can only do that for one or two extraneous variables.1747

Another method of matching is that what you have is for every person you have here, you have sort of equivalent person on in the other group.1752

Sometimes we might have age matching. 1764

For every 23-year-old, you have a 23-year-old in the other group.1766

For every 30-year-old, you have a 30-year-old in this group.1771

For every 28-year-old, you have a 28 year old in this group.1774

For every 41-year-old, you have a 41 year old in this group.1778

And then one group gets radios.1781

In this case, we are not holding some variable constant but basically roughly on average these participants 1785

are roughly equal to these participants on some third variable.1798

You can do point by point matching or you could do by just having roughly even matching.1806

For here we might have a 38-year-old, but we might have a 36-year-old.1815

If we take the means of these two groups we will be roughly equivalent on this variable of age.1820

Making sure on average, participants in both groups, equivalent on some third variable.1830

It might be how much many people make in a year.1861

It might be whether they are married or single.1865

You might have a group of women who are half married, half single but then here also you have a group that half married, half single.1868

Matching is a little bit more loose than holding constant but matching still has the same problems.1876

You can only really match on a couple of extraneous variables that you have to plan ahead of time.1885

There might be other variables that are important, but you just did not know how important they are.1891

Maybe whether they live in an apartment or house or have roommates or not.1896

Maybe those are really important but perhaps you did not match for those variables.1900

But if you think about it, there is a big problem.1908

There are so many variables that might be important like something we do not even know like 1911

having some fold in the powerless region of your brain might be really important to birth control.1921

Or maybe you know some prior experience with your first boyfriend is really important or maybe how hungry you tend to be is really important.1929

Who knows?1939

They are billion extraneous variables.1940

They are infinite number of them.1943

How could we possibly match for all of them?1945

It is impossible.1948

We cannot hold those things all constant.1949

We cannot match for all of those things.1951

What are we going to do?1953

One thing is to trust again in randomness. 1955

One way of doing it is create two groups and women are randomly assigned to these two groups.1959

It is like flipping a coin for each person like heads, in the other group tails.1968

By doing it randomly, hopefully you will have an even mix of women who are hungry and not hungry here and hungry and not hungry and here.1973

Or people who had bad experiences with their first boyfriend in here but they are randomly put into groups 1983

so that the two groups are roughly equivalent on all kinds of extraneous variables that you did not even know existed. 1993

With random assignment, the issue is you are putting your trust in randomness.2004

Randomness does not always mean that you will get equivalent groups.2009

In both of these methods of control you definitely end up with the equivalent groups for those variables that you control.2014

But for random assignment, you could roughly have 2 equivalent groups for lots of variables 2023

that you do not even know about but in exchange, you do not have a guarantee that these two groups are equivalent.2030

There is no guarantee but often random assignment is used just because there are so many variables that are just really hard to control. 2039

Let us talk about experiment terminology. 2052

When we say it is a true experiment, what I really mean is that we really did use the experimental research strategy.2055

That means manipulation and control.2068

Sometimes people will just say that this test was on the lab, it was an experiment.2073

But that does not necessarily mean that the use of experimental research strategy so you have to check for that.2078

What we call all other studies that seem like you know like they are using some research 2085

and collecting data with variables and all of that stuff, we call all of those a study.2090

If it is a correlational design can it be an experiment?2097

No, it can only be a study.2103

Whereas an experiment can actually also be called a study.2107

You just have to be careful about how you use those words from now on. 2112

One important new piece of terminology is the independent variable. 2118

This is being manipulated variable, categorical variable in an experiment.2124

That variable we are going to give it a special name because we talked about it frequently.2138

We are going to call that the ID or independent variable. 2142

The variable that you are interested in measuring the outcome of, that is called the dependent variable.2146

This is how we measure the outcome of an experiment or a study and that is called the dependent variable.2161

Factors that is just a different name for independent variables.2169

You might have 2 factors that means you have 2 IV.2174

You might have 3 factors which means you have 3 IV.2177

Factor is just another name for IV.2180

Treatment conditions are the different situations caused by manipulating that IV.2185

Once you have an IV, you will end up with multiple treatment conditions.2190

Often at least one treatment, one group that get something special, for instance getting a radio versus the control condition where they do not get anything.2195

Nothing is done to them. 2208

If the control condition, sometimes they just do not have anything else.2212

Often in medical studies, you might see randomized trials where that means they did random assignment for patients 2217

to get into the treatment condition where they get some sort of special drug.2225

But the other group they do not go into the control condition, they go into what we call the placebo condition.2229

One group will get a pill that works and have effect.2235

The other group gets sugar pills.2243

Just so that it is not just the fact of believing that it will help in order to rule that out.2246

Placebo conditions are very similar to controls except that they do get something. 2253

It is just that the thing is chemically inert like sugar.2258

Levels of an IV is the same thing as treatment conditions. 2264

In this IV we have 2 levels.2271

We have this level and that level.2273

It is another way of saying different conditions of IV and confounds or extraneous variables, 2277

these are those variables that you do not measure necessarily but affects your DV.2289

These affect DV but they are not part of the your study design.2297

Those are what we call confounds or extraneous variables. 2314

Sometimes you might hear the terminology that this is a blind experiment.2321

This means that participants in the study do not know what the conditions are and they do not know what the condition they are in.2326

There participants do not know what condition they are in.2335

This is often a piece terminology that you hear in medical studies where patients are blind.2348

It just means that they do not know which condition there in.2362

They do not know whether they are in the treatment condition or the placebo condition2365

Blind experiments take it one step further. 2370

Not only do the participants not know, but also the research team that is interacting with the participant those people do not know either.2373

Also, research team interacts with participant do not know what condition participant is in. 2391

For instance, let us say I am like the pharmacist who gives out the medication so the medication might already be labeled with the person's name. 2414

I do not know what is inside of it. 2423

I do not know if the placebo is inside of that or the drug is actually inside of it. 2425

When I interact with the participants and give them instructions and I say if you want to take this two times a day. 2431

You definitely do not want to drink milk if you are you taking this medication. 2439

Then my interaction with the participant is the same regardless of whether they are in the placebo condition or the drug condition.2446

Those are called double-blind experiments.2455

Double blind experiments are also important in psychology, where off and research assistants administer the study 2458

do not know what condition the participant is that so the computer or some other recording system will record which condition that the purchased an event.2465

The experimenter does not know they usually the person who is in charge of the entire thing they know what condition everybody 2478

but the people interacting with the participants need to be blind in order to be a double-blind experiment.2487

Finally, let us summarize how categories are related to statistics.2499

In category 1, in the script that studies usually all you can do is summarize data which we know how to do or visualize using things like histogram.2504

You know box plots and all-kinds of things by in correlational studies what you can do is also summarize the visualize data. 2523

It could still do not summarize visualize, but now you could also analyze your data with regression because this is part of summary and find correlation. 2535

Now you can apply those as well to correlational study now with experimental quasi experimental and non experimental designs of category three. 2559

Obviously, you could summarize and visualize but we cannot necessarily use regression lines 2570

because regression and correlation are saved for when you are interested in two continuous variables. 2578

But one of them is categorical. 2587

We have learn to deal of those yet we are going to learning them to the next eventually on the next lesson.2592

We got to set up probability ideas and advance but we are going to be learning about t-test and f-test and clusters.2601

These are going to depend on probability some probability principles so we are going to go cover those first.2617

but then we are going to go on and talk about this analysis later on. 2625

Let us move on to some examples.2631

Example 1, an educational psychologist has found a significant relationship between college students grade point average and their parents annual income.2634

Students with affluent parents have a higher grade point averages and students with poor parents.2645

She concluded that a student level of academic success depends on how much many the student’s parents earn.2651

What research strategy was used?2657

While we know that grade point average is continuous and annual income is continuous, so this must be category two correlational.2660

What statistical analysis were probably conducted?2673

This sounds like a positive relationship and that seems like it is correlation.2677

Here it says the academic success depends on the money from their parents.2699

That is a causal word.2707

You have to be careful because a lot of times, people would not just come out and say straight that the cause is this.2710

There are a lot of different ways of saying causes impacts depends on these code words for causality.2715

Is it true that money causes academic success?2725

Well, in a correlational design can you look at causality?2732

No why?2737

Because there is the third variable problem. 2741

It might be other things.2750

For instance, their parents might have really good work ethic and because of that they have higher incomes.2752

Maybe they insist that work ethic in their children so that they have higher grade point averages.2764

That might be one explanation of a third variable work ethic that really explains it.2769

Another thing might be that the parents income gives these kids access to extra tutoring.2775

It is really the tutoring that helps their grade point averages.2783

Let us say they have free tutoring available then maybe this relationship would just go away. 2791

Another thing might be that maybe parents income is affected by you know by their ability to delay gratification.2797

They have these values about delaying gratification and how education delays gratification.2809

Maybe because that these college students have been raised in a household where delay gratification has been highly valued 2816

and because of that, there are able to say, although I'm not going to drink because I'm going to take my final tomorrow.2825

They can delay gratification.2832

It might be that these third variable are at play.2834

Probably directionality is not as big of an issue because probably in common is not quite affected by their student’s scores.2837

GPA affecting current income that the harder link to sort of imagine.2850

That is probably not likely to be the case.2854

Example 2, a psychologist wanted to compare children in the first, third and fifth grade on their persistence on a difficult task.2859

What kind of research design is this?2867

They want to know about two variables.2870

Variable one is age or grade level right and variable 2 is persistence.2873

This is a categorical variable and maybe persistence since they might measure it by saying times spent on a difficult task.2886

It is like a really hard puzzle but they look at how long kids spent on it.2898

Maybe times spent and if that is the case, that is a continuous variable.2903

We know that we are in category three.2911

Is this an experiment?2919

Okay is this categorical variable manipulated experimenters or control it.2922

And if everything except for age controlled for.2929

No, that is not the case.2934

They did not randomly assigned these children to be these ages.2936

The children were not made to be instantaneously older.2941

They do not start off as a major just magically made older.2944

It is not experimental. 2948

It is not quasi experimental.2951

It is not really pre-post.2953

It is not really cluster experiment.2954

This is non experimental.2955

Can we conclude that age, or rather like experience causes persistence?2966

No, not necessarily.2972

It might be that somehow the third grade curriculum causes persistent or maybe it is time spent in school causes persistence.2974

There might be all these other variables involved, and so we do not necessarily know that age causes persistence.2984

Is it that age does not cause persistence?3004

That is not what I’m saying.3006

It is just that we do now know.3008

We cannot say that it does or does not.3009

Example 3, a biologist wanted to know whether complex sugars can sustain life longer than simple sugars.3013

She prepares six petri dishes, each containing 10 bee insects.3020

2 dishes are assigned to the control group and 2 are assigned to simple sugars and 2 our assigned to complex sugars.3026

She took 2 dishes and put nothing in one.3034

Simple sugars in 2 of them and complex sugars on the other 2.3036

The DV is the time it takes for half of the leaf hopper insects to die.3041

What are the cases?3047

Are the cases that the leaf hoppers?3052

No because what we are looking for is how fast half of the leaf hoppers dies?3053

It is not about the leaf hoppers itself.3063

It is actually about the dishes. The petri dishes.3068

Each of those is the case, and for each of those petri dishes they are going to have a dependent variable.3072

How long it took for half of the week the leaf hoppers to die?3080

What is the sample size?3089

The sample size is 6 dishes and is this an experiment.3091

Yes, it is presumably these dishes all started off the same and then 2 we are randomly like this that the two were special, 3097

but 2 are randomly assigned to have nothing.3105

2 randomly assigned to artificially by the experiment with extra to put simple sugars in it.3108

The experimenter put complex sugars and the other one. 3114

This is an experiment.3117

Example 4, if you wanted to test the hypothesis that hamster were raised in less daylight 3119

have higher hormone concentrations than hamsters raised in more daylight.3129

What would you do to show that daylight exposure causes hormone concentrations to increase?3136

What you would want to do is first start off with hamsters that were always sort of similar?3144

Start with similar hamsters.3153

Maybe I will get hundred hamsters that were raised similarly.3157

They are the same amount of day light.3161

I would randomly assigned these hamsters to two groups.3164

In one of the groups I would raise in less daylight and the other group would get more daylight.3171

The IV is daylight exposure.3189

How much daylight they got?3193

The DV that I'm interested in is their hormone concentrations.3195

Once they are in these two groups and now I change the way that their raised.3204

One gets more daylight.3209

One gets less daylight.3210

Than I would measure their hormone concentrations and see if there are many changes. 3213

That is our research strategies. 3219

Join us next time on www.educator.com3223