PSY 3801

Study Design


Uses for Statistics: 

  • Estimates- What is the value of a specific variable? 

  • Hypothesis testing- Forming an educated guess and using data to determine how likely it is

  • Prediction- attempting to determine future values of data using one variable to predict values of another 


Types of Statistics and Study Designs

  • Descriptive Statistics- Allow us to describe patterns we see in the data we have access to

  • Inferential Statistics- Allow us to take information from the data we collect and generalize to a larger group of people


Samples vs Populations 

  • Samples- A sample is drawn from a population (descriptive) 

  • Populations- Inferential 


Parameters and Statistics

  • Parameter- Population 

Written in Greek letters 

The measure you want to know

Often can't measure because you can't measure the whole population


# of times Michael Scott blames Toby for something across all episodes of The Office

  • Statistics- Sample 

Characteristics of your sample 

Can be measured 


25 Children’s ability to identify emotions before versus after watching Inside Out



Main Takeaways 

  • Taking samples to make claims about a population - we want  to say something about a population  

  • Can only observe a sample

What Factors Influence Our Inferences?

  • Sample size 

  • Who is in our sample

  • How the study is designed 

  • Extraneous and confounding variables 


Extraneous vs Confounding Variables

  • Extraneous Variable- variable other than the Explanatory Variable (Independent Variables) that may affect the Response Variable (Dependent Variable) 


  • Confounding Variable- an Extraneous Variable that varies systematically (it will never go away) with the Explanatory Variable (IV) 

Can’t tell if change in Response Variable (DV) is due to the Explanatory or Confounding Variable 


Example: 

Goals: Discover relationships between variables

Does caffeine improve final exam scores?


Explanatory Variable (IV) thing being manipulated: What the experimenter manipulates - amount of caffeine 


Response Variable (DV) what the experimenter measures; expected to change as a result of the explanatory variable - final exam scores 


The caffeine group all participated in the morning 

The no-caffeine group all participated in the afternoon 

Therefore the time of day would be an extraneous variable AND confounding variable

Measurement:

  • Observing and recording values for a given variable 

    • True level/amount + error 

      • What do we mean by error? measuring height (wearing heels, posture) 

      • Observed score = true ability + random error


Scales of Measurement:

Nominal (discrete/no order):

Characteristics: categories but no order, no quantitative distinctions 


Examples: married/single, zip code, experimental or control (not better than another based on the number/name) 


Ordinal (discrete/ordered):

Characteristics: ordered categories, categories ordered by size or magnitude, no distance known, why is one better than two   


Examples: clothing sizes, Olympic medals, amazon star ratings, 1-5 happiness scale 


Interval (continuous/ no meaningful zero): 

Characteristics: ordered categories, equal interval between categories, differences are comparable, no zero reference point 


Examples: calendar dates, temp, IQ (average is 100)


Ratio continuous/real/zero:

Characteristics: ordered categories, equal interval between categories, absolute zero point, natural reference point


Examples: reaction time, gain in height since last year

Explanatory: joke presented (mv) - nominal 

Response: dv thing we’re measuring - ordinal, ordered categories no equal distance 

Extraneous variable: other variables - how they tell the joke (nominal) have they seen the show or not (nominal) 


  • We want to study how interested students are in studying abroad 

How can we measure this: ordinal (lickert scale), nominal (yes or no) 

Would this be qualitative or quantitative? qualitative


What benefits or drawbacks exist for your choice? 


Scales of measurement are important: 

  • Affect the stats we can use/report

  • We can measure caffeine use before and after finals 

    • Nominal: used caffeine or did not 

    • Ordinal: scale from a little coffee to a lot of caffeine 

    • Ratio: mgs of coffee consumed



Practice Q’s 

  1. We are interested in measuring someone's stress level. We choose to measure this by giving people a questionnaire. What is a possible source of error with this measurement? 

The wording of the question 

  1. What level of measurement is used for letter grades?

Ordinal 

Letter grades dont have a specific zero meaning you cant say smeone with an F is zero performance 

Letter grades are based on intervals