Psychology Trends: How to Design a Self Study (Part 2)

How to Choose and Measure Your Variables

This Part 2 post follows the previous post in this series… Psychology Trends: How to Design a Self Study (Part 1)

Understanding Independent Variable / Dependent Variable

An independent variable is your hypothesized cause and a dependent variable is your hypothesized effect (as in cause-and -effect).

Your study will generally have only 1 independent variable but could have multiple dependent variables (see footnote * for doing a self study with multiple independent variables).

For example, you could investigate the effect of exercising (the independent variable) on your energy, your mood, and your relationship conflict (the dependent variables).

Use a spreadsheet OR draw up a piece of paper with 3 columns.

Column A. the date

Column B. Your Independent Variable (your hypothesized cause)

Column C. Your Dependent Variable (your hypothesized effect).

If you want to measure more than 1 dependent variable, add an additional column for each additional dependent variable.

Tip 1: Use a Continuous Measure If Possible.

If one of your variables is a

Yes vs. No
Happened vs. Didn’t Happen
Didn’t do it vs. Did it.

type variable you can code it as

No = 0, Yes = 1
Didn’t do it = 0, Did it = 1
etc

Enter either 0 or 1 in that column on your spreadsheet each day.

However, if possible you should generally try to use a continuous measure e.g. minutes of vigorous exercise today, happiness rating 0-10, rather than a 0 vs. 1 for did vigorous exercise today/didn’t, happy/not happy.

The reason for this is that you get better data this way. If you measure how many minutes you exercised you can always transform your data later if you want to. You can add an extra column to measure “0 = did no vigorous exercise/ 1 = did some vigorous exercise”. You will then be able to see if the number of minutes was important or just whether you did it or not. Using a continuous measure is also generally better for statistics theory reasons for when you come to analyze your data but I’ll skip going into why.

Ask yourself if there is a way of thinking of your variables as a continuous measure?

For example, for measuring relationship conflict, instead of asking “Did we have any relationship conflict today? Yes vs. No, you might ask the question as

“How much conflict did I have my relationship with my wife today?” 0-10

or

“In the last 24 hours, to what degree was my relationship with my wife characterized by conflict?” 0-10.

This is better than yes vs. no.

Tip 2: Designing Ratings Scales.

Let’s say you are interested in how doing vigorous exercise affects your next day’s energy, your anxiety, and your overall mood.

You could measure these 3 dependent variables using 3 rating scales

At the end of each day you could answer the question “How Do I Feel?” on a

0 – 10 scale, ranging from 0 = I feel no anxiety at all, to 10 = I feel extremely anxious.

0 = no energy at all, 10 = extremely energetic
0 = as miserable as I can imagine ever feeling, 10 = as happy as I can imagine ever feeling.

Tip 3: Creating a Mini-Questionnaire.

Let’s say you are only interested in one dependent variable. In this case, I will use the example of social confidence.

Instead of only completing one rating of your social confidence, there are advantages to making a mini questionnaire.

For example, a mini-questionnaire for social confidence might look like this…

Question 1.

0 = I feel very pessimistic about social situations to 10 = I feel very optimistic about social situations.

Question 2.

0 = I expect that other people will generally dislike me to 10 = I expect that other people will generally like me.

Question 3.

0 = I expect that social situations will generally go very badly to 10 = I expect that social situations will generally go very well.

Each day you would do all 3 ratings.

What are the advantages of creating a mini-questionnaire rather than only using 1 question?

– When you run some basic statistics on your self study data (I’ll show you how in an upcoming post), you can average the results of the 3 questions, or you can look at the results of each question separately, or both.

– Using the average of the 3 questions can result in a statistically more reliable measure of your dependent variable (I won’t go into the stats theory! But, if you have some basic stats knowledge then a good rule of thumb as that if the results of your 3 questions are correlated .55 or greater with EACH OTHER – not with the independent variable – then you can combine them and use the average of the 3).

– Another reason for asking your question several different ways is that sometimes a particular way of asking a question can seem like a good idea, but in fact isn’t. This way if one of your questions turns out to suck for an unanticipated reason you can discard the data for that question and still have other data.

Tip 4: Two sides of the same coin or not?

There aren’t hard and fast rules about this but in general when you create a rating scale its better to NOT assume that two things are opposites of each other.

For example, it’s better to ask 3 questions

1. How calm do I feel? 0 = not at all calm, 10 = extremely calm.
2. How relaxed do I feel? 0 = not at all relaxed, 10 = extremely relaxed.
3. How anxious do I feel? 0 = not at all anxious, 10 = extremely anxious.

Rather than

How do I feel?

0 = extremely calm/relaxed, 10 = extremely anxious.

Its generally best not to ASSUME that calm/relaxed and anxious are direct opposites of each other. Another example would be happy vs. sad. (You might have noticed that I have already violated this in my examples – like I said, no hard and fast rules)

If you are going to go to the trouble of doing a self study than adding a couple of additional rating scale questions takes very little additional time/effort but gives you a lot more data.

Tip 5. Using a More Objective Measure of Your Dependent Variable

In general, well-designed rating scales can work surprisingly well.

However another option is to use a more objective measure.

For example, let’s say you were doing a self study of whether planning your next day’s eating each evening leads you to eat more moderately the following day.

Each evening you could rate how much you ate in the day on a 0-10 scale (e.g. 0 = much less than average, 5 = average, 10 = much more than average). Or, you could use an app or online calculator to count your calories. The rating scale is obviously easier but more vulnerable to your perception.

Tip 6. Using “Collateral Data”

For example, if one of your dependent variables is your general mood, you might do your own rating of your mood and also have your partner provide a rating.

You should let your partner record their rating independently without you interfering or trying to influence their rating.

If you are recording your mood for a month, you might remind them to do the rating but ask them to give you their set of ratings are the end of the month.

I highly recommend this option if it is available to you.

Tip 7. You might want to consider booking a single session with a psychology PhD who has experience with study/questionnaire design to help you design your self study.

You could get help with choosing

– WHICH variables to measure, and

– HOW to measure your variables.

Info for how to analyze your data coming up in a future post….

Footnote* – If you have access to a program to do a regression analysis (e.g. SPSS) then you could do a self study with multiple independent variables. For example, you could enter sleep, exercise, and ratings for social activity and behavioural avoidance as the independent variables, and mood as the dependent variable. You’d then see which of the independent variables had the strongest relationship to your mood.