PSYC399, Construct Validation

Names:

  • The lab you are working in has painstakingly created a 1,000 item test of your construct of interest, “halloweenophilia,” defined “an overwhelming love of all things halloween.” One of your colleagues suspects that the test could be shorter without losing any predictive power. They construct an alternate form (13 questions, see below) and ask you to evaluate how the new test compares to the original, long version. What do you do? What are you assessing?

The Strandsylvania Halloweenophilia Scale (short-form)

 For each item, circle 1-5 where 1 is “strongly disagree” and 5 is “strongly agree”

  1. I usually plan my Halloween costume before October 1st

 1    2    3    4    5

  1. I enjoy spooky movies

 1    2    3    4    5

  1. I love pumpkin spice flavor

 1    2    3    4    5

  1. I think Halloween is the best holiday

 1    2    3    4    5

  1. My favorite artistic activity is carving pumpkins        

 1    2    3    4    5

  1. I enjoy wearing costumes

 1    2    3    4    5

  1. I own objects like fake skulls, spiders, and ghosts

 1    2    3    4    5

  1. November 1st is my least favorite day        

 1    2    3    4    5

  1. My enjoyment of Halloween sometimes interferes with other activities

 1    2    3    4    5

  1. I usually plan my Halloween costume over the summer

1    2    3    4    5

  1. I organize Halloween movie marathons

1    2    3    4    5

  1. I think about what I’ll do on Halloween ahead of time

1    2    3    4    5

  1. I am likely to know what day of the week Halloween is on this year

1    2    3    4    5

  • How might you assess the scale’s convergent validity?
  • How might you assess the scale’s discriminant validity?
  • How would you assess test-retest reliability?
  • How would you assess whether the test is uni-dimensional (that is, whether it is testing one thing or several distinct underlying things? Look closely at the items and think about which would tend to “hang together” or show similar patters of responses).
  • How do convergent and discriminant validity relate to the jingle and jangle fallacies?
  • How might poor construct validity contribute to the problems associated with the replication crisis?