Non-experimental designs 2

Question 1

Single subject designs

Answer 1

To study a unique individual

Data on only one person (many observations) or differences across individuals

Question 2

ABAB Reversal Design

Answer 2

Single subject design
A. Measure baseline performance on DV over time (baseline)
B. Administer treatment (IV) and record DV over time
A. Withdraw treatment (reversal), and record DV over time (back to baseline)
B. Administer treatment again and record DV over time

Control for

• history
• maturation
• testing

Question 3

Multiple Baseline Design

Answer 3

Single subject design
Can be used if treatment creates irreversible change
Attempt to replicate treatment effect by examining several related behaviors in series

Ex- therapy for improving speech articulation for various sounds

Question 4

Caution for using single-subject designs

Answer 4

Cannot counterbalance treatments
Cannot discern small effects
Cannot generalize to other individuals

Question 5

Survey research

Answer 5

Relies on self-reports of behavior
Gathers info from sample, to generalize to pop
-> critical to select representative sample
Provides correlational data only!

Question 6

Selecting samples (for survey research)

Answer 6

Convenience sample

• easy to give questionnaire to
• may be biased, unrepresentative

Random sample

• everyone in pop has equal likelihood of being sampled
• must identify every individual in pop

Stratified sample

• Identify subpopulations (strata)
• select from strata, in proportion to number in population
• > can employ multiple strata (gender, income, race, etc.)

Question 7

Surveys are

Answer 7

Instruments of communication, not just measures of behavior

• communication involves inferring what is intended by statement/questions
• Respondents infer our intentions from phrasing of survey questions or response choices

Question 8

Constructing questionnaires

Answer 8

Open-ended questions
-no constraints on responses

Close-ended questions

• provide alternative responses
• helps respondents understand what info is important
• multiple choice
• rating scale

Question 9

Info respondents use to interpret questions (infer questioner’s intent)

Answer 9

1. Response alternatives
   - the expected response (close-ended) may not reflect how respondent feels (open-ended)
2. Values of rating scales
   - 0-10 single dimension (0 success)
   - -5-+5 bipolar dimension (# of failures; # successes)
3. Frequency ranges
   - must estimate response- assumes middle range is average
4. Context question
   - letter head
   - question order (specific -> general)

Question 10

Guidelines for writing questions

Answer 10

Use interviews to “think aloud” protocols during questionnaire pretesting (check understanding of question)
Request frequency info in open-ended format
Avoid loaded or leading questions
Present conditional information FIRST (because begin to form response before finish reading)

Question 11

Admin survey by: Mail

Answer 11

Adv.

• no interviewer bias
• contact distant people
• anonymity -> honesty?

Dis

• low response rate
• nonresponse bias

Question 12

Admin survey by: Phone

Answer 12

Avd

• follow-up questions
• better response rate

Dis

• interviewer bias
• brief questions only
• nonresponse bias
• distractions

Question 13

Admin survey by: Face to face interview

Answer 13

Adv

• natural way to provide info
• high response rate

Dis

• interviewer bias
• expensive
• loss of anonymity
• sample may not be representative

Question 14

Admin survey by: Internet

Answer 14

Adv

• fast
• cheap
• contact distant people
• high response rate

Dis

• sample uncontrolled
• may be biased