Ch. 6-12 Flashcards
(94 cards)
Opportunities of Internet Research
Compared to other modes of collecting data is:
- Less expensive
- Easier to conduct
How does internet research work?
- Post ad on specialized website, ie. Social Psych Network or American Psychological Society
- Send email to online community (ie. Listserve)
Opportunities of internet research:
Decreasing the costs
Decrease the Costs
- of recruiting large, diverse, or specialized samples of research subjects
- Ie. Nosek et al. (2002) collected over 2.5 million responses on IAT in 5 years
Internet Research: Easier to Conduct
- No longer need access into intro psych courses
- Do NOT need grant money to pay subjects
- Demographic data collection – researchers at small schools or independent research
- Avoidance of time constraints
- Round the clock data collection!
- Avoidance of organizational difficulties such as scheduling because thousands can participate at once
Easier to Conduct: Observing Social Behavior
- Study on-line groups to understand group formation and change
- Conversations are pre-transcribed
- There are automated coding and content-analysis tools
Examples:
- Baym (1998): studied way groups develop sense of community by examining an email distribution list about soap operas over several years
- Bruckmann (1999): studied influence of groups on long-term learning by examining online conservations of 475 children learning a programming language over 5 years
Easier to conduct: Access to Archival Data
- Detailed transaction logs that people leave when using the internet can be used for studies of preference and choice
- Ie. Browsing behavior, software use, purchasing behavior, file uploads and downloads, subscription to communication forums
Easier to conduct: Automated
- Low marginal cost of each additional subject, close to zero!
- COMPARE TO: postal mail surveys may cost close to $2 per subject; telephone surveys cost about $40-100 per subject
Easier to conduct: Experimental Control
- COMPARED TO paper based surveys: web surveys are flexible
- Less error prone (b/c don’t require human transcription – no data entry)
- Reduction in experimenter effects
- Reduction in demand characteristics
Challenges of Internet Research: Data Quality
-
Sample biases
- To whom does web-based research generalize?
- Good external validity new for social psych who often use colleges students
- But internet users tend to be White, young and have children
- Lower response rates than telephone & paper surveys
- Dropouts: esp. for longitudinal data b/c ppl frequently change email addresses
- Possible multiple submissions by SAME person
- Anonymous nature allows people to participate frivolously or with malicious intent
Challenges of Internet Research:
Lack of Control over Data Collection Setting
In Lab we can:
- Verify subjects’ identities, age or gender
- Monitor behavior to ensure that they’re involved & serious
- Intervene if perceive undesirable effects
How to do deal with malicious responses
- May require LARGER samples to compensate for great error rate
- But can set up identifiers for invited subjects and do systematic data mining for anomalous data patterns
Challenges of Internet Research:
Protection of Human Subjects
- NOT riskier than research through other means
- But, online research changes the nature of risks and investigator’s ability to assess them
Ambiguities of Internet Research
-
Is the subject identifiable or anonymous?
- even pseudonyms can be linked to real names & responses can be tracked to individuals
- Breach of confidentiality (GREATEST RISK)
-
Do we need consent for online communication forums with unrestrcted memberships?
- Posters do NOT have expectation of privacy or do they?
-
Harm resulting from direct participation in research (ie. emotional reactions)
- Can post debrief materials & even tailor info to specific consition of subject (but cannot assess subject’s state)
In Summary
- online research has some major advantages (Low cost, diverse subject pool) to traditional means of conducting social psych research)
What is a Longitudinal Study?
- studying something over time…at least 2 time points
- can study stability vs. change
- Ex) Time 1 = beginning of school
- Time 2 = end of school year
Longitudinal Study: Advantage of Sampling More than Once
- does NOT rely on retrospective (thinking back) reports
- you report immediately
- more accurate
Diary or Experience Sampling Methods (ESM)
- can record daily events
- go about everyday lives except that asked to report what you have done or experienced or felt (via electronic or paper diary) at some interval
- report at specific intervals: once a day, or specific times a day (after each mean) or specific time intervals (every hour)
- at a signal from the researcher (email prompt, text message)
- when an event occurs (when on a date)
- *depends on phenomenon under study
Paper & Pencil Diaries
**easy to implement, **but:
- participants may forget to write in their diary
- potential for retrospection error for missed entries
- burden of data entry and handling
Benefits:
- can reprint the dates & times of expected responses on diary sheets
- prompt entries with pagers, preprogrammed wristwates, phone calls, or mobile phones (alarm, email, text msgs)
Electronic Diaries
- NO data entry
- allow for signaling/reminders
- date & time-stamp entries
- better response rates
- later questions can be sensitive to earlier responses
- **BEST/PREFERRED method!!
Which Kind of Diary Do You Use?
- depends on sample (practical) & goals of study (theoretical)
Meta-Analysis
- quanititative approach for systematically combining results of previous research to arrive at conclusions about the body of research
- can combine heterogeneous data from heterogeneous studies
- POPULAR - they have steadily increased
Origins of Meta Analysis: The Great Debate
- 1951: Hans J. Eysenck → NO favorable effects of psychotherapy
- 1978: Gene V. Glass → statistically aggregated the findings of 375 psychotherapy outcome studies, concluded that psychotherapy DID work
Why Do We Use Meta Analysis?
- traditional methods of review focus on statistical significance testing
- significance testing is NOT well suited to this task
- highly dependent on sample size
- null finding does NOT carry same “weight” as a significant finding
Why we use meta analysis:
- to synthesize results once a large body of work has been conducted
- to investigate large-scale patterns beyond the scope of individual experiments
When do we use meta analysis?
collections of research that:
- are empirical, rather than theoretical
- produce quantitative results, rather than qualitative findings
- examine the same constructs & relationships
- have findings that can be configured in a comparable statistical form (ie. a correlation coefficient is used for effect sizes)